system
stringclasses
7 values
input
stringlengths
10
94.5k
output
stringlengths
14
55.5k
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z8calcAreaPiPfS0_ .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ MOV R4, c[0x0][0x160] ; /* 0x0000580000047a02 */ /* 0x000fe20000000f00 */ /*0020*/ ULDC.64 UR6, c[0x0][0x118] ; /* 0x0000460000067ab9 */ /* 0x000fe20000000a00 */ /*0030*/ MOV R5, c[0x0][0x164] ; /* 0x0000590000057a02 */ /* 0x000fca0000000f00 */ /*0040*/ LDG.E R8, [R4.64] ; /* 0x0000000604087981 */ /* 0x000ea2000c1e1900 */ /*0050*/ HFMA2.MMA R10, -RZ, RZ, 1.25, 0 ; /* 0x3d000000ff0a7435 */ /* 0x000fe200000001ff */ /*0060*/ ULDC UR4, c[0x0][0x0] ; /* 0x0000000000047ab9 */ /* 0x000fe20000000800 */ /*0070*/ HFMA2.MMA R9, -RZ, RZ, 0, 0 ; /* 0x00000000ff097435 */ /* 0x000fe200000001ff */ /*0080*/ S2R R0, SR_CTAID.X ; /* 0x0000000000007919 */ /* 0x000e220000002500 */ /*0090*/ USHF.R.U32.HI UR4, URZ, 0x1, UR4 ; /* 0x000000013f047899 */ /* 0x000fc60008011604 */ /*00a0*/ S2R R2, SR_TID.X ; /* 0x0000000000027919 */ /* 0x000e620000002100 */ /*00b0*/ I2F.U32 R3, R0 ; /* 0x0000000000037306 */ /* 0x001e300000201000 */ /*00c0*/ I2F.U32 R6, R2 ; /* 0x0000000200067306 */ /* 0x002e620000201000 */ /*00d0*/ FADD R3, R3, R3 ; /* 0x0000000303037221 */ /* 0x001fc80000000000 */ /*00e0*/ FFMA R3, R3, R10, -1 ; /* 0xbf80000003037423 */ /* 0x000fe4000000000a */ /*00f0*/ FMUL R6, R6, 0.0625 ; /* 0x3d80000006067820 */ /* 0x002fc80000400000 */ /*0100*/ FFMA R7, R6, 0.00390625, R3 ; /* 0x3b80000006077823 */ /* 0x000fe20000000003 */ /*0110*/ MOV R3, UR4 ; /* 0x0000000400037c02 */ /* 0x000fe40008000f00 */ /*0120*/ ISETP.GE.AND P0, PT, R8, 0x1, PT ; /* 0x000000010800780c */ /* 0x004fda0003f06270 */ /*0130*/ @!P0 BRA 0x790 ; /* 0x0000065000008947 */ /* 0x000fea0003800000 */ /*0140*/ IMAD.MOV.U32 R10, RZ, RZ, c[0x0][0x168] ; /* 0x00005a00ff0a7624 */ /* 0x000fe200078e00ff */ /*0150*/ MOV R11, c[0x0][0x16c] ; /* 0x00005b00000b7a02 */ /* 0x000fca0000000f00 */ /*0160*/ LDG.E R4, [R10.64] ; /* 0x000000060a047981 */ /* 0x000ea2000c1e1900 */ /*0170*/ IADD3 R5, R8.reuse, -0x1, RZ ; /* 0xffffffff08057810 */ /* 0x040fe40007ffe0ff */ /*0180*/ MOV R9, RZ ; /* 0x000000ff00097202 */ /* 0x000fe40000000f00 */ /*0190*/ ISETP.GE.U32.AND P0, PT, R5, 0x3, PT ; /* 0x000000030500780c */ /* 0x000fe40003f06070 */ /*01a0*/ LOP3.LUT R5, R8, 0x3, RZ, 0xc0, !PT ; /* 0x0000000308057812 */ /* 0x000fe200078ec0ff */ /*01b0*/ FFMA R7, R4, -0.5, R7 ; /* 0xbf00000004077823 */ /* 0x004fd40000000007 */ /*01c0*/ @!P0 BRA 0x630 ; /* 0x0000046000008947 */ /* 0x000fea0003800000 */ /*01d0*/ IADD3 R6, R8, -R5, RZ ; /* 0x8000000508067210 */ /* 0x000fe40007ffe0ff */ /*01e0*/ MOV R9, RZ ; /* 0x000000ff00097202 */ /* 0x000fe40000000f00 */ /*01f0*/ FADD R17, R4, R7 ; /* 0x0000000704117221 */ /* 0x000fe20000000000 */ /*0200*/ IADD3 R6, R6, -0x4, RZ ; /* 0xfffffffc06067810 */ /* 0x000fe20007ffe0ff */ /*0210*/ BSSY B0, 0x300 ; /* 0x000000e000007945 */ /* 0x000fe40003800000 */ /*0220*/ FFMA R14, -R17, R17, 1 ; /* 0x3f800000110e7423 */ /* 0x000fe20000000111 */ /*0230*/ ISETP.NE.AND P0, PT, R6, RZ, PT ; /* 0x000000ff0600720c */ /* 0x000fc60003f05270 */ /*0240*/ MUFU.RSQ R7, R14 ; /* 0x0000000e00077308 */ /* 0x0000620000001400 */ /*0250*/ IADD3 R8, R14, -0xd000000, RZ ; /* 0xf30000000e087810 */ /* 0x000fc80007ffe0ff */ /*0260*/ ISETP.GT.U32.AND P1, PT, R8, 0x727fffff, PT ; /* 0x727fffff0800780c */ /* 0x000fda0003f24070 */ /*0270*/ @!P1 BRA 0x2b0 ; /* 0x0000003000009947 */ /* 0x000fea0003800000 */ /*0280*/ MOV R15, 0x2a0 ; /* 0x000002a0000f7802 */ /* 0x003fe40000000f00 */ /*0290*/ CALL.REL.NOINC 0x900 ; /* 0x0000066000007944 */ /* 0x000fea0003c00000 */ /*02a0*/ BRA 0x2f0 ; /* 0x0000004000007947 */ /* 0x000fea0003800000 */ /*02b0*/ FMUL.FTZ R8, R14, R7 ; /* 0x000000070e087220 */ /* 0x003fe40000410000 */ /*02c0*/ FMUL.FTZ R10, R7, 0.5 ; /* 0x3f000000070a7820 */ /* 0x000fe40000410000 */ /*02d0*/ FFMA R7, -R8, R8, R14 ; /* 0x0000000808077223 */ /* 0x000fc8000000010e */ /*02e0*/ FFMA R8, R7, R10, R8 ; /* 0x0000000a07087223 */ /* 0x000fe40000000008 */ /*02f0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*0300*/ FADD R7, R4.reuse, R17 ; /* 0x0000001104077221 */ /* 0x040fe20000000000 */ /*0310*/ BSSY B0, 0x400 ; /* 0x000000e000007945 */ /* 0x000fe20003800000 */ /*0320*/ FFMA R9, R4, R8, R9 ; /* 0x0000000804097223 */ /* 0x000fe40000000009 */ /*0330*/ FFMA R14, -R7, R7, 1 ; /* 0x3f800000070e7423 */ /* 0x000fc80000000107 */ /*0340*/ MUFU.RSQ R11, R14 ; /* 0x0000000e000b7308 */ /* 0x0000620000001400 */ /*0350*/ IADD3 R10, R14, -0xd000000, RZ ; /* 0xf30000000e0a7810 */ /* 0x000fc80007ffe0ff */ /*0360*/ ISETP.GT.U32.AND P1, PT, R10, 0x727fffff, PT ; /* 0x727fffff0a00780c */ /* 0x000fda0003f24070 */ /*0370*/ @!P1 BRA 0x3b0 ; /* 0x0000003000009947 */ /* 0x000fea0003800000 */ /*0380*/ MOV R15, 0x3a0 ; /* 0x000003a0000f7802 */ /* 0x003fe40000000f00 */ /*0390*/ CALL.REL.NOINC 0x900 ; /* 0x0000056000007944 */ /* 0x000fea0003c00000 */ /*03a0*/ BRA 0x3f0 ; /* 0x0000004000007947 */ /* 0x000fea0003800000 */ /*03b0*/ FMUL.FTZ R8, R14, R11 ; /* 0x0000000b0e087220 */ /* 0x003fe40000410000 */ /*03c0*/ FMUL.FTZ R10, R11, 0.5 ; /* 0x3f0000000b0a7820 */ /* 0x000fe40000410000 */ /*03d0*/ FFMA R11, -R8, R8, R14 ; /* 0x00000008080b7223 */ /* 0x000fc8000000010e */ /*03e0*/ FFMA R8, R11, R10, R8 ; /* 0x0000000a0b087223 */ /* 0x000fe40000000008 */ /*03f0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*0400*/ FADD R7, R4.reuse, R7 ; /* 0x0000000704077221 */ /* 0x040fe20000000000 */ /*0410*/ BSSY B0, 0x500 ; /* 0x000000e000007945 */ /* 0x000fe20003800000 */ /*0420*/ FFMA R9, R4, R8, R9 ; /* 0x0000000804097223 */ /* 0x000fe40000000009 */ /*0430*/ FFMA R14, -R7, R7, 1 ; /* 0x3f800000070e7423 */ /* 0x000fc80000000107 */ /*0440*/ MUFU.RSQ R11, R14 ; /* 0x0000000e000b7308 */ /* 0x0000620000001400 */ /*0450*/ IADD3 R10, R14, -0xd000000, RZ ; /* 0xf30000000e0a7810 */ /* 0x000fc80007ffe0ff */ /*0460*/ ISETP.GT.U32.AND P1, PT, R10, 0x727fffff, PT ; /* 0x727fffff0a00780c */ /* 0x000fda0003f24070 */ /*0470*/ @!P1 BRA 0x4b0 ; /* 0x0000003000009947 */ /* 0x000fea0003800000 */ /*0480*/ MOV R15, 0x4a0 ; /* 0x000004a0000f7802 */ /* 0x003fe40000000f00 */ /*0490*/ CALL.REL.NOINC 0x900 ; /* 0x0000046000007944 */ /* 0x000fea0003c00000 */ /*04a0*/ BRA 0x4f0 ; /* 0x0000004000007947 */ /* 0x000fea0003800000 */ /*04b0*/ FMUL.FTZ R8, R14, R11 ; /* 0x0000000b0e087220 */ /* 0x003fe40000410000 */ /*04c0*/ FMUL.FTZ R10, R11, 0.5 ; /* 0x3f0000000b0a7820 */ /* 0x000fe40000410000 */ /*04d0*/ FFMA R11, -R8, R8, R14 ; /* 0x00000008080b7223 */ /* 0x000fc8000000010e */ /*04e0*/ FFMA R8, R11, R10, R8 ; /* 0x0000000a0b087223 */ /* 0x000fe40000000008 */ /*04f0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*0500*/ FADD R7, R4.reuse, R7 ; /* 0x0000000704077221 */ /* 0x040fe20000000000 */ /*0510*/ BSSY B0, 0x610 ; /* 0x000000f000007945 */ /* 0x000fe20003800000 */ /*0520*/ FFMA R16, R4, R8, R9 ; /* 0x0000000804107223 */ /* 0x000fe40000000009 */ /*0530*/ FFMA R14, -R7, R7, 1 ; /* 0x3f800000070e7423 */ /* 0x000fc80000000107 */ /*0540*/ MUFU.RSQ R11, R14 ; /* 0x0000000e000b7308 */ /* 0x0000620000001400 */ /*0550*/ IADD3 R10, R14, -0xd000000, RZ ; /* 0xf30000000e0a7810 */ /* 0x000fc80007ffe0ff */ /*0560*/ ISETP.GT.U32.AND P1, PT, R10, 0x727fffff, PT ; /* 0x727fffff0a00780c */ /* 0x000fda0003f24070 */ /*0570*/ @!P1 BRA 0x5c0 ; /* 0x0000004000009947 */ /* 0x000fea0003800000 */ /*0580*/ MOV R15, 0x5a0 ; /* 0x000005a0000f7802 */ /* 0x003fe40000000f00 */ /*0590*/ CALL.REL.NOINC 0x900 ; /* 0x0000036000007944 */ /* 0x000fea0003c00000 */ /*05a0*/ IMAD.MOV.U32 R9, RZ, RZ, R8 ; /* 0x000000ffff097224 */ /* 0x000fe200078e0008 */ /*05b0*/ BRA 0x600 ; /* 0x0000004000007947 */ /* 0x000fea0003800000 */ /*05c0*/ FMUL.FTZ R9, R14, R11 ; /* 0x0000000b0e097220 */ /* 0x003fe40000410000 */ /*05d0*/ FMUL.FTZ R10, R11, 0.5 ; /* 0x3f0000000b0a7820 */ /* 0x000fe40000410000 */ /*05e0*/ FFMA R8, -R9, R9, R14 ; /* 0x0000000909087223 */ /* 0x000fc8000000010e */ /*05f0*/ FFMA R9, R8, R10, R9 ; /* 0x0000000a08097223 */ /* 0x000fe40000000009 */ /*0600*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*0610*/ FFMA R9, R4, R9, R16 ; /* 0x0000000904097223 */ /* 0x000fe20000000010 */ /*0620*/ @P0 BRA 0x1f0 ; /* 0xfffffbc000000947 */ /* 0x000fea000383ffff */ /*0630*/ ISETP.NE.AND P0, PT, R5, RZ, PT ; /* 0x000000ff0500720c */ /* 0x000fda0003f05270 */ /*0640*/ @!P0 BRA 0x790 ; /* 0x0000014000008947 */ /* 0x000fea0003800000 */ /*0650*/ FADD R7, R4, R7 ; /* 0x0000000704077221 */ /* 0x000fe20000000000 */ /*0660*/ BSSY B0, 0x750 ; /* 0x000000e000007945 */ /* 0x000fe60003800000 */ /*0670*/ FFMA R14, -R7, R7, 1 ; /* 0x3f800000070e7423 */ /* 0x000fc80000000107 */ /*0680*/ MUFU.RSQ R11, R14 ; /* 0x0000000e000b7308 */ /* 0x0000620000001400 */ /*0690*/ IADD3 R6, R14, -0xd000000, RZ ; /* 0xf30000000e067810 */ /* 0x000fc80007ffe0ff */ /*06a0*/ ISETP.GT.U32.AND P0, PT, R6, 0x727fffff, PT ; /* 0x727fffff0600780c */ /* 0x000fda0003f04070 */ /*06b0*/ @!P0 BRA 0x700 ; /* 0x0000004000008947 */ /* 0x000fea0003800000 */ /*06c0*/ MOV R15, 0x6e0 ; /* 0x000006e0000f7802 */ /* 0x003fe40000000f00 */ /*06d0*/ CALL.REL.NOINC 0x900 ; /* 0x0000022000007944 */ /* 0x000fea0003c00000 */ /*06e0*/ MOV R6, R8 ; /* 0x0000000800067202 */ /* 0x000fe20000000f00 */ /*06f0*/ BRA 0x740 ; /* 0x0000004000007947 */ /* 0x000fea0003800000 */ /*0700*/ FMUL.FTZ R6, R14, R11 ; /* 0x0000000b0e067220 */ /* 0x003fe40000410000 */ /*0710*/ FMUL.FTZ R8, R11, 0.5 ; /* 0x3f0000000b087820 */ /* 0x000fe40000410000 */ /*0720*/ FFMA R11, -R6, R6, R14 ; /* 0x00000006060b7223 */ /* 0x000fc8000000010e */ /*0730*/ FFMA R6, R11, R8, R6 ; /* 0x000000080b067223 */ /* 0x000fe40000000006 */ /*0740*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*0750*/ IADD3 R5, R5, -0x1, RZ ; /* 0xffffffff05057810 */ /* 0x000fe20007ffe0ff */ /*0760*/ FFMA R9, R4, R6, R9 ; /* 0x0000000604097223 */ /* 0x000fc60000000009 */ /*0770*/ ISETP.NE.AND P0, PT, R5, RZ, PT ; /* 0x000000ff0500720c */ /* 0x000fda0003f05270 */ /*0780*/ @P0 BRA 0x650 ; /* 0xfffffec000000947 */ /* 0x000fea000383ffff */ /*0790*/ STS [R2.X4], R9 ; /* 0x0000000902007388 */ /* 0x0001e20000004800 */ /*07a0*/ ISETP.NE.AND P1, PT, R3, RZ, PT ; /* 0x000000ff0300720c */ /* 0x000fe40003f25270 */ /*07b0*/ ISETP.NE.AND P0, PT, R2, RZ, PT ; /* 0x000000ff0200720c */ /* 0x000fd60003f05270 */ /*07c0*/ @!P1 BRA 0x8a0 ; /* 0x000000d000009947 */ /* 0x000fea0003800000 */ /*07d0*/ SHF.L.U32 R4, R2, 0x2, RZ ; /* 0x0000000202047819 */ /* 0x001fe400000006ff */ /*07e0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe20000010000 */ /*07f0*/ ISETP.GE.U32.AND P1, PT, R2, R3, PT ; /* 0x000000030200720c */ /* 0x000fda0003f26070 */ /*0800*/ @!P1 LEA R5, R3.reuse, R4, 0x2 ; /* 0x0000000403059211 */ /* 0x040fe200078e10ff */ /*0810*/ @!P1 LDS R6, [R2.X4] ; /* 0x0000000002069984 */ /* 0x000fea0000004800 */ /*0820*/ @!P1 LDS R5, [R5] ; /* 0x0000000005059984 */ /* 0x000e240000000800 */ /*0830*/ @!P1 FADD R7, R6, R5 ; /* 0x0000000506079221 */ /* 0x001fe20000000000 */ /*0840*/ IADD3 R6, R3.reuse, 0x1, RZ ; /* 0x0000000103067810 */ /* 0x040fe40007ffe0ff */ /*0850*/ LEA.HI R3, R3, R3, RZ, 0x1 ; /* 0x0000000303037211 */ /* 0x000fc400078f08ff */ /*0860*/ @!P1 STS [R2.X4], R7 ; /* 0x0000000702009388 */ /* 0x0001e20000004800 */ /*0870*/ ISETP.GT.U32.AND P1, PT, R6, 0x2, PT ; /* 0x000000020600780c */ /* 0x000fe40003f24070 */ /*0880*/ SHF.R.S32.HI R3, RZ, 0x1, R3 ; /* 0x00000001ff037819 */ /* 0x000fd60000011403 */ /*0890*/ @P1 BRA 0x7e0 ; /* 0xffffff4000001947 */ /* 0x001fea000383ffff */ /*08a0*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x001fea0003800000 */ /*08b0*/ LDS R5, [RZ] ; /* 0x00000000ff057984 */ /* 0x000e220000000800 */ /*08c0*/ HFMA2.MMA R3, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff037435 */ /* 0x000fd400000001ff */ /*08d0*/ IMAD.WIDE.U32 R2, R0, R3, c[0x0][0x170] ; /* 0x00005c0000027625 */ /* 0x000fca00078e0003 */ /*08e0*/ STG.E [R2.64], R5 ; /* 0x0000000502007986 */ /* 0x001fe2000c101906 */ /*08f0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0900*/ LOP3.LUT P1, RZ, R14, 0x7fffffff, RZ, 0xc0, !PT ; /* 0x7fffffff0eff7812 */ /* 0x000fda000782c0ff */ /*0910*/ @!P1 IMAD.MOV.U32 R8, RZ, RZ, R14 ; /* 0x000000ffff089224 */ /* 0x000fe200078e000e */ /*0920*/ @!P1 BRA 0xa30 ; /* 0x0000010000009947 */ /* 0x000fea0003800000 */ /*0930*/ FSETP.GEU.FTZ.AND P1, PT, R14, RZ, PT ; /* 0x000000ff0e00720b */ /* 0x000fda0003f3e000 */ /*0940*/ @!P1 MOV R8, 0x7fffffff ; /* 0x7fffffff00089802 */ /* 0x000fe20000000f00 */ /*0950*/ @!P1 BRA 0xa30 ; /* 0x000000d000009947 */ /* 0x000fea0003800000 */ /*0960*/ FSETP.GTU.FTZ.AND P1, PT, |R14|, +INF , PT ; /* 0x7f8000000e00780b */ /* 0x000fda0003f3c200 */ /*0970*/ @P1 FADD.FTZ R8, R14, 1 ; /* 0x3f8000000e081421 */ /* 0x000fe20000010000 */ /*0980*/ @P1 BRA 0xa30 ; /* 0x000000a000001947 */ /* 0x000fea0003800000 */ /*0990*/ FSETP.NEU.FTZ.AND P1, PT, |R14|, +INF , PT ; /* 0x7f8000000e00780b */ /* 0x000fda0003f3d200 */ /*09a0*/ @P1 FFMA R10, R14, 1.84467440737095516160e+19, RZ ; /* 0x5f8000000e0a1823 */ /* 0x000fc800000000ff */ /*09b0*/ @P1 MUFU.RSQ R11, R10 ; /* 0x0000000a000b1308 */ /* 0x000e240000001400 */ /*09c0*/ @P1 FMUL.FTZ R13, R10, R11 ; /* 0x0000000b0a0d1220 */ /* 0x001fe40000410000 */ /*09d0*/ @P1 FMUL.FTZ R11, R11, 0.5 ; /* 0x3f0000000b0b1820 */ /* 0x000fe40000410000 */ /*09e0*/ @P1 FADD.FTZ R8, -R13, -RZ ; /* 0x800000ff0d081221 */ /* 0x000fc80000010100 */ /*09f0*/ @P1 FFMA R12, R13, R8, R10 ; /* 0x000000080d0c1223 */ /* 0x000fe2000000000a */ /*0a00*/ @!P1 MOV R8, R14 ; /* 0x0000000e00089202 */ /* 0x000fc60000000f00 */ /*0a10*/ @P1 FFMA R11, R12, R11, R13 ; /* 0x0000000b0c0b1223 */ /* 0x000fc8000000000d */ /*0a20*/ @P1 FMUL.FTZ R8, R11, 2.3283064365386962891e-10 ; /* 0x2f8000000b081820 */ /* 0x000fe40000410000 */ /*0a30*/ HFMA2.MMA R11, -RZ, RZ, 0, 0 ; /* 0x00000000ff0b7435 */ /* 0x000fe200000001ff */ /*0a40*/ MOV R10, R15 ; /* 0x0000000f000a7202 */ /* 0x000fca0000000f00 */ /*0a50*/ RET.REL.NODEC R10 0x0 ; /* 0xfffff5a00a007950 */ /* 0x000fea0003c3ffff */ /*0a60*/ BRA 0xa60; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0a70*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a80*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a90*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0aa0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0ab0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0ac0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0ad0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0ae0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0af0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z8calcAreaPiPfS0_ .globl _Z8calcAreaPiPfS0_ .p2align 8 .type _Z8calcAreaPiPfS0_,@function _Z8calcAreaPiPfS0_: s_load_b64 s[2:3], s[0:1], 0x0 s_mov_b32 s4, s15 s_waitcnt lgkmcnt(0) s_load_b32 s3, s[2:3], 0x0 s_waitcnt lgkmcnt(0) s_cmp_lt_i32 s3, 1 s_cbranch_scc1 .LBB0_3 s_load_b64 s[6:7], s[0:1], 0x8 v_cvt_f32_u32_e32 v1, s4 v_cvt_f32_u32_e32 v2, v0 v_mov_b32_e32 v3, -1.0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_dual_add_f32 v1, v1, v1 :: v_dual_mul_f32 v2, 0x3d800000, v2 v_fmamk_f32 v1, v1, 0x3d000000, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_dual_fmamk_f32 v2, v2, 0x3b800000, v1 :: v_dual_mov_b32 v1, 0 s_waitcnt lgkmcnt(0) s_load_b32 s5, s[6:7], 0x0 s_waitcnt lgkmcnt(0) v_fmac_f32_e64 v2, s5, -0.5 s_set_inst_prefetch_distance 0x1 .p2align 6 .LBB0_2: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_add_f32_e32 v2, s5, v2 s_add_i32 s3, s3, -1 s_cmp_eq_u32 s3, 0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fma_f32 v3, -v2, v2, 1.0 v_mul_f32_e32 v4, 0x4f800000, v3 v_cmp_gt_f32_e32 vcc_lo, 0xf800000, v3 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_cndmask_b32_e32 v3, v3, v4, vcc_lo v_sqrt_f32_e32 v4, v3 s_waitcnt_depctr 0xfff v_add_nc_u32_e32 v5, -1, v4 v_add_nc_u32_e32 v6, 1, v4 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_fma_f32 v7, -v5, v4, v3 v_fma_f32 v8, -v6, v4, v3 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_ge_f32_e64 s2, 0, v7 v_cndmask_b32_e64 v4, v4, v5, s2 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_lt_f32_e64 s2, 0, v8 v_cndmask_b32_e64 v4, v4, v6, s2 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f32_e32 v5, 0x37800000, v4 v_cndmask_b32_e32 v4, v4, v5, vcc_lo v_cmp_class_f32_e64 vcc_lo, v3, 0x260 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_cndmask_b32_e32 v3, v4, v3, vcc_lo v_fmac_f32_e32 v1, s5, v3 s_cbranch_scc0 .LBB0_2 s_branch .LBB0_4 .LBB0_3: v_mov_b32_e32 v1, 0 .LBB0_4: s_set_inst_prefetch_distance 0x2 s_load_b32 s2, s[0:1], 0x24 v_lshlrev_b32_e32 v2, 2, v0 ds_store_b32 v2, v1 s_waitcnt lgkmcnt(0) v_cmp_lt_u16_e64 s3, s2, 2 s_delay_alu instid0(VALU_DEP_1) s_and_b32 vcc_lo, exec_lo, s3 s_cbranch_vccnz .LBB0_9 s_and_b32 s2, 0xffff, s2 s_delay_alu instid0(SALU_CYCLE_1) s_lshr_b32 s2, s2, 1 s_branch .LBB0_7 .p2align 6 .LBB0_6: s_or_b32 exec_lo, exec_lo, s3 s_lshr_b32 s3, s2, 1 s_cmp_gt_u32 s2, 1 s_mov_b32 s2, s3 s_cbranch_scc0 .LBB0_9 .LBB0_7: s_mov_b32 s3, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e64 s2, v0 s_cbranch_execz .LBB0_6 v_add_lshl_u32 v1, s2, v0, 2 ds_load_b32 v1, v1 ds_load_b32 v3, v2 s_waitcnt lgkmcnt(0) v_add_f32_e32 v1, v1, v3 ds_store_b32 v2, v1 s_branch .LBB0_6 .LBB0_9: s_mov_b32 s5, 0 s_mov_b32 s2, exec_lo v_cmpx_eq_u32_e32 0, v0 s_cbranch_execz .LBB0_11 v_mov_b32_e32 v0, 0 s_load_b64 s[0:1], s[0:1], 0x10 s_lshl_b64 s[2:3], s[4:5], 2 ds_load_b32 v1, v0 s_waitcnt lgkmcnt(0) s_add_u32 s0, s0, s2 s_addc_u32 s1, s1, s3 global_store_b32 v0, v1, s[0:1] .LBB0_11: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z8calcAreaPiPfS0_ .amdhsa_group_segment_fixed_size 1024 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 280 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 9 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z8calcAreaPiPfS0_, .Lfunc_end0-_Z8calcAreaPiPfS0_ .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: hidden_block_count_x - .offset: 28 .size: 4 .value_kind: hidden_block_count_y - .offset: 32 .size: 4 .value_kind: hidden_block_count_z - .offset: 36 .size: 2 .value_kind: hidden_group_size_x - .offset: 38 .size: 2 .value_kind: hidden_group_size_y - .offset: 40 .size: 2 .value_kind: hidden_group_size_z - .offset: 42 .size: 2 .value_kind: hidden_remainder_x - .offset: 44 .size: 2 .value_kind: hidden_remainder_y - .offset: 46 .size: 2 .value_kind: hidden_remainder_z - .offset: 64 .size: 8 .value_kind: hidden_global_offset_x - .offset: 72 .size: 8 .value_kind: hidden_global_offset_y - .offset: 80 .size: 8 .value_kind: hidden_global_offset_z - .offset: 88 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 1024 .kernarg_segment_align: 8 .kernarg_segment_size: 280 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z8calcAreaPiPfS0_ .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z8calcAreaPiPfS0_.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 9 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_00162233_00000000-6_pi_area_cuda.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2060: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2060: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z32__device_stub__Z8calcAreaPiPfS0_PiPfS0_ .type _Z32__device_stub__Z8calcAreaPiPfS0_PiPfS0_, @function _Z32__device_stub__Z8calcAreaPiPfS0_PiPfS0_: .LFB2082: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movq %fs:40, %rax movq %rax, 120(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 120(%rsp), %rax subq %fs:40, %rax jne .L8 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 152 pushq 40(%rsp) .cfi_def_cfa_offset 160 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z8calcAreaPiPfS0_(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2082: .size _Z32__device_stub__Z8calcAreaPiPfS0_PiPfS0_, .-_Z32__device_stub__Z8calcAreaPiPfS0_PiPfS0_ .globl _Z8calcAreaPiPfS0_ .type _Z8calcAreaPiPfS0_, @function _Z8calcAreaPiPfS0_: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z32__device_stub__Z8calcAreaPiPfS0_PiPfS0_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z8calcAreaPiPfS0_, .-_Z8calcAreaPiPfS0_ .section .rodata.str1.1,"aMS",@progbits,1 .LC2: .string "pi" .LC3: .string "\n==\n==\t%20s = %15.10f\n" .LC4: .string "total rectangles" .LC5: .string "==\t%20s = %15d\n" .LC6: .string "CUDA threads" .LC7: .string "==\t%20s = %15d\n==\n\n" .text .globl main .type main, @function main: .LFB2057: .cfi_startproc endbr64 pushq %rbp .cfi_def_cfa_offset 16 .cfi_offset 6, -16 pushq %rbx .cfi_def_cfa_offset 24 .cfi_offset 3, -24 subq $216, %rsp .cfi_def_cfa_offset 240 movq %fs:40, %rax movq %rax, 200(%rsp) xorl %eax, %eax movl $10000000, %eax cmpl $2, %edi je .L19 .L12: leal 8191(%rax), %ebp testl %eax, %eax cmovns %eax, %ebp sarl $13, %ebp movl %ebp, 8(%rsp) sall $13, %ebp pxor %xmm1, %xmm1 cvtsi2ssl %ebp, %xmm1 movss .LC1(%rip), %xmm0 divss %xmm1, %xmm0 movss %xmm0, 12(%rsp) leaq 16(%rsp), %rdi movl $4, %esi call cudaMalloc@PLT leaq 24(%rsp), %rdi movl $4, %esi call cudaMalloc@PLT leaq 32(%rsp), %rdi movl $128, %esi call cudaMalloc@PLT leaq 8(%rsp), %rsi movl $1, %ecx movl $4, %edx movq 16(%rsp), %rdi call cudaMemcpy@PLT leaq 12(%rsp), %rsi movl $1, %ecx movl $4, %edx movq 24(%rsp), %rdi call cudaMemcpy@PLT movl $256, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $32, 40(%rsp) movl $1, 44(%rsp) movl $1, 48(%rsp) movl $0, %r9d movl $0, %r8d movq 52(%rsp), %rdx movl $1, %ecx movq 40(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L20 .L13: leaq 64(%rsp), %rbx movl $2, %ecx movl $128, %edx movq 32(%rsp), %rsi movq %rbx, %rdi call cudaMemcpy@PLT movq %rbx, %rax leaq 192(%rsp), %rdx pxor %xmm0, %xmm0 .L14: addss (%rax), %xmm0 addq $4, %rax cmpq %rdx, %rax jne .L14 addss %xmm0, %xmm0 cvtss2sd %xmm0, %xmm0 leaq .LC2(%rip), %rdx leaq .LC3(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movl %ebp, %ecx leaq .LC4(%rip), %rdx leaq .LC5(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl $8192, %ecx leaq .LC6(%rip), %rdx leaq .LC7(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq 200(%rsp), %rax subq %fs:40, %rax jne .L21 movl $0, %eax addq $216, %rsp .cfi_remember_state .cfi_def_cfa_offset 24 popq %rbx .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 ret .L19: .cfi_restore_state movq 8(%rsi), %rdi movl $10, %edx movl $0, %esi call __isoc23_strtol@PLT jmp .L12 .L20: movq 32(%rsp), %rdx movq 24(%rsp), %rsi movq 16(%rsp), %rdi call _Z32__device_stub__Z8calcAreaPiPfS0_PiPfS0_ jmp .L13 .L21: call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1 .LC8: .string "_Z8calcAreaPiPfS0_" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2085: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC8(%rip), %rdx movq %rdx, %rcx leaq _Z8calcAreaPiPfS0_(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2085: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC1: .long 1073741824 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "pi_area_cuda.hip" .globl _Z23__device_stub__calcAreaPiPfS0_ # -- Begin function _Z23__device_stub__calcAreaPiPfS0_ .p2align 4, 0x90 .type _Z23__device_stub__calcAreaPiPfS0_,@function _Z23__device_stub__calcAreaPiPfS0_: # @_Z23__device_stub__calcAreaPiPfS0_ .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movq %rdx, 56(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z8calcAreaPiPfS0_, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $120, %rsp .cfi_adjust_cfa_offset -120 retq .Lfunc_end0: .size _Z23__device_stub__calcAreaPiPfS0_, .Lfunc_end0-_Z23__device_stub__calcAreaPiPfS0_ .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function main .LCPI1_0: .long 0x40000000 # float 2 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 subq $240, %rsp .cfi_def_cfa_offset 256 .cfi_offset %rbx, -16 movl $1220, %ebx # imm = 0x4C4 cmpl $2, %edi jne .LBB1_2 # %bb.1: movq 8(%rsi), %rdi xorl %esi, %esi movl $10, %edx callq __isoc23_strtol leal 8191(%rax), %ebx testl %eax, %eax cmovnsl %eax, %ebx sarl $13, %ebx .LBB1_2: movl %ebx, 12(%rsp) shll $13, %ebx cvtsi2ss %ebx, %xmm0 movss .LCPI1_0(%rip), %xmm1 # xmm1 = mem[0],zero,zero,zero divss %xmm0, %xmm1 movss %xmm1, 8(%rsp) leaq 32(%rsp), %rdi movl $4, %esi callq hipMalloc leaq 24(%rsp), %rdi movl $4, %esi callq hipMalloc leaq 16(%rsp), %rdi movl $128, %esi callq hipMalloc movq 32(%rsp), %rdi leaq 12(%rsp), %rsi movl $4, %edx movl $1, %ecx callq hipMemcpy movq 24(%rsp), %rdi leaq 8(%rsp), %rsi movl $4, %edx movl $1, %ecx callq hipMemcpy movabsq $4294967328, %rdi # imm = 0x100000020 leaq 224(%rdi), %rdx movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_4 # %bb.3: movq 32(%rsp), %rax movq 24(%rsp), %rcx movq 16(%rsp), %rdx movq %rax, 104(%rsp) movq %rcx, 96(%rsp) movq %rdx, 88(%rsp) leaq 104(%rsp), %rax movq %rax, 112(%rsp) leaq 96(%rsp), %rax movq %rax, 120(%rsp) leaq 88(%rsp), %rax movq %rax, 128(%rsp) leaq 72(%rsp), %rdi leaq 56(%rsp), %rsi leaq 48(%rsp), %rdx leaq 40(%rsp), %rcx callq __hipPopCallConfiguration movq 72(%rsp), %rsi movl 80(%rsp), %edx movq 56(%rsp), %rcx movl 64(%rsp), %r8d leaq 112(%rsp), %r9 movl $_Z8calcAreaPiPfS0_, %edi pushq 40(%rsp) .cfi_adjust_cfa_offset 8 pushq 56(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_4: movq 16(%rsp), %rsi leaq 112(%rsp), %rdi movl $128, %edx movl $2, %ecx callq hipMemcpy xorps %xmm0, %xmm0 xorl %eax, %eax .p2align 4, 0x90 .LBB1_5: # =>This Inner Loop Header: Depth=1 addss 112(%rsp,%rax,4), %xmm0 incq %rax cmpq $32, %rax jne .LBB1_5 # %bb.6: addss %xmm0, %xmm0 cvtss2sd %xmm0, %xmm0 movl $.L.str, %edi movl $.L.str.1, %esi movb $1, %al callq printf movl $.L.str.2, %edi movl $.L.str.3, %esi movl %ebx, %edx xorl %eax, %eax callq printf movl $.L.str.4, %edi movl $.L.str.5, %esi movl $8192, %edx # imm = 0x2000 xorl %eax, %eax callq printf xorl %eax, %eax addq $240, %rsp .cfi_def_cfa_offset 16 popq %rbx .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z8calcAreaPiPfS0_, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z8calcAreaPiPfS0_,@object # @_Z8calcAreaPiPfS0_ .section .rodata,"a",@progbits .globl _Z8calcAreaPiPfS0_ .p2align 3, 0x0 _Z8calcAreaPiPfS0_: .quad _Z23__device_stub__calcAreaPiPfS0_ .size _Z8calcAreaPiPfS0_, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "\n==\n==\t%20s = %15.10f\n" .size .L.str, 23 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz "pi" .size .L.str.1, 3 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "==\t%20s = %15d\n" .size .L.str.2, 16 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "total rectangles" .size .L.str.3, 17 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "==\t%20s = %15d\n==\n\n" .size .L.str.4, 20 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "CUDA threads" .size .L.str.5, 13 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z8calcAreaPiPfS0_" .size .L__unnamed_1, 19 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z23__device_stub__calcAreaPiPfS0_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z8calcAreaPiPfS0_ .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
struct ProgramGPUColorRGB { __device__ ProgramGPUColorRGB() { } unsigned char Blue; unsigned char Green; unsigned char Red; unsigned char Alpha; }; // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0); // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0) { __syncthreads(); indices[(threadIdx.x)]++; }
code for sm_80 Function : FindPixel .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R2, SR_TID.X ; /* 0x0000000000027919 */ /* 0x000e220000002100 */ /*0020*/ HFMA2.MMA R3, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff037435 */ /* 0x000fe200000001ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe40000000a00 */ /*0040*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fee0000010000 */ /*0050*/ IMAD.WIDE.U32 R2, R2, R3, c[0x0][0x180] ; /* 0x0000600002027625 */ /* 0x001fca00078e0003 */ /*0060*/ LDG.E R0, [R2.64] ; /* 0x0000000402007981 */ /* 0x000ea4000c1e1900 */ /*0070*/ IADD3 R5, R0, 0x1, RZ ; /* 0x0000000100057810 */ /* 0x004fca0007ffe0ff */ /*0080*/ STG.E [R2.64], R5 ; /* 0x0000000502007986 */ /* 0x000fe2000c101904 */ /*0090*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00a0*/ BRA 0xa0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*00b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0100*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0110*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0120*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0130*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0140*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0150*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0160*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0170*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
struct ProgramGPUColorRGB { __device__ ProgramGPUColorRGB() { } unsigned char Blue; unsigned char Green; unsigned char Red; unsigned char Alpha; }; // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0); // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0) { __syncthreads(); indices[(threadIdx.x)]++; }
.file "tmpxft_000a236f_00000000-6_CUDAFYSOURCETEMP.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2032: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2032: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi .type _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi, @function _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi: .LFB2054: .cfi_startproc endbr64 subq $200, %rsp .cfi_def_cfa_offset 208 movq %rdi, 40(%rsp) movl %esi, 36(%rsp) movq %rdx, 24(%rsp) movl %ecx, 32(%rsp) movq %r8, 16(%rsp) movl %r9d, 12(%rsp) movq 208(%rsp), %rax movq %rax, (%rsp) movq %fs:40, %rax movq %rax, 184(%rsp) xorl %eax, %eax leaq 40(%rsp), %rax movq %rax, 112(%rsp) leaq 36(%rsp), %rax movq %rax, 120(%rsp) leaq 24(%rsp), %rax movq %rax, 128(%rsp) leaq 32(%rsp), %rax movq %rax, 136(%rsp) leaq 16(%rsp), %rax movq %rax, 144(%rsp) leaq 12(%rsp), %rax movq %rax, 152(%rsp) movq %rsp, %rax movq %rax, 160(%rsp) leaq 216(%rsp), %rax movq %rax, 168(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) movl $1, 72(%rsp) movl $1, 76(%rsp) movl $1, 80(%rsp) movl $1, 84(%rsp) leaq 56(%rsp), %rcx leaq 48(%rsp), %rdx leaq 76(%rsp), %rsi leaq 64(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 184(%rsp), %rax subq %fs:40, %rax jne .L8 addq $200, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 56(%rsp) .cfi_def_cfa_offset 216 pushq 56(%rsp) .cfi_def_cfa_offset 224 leaq 128(%rsp), %r9 movq 92(%rsp), %rcx movl 100(%rsp), %r8d movq 80(%rsp), %rsi movl 88(%rsp), %edx leaq FindPixel(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 208 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2054: .size _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi, .-_Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi .globl FindPixel .type FindPixel, @function FindPixel: .LFB2055: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movl 24(%rsp), %eax pushq %rax .cfi_def_cfa_offset 24 pushq 24(%rsp) .cfi_def_cfa_offset 32 call _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi addq $24, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2055: .size FindPixel, .-FindPixel .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "FindPixel" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2057: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq FindPixel(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2057: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
struct ProgramGPUColorRGB { __device__ ProgramGPUColorRGB() { } unsigned char Blue; unsigned char Green; unsigned char Red; unsigned char Alpha; }; // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0); // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0) { __syncthreads(); indices[(threadIdx.x)]++; }
#include <hip/hip_runtime.h> struct ProgramGPUColorRGB { __device__ ProgramGPUColorRGB() { } unsigned char Blue; unsigned char Green; unsigned char Red; unsigned char Alpha; }; // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0); // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0) { __syncthreads(); indices[(threadIdx.x)]++; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> struct ProgramGPUColorRGB { __device__ ProgramGPUColorRGB() { } unsigned char Blue; unsigned char Green; unsigned char Red; unsigned char Alpha; }; // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0); // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0) { __syncthreads(); indices[(threadIdx.x)]++; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected FindPixel .globl FindPixel .p2align 8 .type FindPixel,@function FindPixel: s_load_b64 s[0:1], s[0:1], 0x20 v_lshlrev_b32_e32 v0, 2, v0 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv global_load_b32 v1, v0, s[0:1] s_waitcnt vmcnt(0) v_add_nc_u32_e32 v1, 1, v1 global_store_b32 v0, v1, s[0:1] s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel FindPixel .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 60 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 2 .amdhsa_next_free_sgpr 2 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size FindPixel, .Lfunc_end0-FindPixel .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .offset: 8 .size: 4 .value_kind: by_value - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: by_value - .address_space: global .offset: 32 .size: 8 .value_kind: global_buffer - .offset: 40 .size: 4 .value_kind: by_value - .address_space: global .offset: 48 .size: 8 .value_kind: global_buffer - .offset: 56 .size: 4 .value_kind: by_value .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 60 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: FindPixel .private_segment_fixed_size: 0 .sgpr_count: 2 .sgpr_spill_count: 0 .symbol: FindPixel.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 2 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <hip/hip_runtime.h> struct ProgramGPUColorRGB { __device__ ProgramGPUColorRGB() { } unsigned char Blue; unsigned char Green; unsigned char Red; unsigned char Alpha; }; // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0); // Insaniquarium_Deluxe_Bot.Program extern "C" __global__ void FindPixel( ProgramGPUColorRGB* rgbColors, int rgbColorsLen0, ProgramGPUColorRGB* colors, int colorsLen0, int* indices, int indicesLen0, float* output, int outputLen0) { __syncthreads(); indices[(threadIdx.x)]++; }
.text .file "CUDAFYSOURCETEMP.hip" .globl __device_stub__FindPixel # -- Begin function __device_stub__FindPixel .p2align 4, 0x90 .type __device_stub__FindPixel,@function __device_stub__FindPixel: # @__device_stub__FindPixel .cfi_startproc # %bb.0: subq $168, %rsp .cfi_def_cfa_offset 176 movq %rdi, 88(%rsp) movl %esi, 20(%rsp) movq %rdx, 80(%rsp) movl %ecx, 16(%rsp) movq %r8, 72(%rsp) movl %r9d, 12(%rsp) leaq 88(%rsp), %rax movq %rax, 96(%rsp) leaq 20(%rsp), %rax movq %rax, 104(%rsp) leaq 80(%rsp), %rax movq %rax, 112(%rsp) leaq 16(%rsp), %rax movq %rax, 120(%rsp) leaq 72(%rsp), %rax movq %rax, 128(%rsp) leaq 12(%rsp), %rax movq %rax, 136(%rsp) leaq 176(%rsp), %rax movq %rax, 144(%rsp) leaq 184(%rsp), %rax movq %rax, 152(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 96(%rsp), %r9 movl $FindPixel, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $184, %rsp .cfi_adjust_cfa_offset -184 retq .Lfunc_end0: .size __device_stub__FindPixel, .Lfunc_end0-__device_stub__FindPixel .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $FindPixel, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type FindPixel,@object # @FindPixel .section .rodata,"a",@progbits .globl FindPixel .p2align 3, 0x0 FindPixel: .quad __device_stub__FindPixel .size FindPixel, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "FindPixel" .size .L__unnamed_1, 10 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __device_stub__FindPixel .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym FindPixel .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : FindPixel .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R2, SR_TID.X ; /* 0x0000000000027919 */ /* 0x000e220000002100 */ /*0020*/ HFMA2.MMA R3, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff037435 */ /* 0x000fe200000001ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe40000000a00 */ /*0040*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fee0000010000 */ /*0050*/ IMAD.WIDE.U32 R2, R2, R3, c[0x0][0x180] ; /* 0x0000600002027625 */ /* 0x001fca00078e0003 */ /*0060*/ LDG.E R0, [R2.64] ; /* 0x0000000402007981 */ /* 0x000ea4000c1e1900 */ /*0070*/ IADD3 R5, R0, 0x1, RZ ; /* 0x0000000100057810 */ /* 0x004fca0007ffe0ff */ /*0080*/ STG.E [R2.64], R5 ; /* 0x0000000502007986 */ /* 0x000fe2000c101904 */ /*0090*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00a0*/ BRA 0xa0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*00b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0100*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0110*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0120*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0130*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0140*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0150*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0160*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0170*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected FindPixel .globl FindPixel .p2align 8 .type FindPixel,@function FindPixel: s_load_b64 s[0:1], s[0:1], 0x20 v_lshlrev_b32_e32 v0, 2, v0 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv global_load_b32 v1, v0, s[0:1] s_waitcnt vmcnt(0) v_add_nc_u32_e32 v1, 1, v1 global_store_b32 v0, v1, s[0:1] s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel FindPixel .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 60 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 2 .amdhsa_next_free_sgpr 2 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size FindPixel, .Lfunc_end0-FindPixel .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .offset: 8 .size: 4 .value_kind: by_value - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: by_value - .address_space: global .offset: 32 .size: 8 .value_kind: global_buffer - .offset: 40 .size: 4 .value_kind: by_value - .address_space: global .offset: 48 .size: 8 .value_kind: global_buffer - .offset: 56 .size: 4 .value_kind: by_value .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 60 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: FindPixel .private_segment_fixed_size: 0 .sgpr_count: 2 .sgpr_spill_count: 0 .symbol: FindPixel.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 2 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_000a236f_00000000-6_CUDAFYSOURCETEMP.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2032: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2032: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi .type _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi, @function _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi: .LFB2054: .cfi_startproc endbr64 subq $200, %rsp .cfi_def_cfa_offset 208 movq %rdi, 40(%rsp) movl %esi, 36(%rsp) movq %rdx, 24(%rsp) movl %ecx, 32(%rsp) movq %r8, 16(%rsp) movl %r9d, 12(%rsp) movq 208(%rsp), %rax movq %rax, (%rsp) movq %fs:40, %rax movq %rax, 184(%rsp) xorl %eax, %eax leaq 40(%rsp), %rax movq %rax, 112(%rsp) leaq 36(%rsp), %rax movq %rax, 120(%rsp) leaq 24(%rsp), %rax movq %rax, 128(%rsp) leaq 32(%rsp), %rax movq %rax, 136(%rsp) leaq 16(%rsp), %rax movq %rax, 144(%rsp) leaq 12(%rsp), %rax movq %rax, 152(%rsp) movq %rsp, %rax movq %rax, 160(%rsp) leaq 216(%rsp), %rax movq %rax, 168(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) movl $1, 72(%rsp) movl $1, 76(%rsp) movl $1, 80(%rsp) movl $1, 84(%rsp) leaq 56(%rsp), %rcx leaq 48(%rsp), %rdx leaq 76(%rsp), %rsi leaq 64(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 184(%rsp), %rax subq %fs:40, %rax jne .L8 addq $200, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 56(%rsp) .cfi_def_cfa_offset 216 pushq 56(%rsp) .cfi_def_cfa_offset 224 leaq 128(%rsp), %r9 movq 92(%rsp), %rcx movl 100(%rsp), %r8d movq 80(%rsp), %rsi movl 88(%rsp), %edx leaq FindPixel(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 208 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2054: .size _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi, .-_Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi .globl FindPixel .type FindPixel, @function FindPixel: .LFB2055: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movl 24(%rsp), %eax pushq %rax .cfi_def_cfa_offset 24 pushq 24(%rsp) .cfi_def_cfa_offset 32 call _Z58__device_stub__Z9FindPixelP18ProgramGPUColorRGBiS0_iPiiPfiP18ProgramGPUColorRGBiS0_iPiiPfi addq $24, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2055: .size FindPixel, .-FindPixel .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "FindPixel" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2057: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq FindPixel(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2057: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "CUDAFYSOURCETEMP.hip" .globl __device_stub__FindPixel # -- Begin function __device_stub__FindPixel .p2align 4, 0x90 .type __device_stub__FindPixel,@function __device_stub__FindPixel: # @__device_stub__FindPixel .cfi_startproc # %bb.0: subq $168, %rsp .cfi_def_cfa_offset 176 movq %rdi, 88(%rsp) movl %esi, 20(%rsp) movq %rdx, 80(%rsp) movl %ecx, 16(%rsp) movq %r8, 72(%rsp) movl %r9d, 12(%rsp) leaq 88(%rsp), %rax movq %rax, 96(%rsp) leaq 20(%rsp), %rax movq %rax, 104(%rsp) leaq 80(%rsp), %rax movq %rax, 112(%rsp) leaq 16(%rsp), %rax movq %rax, 120(%rsp) leaq 72(%rsp), %rax movq %rax, 128(%rsp) leaq 12(%rsp), %rax movq %rax, 136(%rsp) leaq 176(%rsp), %rax movq %rax, 144(%rsp) leaq 184(%rsp), %rax movq %rax, 152(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 96(%rsp), %r9 movl $FindPixel, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $184, %rsp .cfi_adjust_cfa_offset -184 retq .Lfunc_end0: .size __device_stub__FindPixel, .Lfunc_end0-__device_stub__FindPixel .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $FindPixel, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type FindPixel,@object # @FindPixel .section .rodata,"a",@progbits .globl FindPixel .p2align 3, 0x0 FindPixel: .quad __device_stub__FindPixel .size FindPixel, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "FindPixel" .size .L__unnamed_1, 10 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __device_stub__FindPixel .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym FindPixel .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
#include <time.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <cuda_runtime.h> __global__ void GPuEuler(float *y, float t_0, float y_0 ,int N, float delta) { int myID = threadIdx.x + blockDim.x * blockIdx.x; if (myID <N) { y[myID] = y_0; for(int i=0;i<myID;i++){ float j_del = delta * i+ t_0; y[myID] =y[myID]+ delta * (9*j_del*j_del-4*j_del+5); } } } int main(int argc, char const *argv[]) { printf("seccion 1.b\n"); int hilos1b = 256,n1b,bloque1b; float delta_t1b,tiempoGPU1b; float *dev_e1b,*hst_y; cudaEvent_t start1b, end1b; for(int i=1;i<5;i++) { delta_t1b=powf(10,-i); n1b=10/delta_t1b +1; hst_y = (float*) malloc(n1b*sizeof(float)); bloque1b = ceil((float) n1b /hilos1b); cudaEventCreate(&start1b); cudaEventCreate(&end1b); cudaEventRecord(start1b,0); cudaMalloc( (void**) &dev_e1b, n1b*sizeof(float)); GPuEuler<<<bloque1b,hilos1b>>>(dev_e1b,0,4,n1b,delta_t1b); cudaEventRecord(end1b,0); cudaEventSynchronize(end1b); cudaEventElapsedTime(&tiempoGPU1b,start1b,end1b); cudaMemcpy(hst_y,dev_e1b,n1b*sizeof(float),cudaMemcpyDeviceToHost); cudaFree(dev_e1b); free(hst_y); cudaEventDestroy(start1b); cudaEventDestroy(end1b); printf("%f\n",tiempoGPU1b); } return 0; }
code for sm_80 Function : _Z8GPuEulerPfffif .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R5, SR_TID.X ; /* 0x0000000000057919 */ /* 0x000e280000002100 */ /*0020*/ S2R R0, SR_CTAID.X ; /* 0x0000000000007919 */ /* 0x000e240000002500 */ /*0030*/ IMAD R5, R0, c[0x0][0x0], R5 ; /* 0x0000000000057a24 */ /* 0x001fca00078e0205 */ /*0040*/ ISETP.GE.AND P0, PT, R5, c[0x0][0x170], PT ; /* 0x00005c0005007a0c */ /* 0x000fda0003f06270 */ /*0050*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0060*/ HFMA2.MMA R2, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff027435 */ /* 0x000fe200000001ff */ /*0070*/ ISETP.GE.AND P0, PT, R5, 0x1, PT ; /* 0x000000010500780c */ /* 0x000fe20003f06270 */ /*0080*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe20000000a00 */ /*0090*/ MOV R7, c[0x0][0x16c] ; /* 0x00005b0000077a02 */ /* 0x000fce0000000f00 */ /*00a0*/ IMAD.WIDE R2, R5, R2, c[0x0][0x160] ; /* 0x0000580005027625 */ /* 0x000fca00078e0202 */ /*00b0*/ STG.E [R2.64], R7 ; /* 0x0000000702007986 */ /* 0x0001e2000c101904 */ /*00c0*/ @!P0 EXIT ; /* 0x000000000000894d */ /* 0x000fea0003800000 */ /*00d0*/ IADD3 R0, R5.reuse, -0x1, RZ ; /* 0xffffffff05007810 */ /* 0x040fe20007ffe0ff */ /*00e0*/ BSSY B0, 0x3c0 ; /* 0x000002d000007945 */ /* 0x000fe20003800000 */ /*00f0*/ LOP3.LUT R4, R5, 0x3, RZ, 0xc0, !PT ; /* 0x0000000305047812 */ /* 0x000fe400078ec0ff */ /*0100*/ ISETP.GE.U32.AND P1, PT, R0, 0x3, PT ; /* 0x000000030000780c */ /* 0x000fe40003f26070 */ /*0110*/ ISETP.NE.AND P0, PT, R4, RZ, PT ; /* 0x000000ff0400720c */ /* 0x000fe40003f05270 */ /*0120*/ MOV R7, c[0x0][0x16c] ; /* 0x00005b0000077a02 */ /* 0x001fe40000000f00 */ /*0130*/ MOV R0, RZ ; /* 0x000000ff00007202 */ /* 0x000fce0000000f00 */ /*0140*/ @!P1 BRA 0x3b0 ; /* 0x0000026000009947 */ /* 0x000fea0003800000 */ /*0150*/ HFMA2.MMA R0, -RZ, RZ, 0, 0 ; /* 0x00000000ff007435 */ /* 0x000fe200000001ff */ /*0160*/ IADD3 R5, R5, -R4, RZ ; /* 0x8000000405057210 */ /* 0x000fd20007ffe0ff */ /*0170*/ IADD3 R10, R0.reuse, 0x1, RZ ; /* 0x00000001000a7810 */ /* 0x040fe20007ffe0ff */ /*0180*/ I2F R6, R0 ; /* 0x0000000000067306 */ /* 0x0000620000201400 */ /*0190*/ IADD3 R13, R0.reuse, 0x2, RZ ; /* 0x00000002000d7810 */ /* 0x040fe40007ffe0ff */ /*01a0*/ IADD3 R15, R0.reuse, 0x3, RZ ; /* 0x00000003000f7810 */ /* 0x040fe40007ffe0ff */ /*01b0*/ MOV R16, c[0x0][0x174] ; /* 0x00005d0000107a02 */ /* 0x000fe40000000f00 */ /*01c0*/ IADD3 R5, R5, -0x4, RZ ; /* 0xfffffffc05057810 */ /* 0x000fe20007ffe0ff */ /*01d0*/ I2F R10, R10 ; /* 0x0000000a000a7306 */ /* 0x000ea20000201400 */ /*01e0*/ IADD3 R0, R0, 0x4, RZ ; /* 0x0000000400007810 */ /* 0x001fc40007ffe0ff */ /*01f0*/ ISETP.NE.AND P1, PT, R5, RZ, PT ; /* 0x000000ff0500720c */ /* 0x000fca0003f25270 */ /*0200*/ I2F R13, R13 ; /* 0x0000000d000d7306 */ /* 0x000e220000201400 */ /*0210*/ FFMA R6, R6, R16, c[0x0][0x168] ; /* 0x00005a0006067623 */ /* 0x002fc80000000010 */ /*0220*/ FMUL R9, R6.reuse, 9 ; /* 0x4110000006097820 */ /* 0x040fe40000400000 */ /*0230*/ FMUL R8, R6, 4 ; /* 0x4080000006087820 */ /* 0x000fe20000400000 */ /*0240*/ I2F R15, R15 ; /* 0x0000000f000f7306 */ /* 0x000e620000201400 */ /*0250*/ FFMA R11, R10, R16, c[0x0][0x168] ; /* 0x00005a000a0b7623 */ /* 0x004fe40000000010 */ /*0260*/ FFMA R8, R6, R9, -R8 ; /* 0x0000000906087223 */ /* 0x000fe40000000808 */ /*0270*/ FMUL R12, R11.reuse, 9 ; /* 0x411000000b0c7820 */ /* 0x040fe40000400000 */ /*0280*/ FMUL R10, R11, 4 ; /* 0x408000000b0a7820 */ /* 0x000fc40000400000 */ /*0290*/ FFMA R14, R13, R16, c[0x0][0x168] ; /* 0x00005a000d0e7623 */ /* 0x001fe40000000010 */ /*02a0*/ FADD R8, R8, 5 ; /* 0x40a0000008087421 */ /* 0x000fe40000000000 */ /*02b0*/ FFMA R10, R11, R12, -R10 ; /* 0x0000000c0b0a7223 */ /* 0x000fe4000000080a */ /*02c0*/ FMUL R9, R14.reuse, 9 ; /* 0x411000000e097820 */ /* 0x040fe40000400000 */ /*02d0*/ FFMA R13, R15, R16, c[0x0][0x168] ; /* 0x00005a000f0d7623 */ /* 0x002fe40000000010 */ /*02e0*/ FMUL R6, R14, 4 ; /* 0x408000000e067820 */ /* 0x000fc40000400000 */ /*02f0*/ FFMA R8, R8, c[0x0][0x174], R7 ; /* 0x00005d0008087a23 */ /* 0x000fe40000000007 */ /*0300*/ FMUL R12, R13.reuse, 9 ; /* 0x411000000d0c7820 */ /* 0x040fe40000400000 */ /*0310*/ FMUL R11, R13.reuse, 4 ; /* 0x408000000d0b7820 */ /* 0x040fe40000400000 */ /*0320*/ FADD R7, R10, 5 ; /* 0x40a000000a077421 */ /* 0x000fe40000000000 */ /*0330*/ FFMA R6, R14, R9, -R6 ; /* 0x000000090e067223 */ /* 0x000fe40000000806 */ /*0340*/ FFMA R11, R13, R12, -R11 ; /* 0x0000000c0d0b7223 */ /* 0x000fc4000000080b */ /*0350*/ FFMA R7, R7, c[0x0][0x174], R8 ; /* 0x00005d0007077a23 */ /* 0x000fe40000000008 */ /*0360*/ FADD R6, R6, 5 ; /* 0x40a0000006067421 */ /* 0x000fe40000000000 */ /*0370*/ FADD R8, R11, 5 ; /* 0x40a000000b087421 */ /* 0x000fe40000000000 */ /*0380*/ FFMA R7, R6, c[0x0][0x174], R7 ; /* 0x00005d0006077a23 */ /* 0x000fc80000000007 */ /*0390*/ FFMA R7, R8, c[0x0][0x174], R7 ; /* 0x00005d0008077a23 */ /* 0x000fe20000000007 */ /*03a0*/ @P1 BRA 0x170 ; /* 0xfffffdc000001947 */ /* 0x000fea000383ffff */ /*03b0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*03c0*/ BSSY B0, 0x4b0 ; /* 0x000000e000007945 */ /* 0x000fe20003800000 */ /*03d0*/ @!P0 BRA 0x4a0 ; /* 0x000000c000008947 */ /* 0x000fea0003800000 */ /*03e0*/ MOV R8, c[0x0][0x174] ; /* 0x00005d0000087a02 */ /* 0x000fe40000000f00 */ /*03f0*/ I2F R5, R0 ; /* 0x0000000000057306 */ /* 0x0000620000201400 */ /*0400*/ IADD3 R4, R4, -0x1, RZ ; /* 0xffffffff04047810 */ /* 0x000fc80007ffe0ff */ /*0410*/ ISETP.NE.AND P0, PT, R4, RZ, PT ; /* 0x000000ff0400720c */ /* 0x000fe40003f05270 */ /*0420*/ IADD3 R0, R0, 0x1, RZ ; /* 0x0000000100007810 */ /* 0x001fe20007ffe0ff */ /*0430*/ FFMA R6, R5, R8, c[0x0][0x168] ; /* 0x00005a0005067623 */ /* 0x002fc80000000008 */ /*0440*/ FMUL R9, R6.reuse, 9 ; /* 0x4110000006097820 */ /* 0x040fe40000400000 */ /*0450*/ FMUL R5, R6, -4 ; /* 0xc080000006057820 */ /* 0x000fc80000400000 */ /*0460*/ FFMA R5, R6, R9, R5 ; /* 0x0000000906057223 */ /* 0x000fc80000000005 */ /*0470*/ FADD R6, R5, 5 ; /* 0x40a0000005067421 */ /* 0x000fc80000000000 */ /*0480*/ FFMA R7, R6, c[0x0][0x174], R7 ; /* 0x00005d0006077a23 */ /* 0x000fe20000000007 */ /*0490*/ @P0 BRA 0x3f0 ; /* 0xffffff5000000947 */ /* 0x000fea000383ffff */ /*04a0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*04b0*/ STG.E [R2.64], R7 ; /* 0x0000000702007986 */ /* 0x000fe2000c101904 */ /*04c0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*04d0*/ BRA 0x4d0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*04e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*04f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0500*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0510*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0520*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0530*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0540*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0550*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0560*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0570*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include <time.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <cuda_runtime.h> __global__ void GPuEuler(float *y, float t_0, float y_0 ,int N, float delta) { int myID = threadIdx.x + blockDim.x * blockIdx.x; if (myID <N) { y[myID] = y_0; for(int i=0;i<myID;i++){ float j_del = delta * i+ t_0; y[myID] =y[myID]+ delta * (9*j_del*j_del-4*j_del+5); } } } int main(int argc, char const *argv[]) { printf("seccion 1.b\n"); int hilos1b = 256,n1b,bloque1b; float delta_t1b,tiempoGPU1b; float *dev_e1b,*hst_y; cudaEvent_t start1b, end1b; for(int i=1;i<5;i++) { delta_t1b=powf(10,-i); n1b=10/delta_t1b +1; hst_y = (float*) malloc(n1b*sizeof(float)); bloque1b = ceil((float) n1b /hilos1b); cudaEventCreate(&start1b); cudaEventCreate(&end1b); cudaEventRecord(start1b,0); cudaMalloc( (void**) &dev_e1b, n1b*sizeof(float)); GPuEuler<<<bloque1b,hilos1b>>>(dev_e1b,0,4,n1b,delta_t1b); cudaEventRecord(end1b,0); cudaEventSynchronize(end1b); cudaEventElapsedTime(&tiempoGPU1b,start1b,end1b); cudaMemcpy(hst_y,dev_e1b,n1b*sizeof(float),cudaMemcpyDeviceToHost); cudaFree(dev_e1b); free(hst_y); cudaEventDestroy(start1b); cudaEventDestroy(end1b); printf("%f\n",tiempoGPU1b); } return 0; }
.file "tmpxft_001b53e4_00000000-6_pregunta1b.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2060: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2060: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z31__device_stub__Z8GPuEulerPfffifPfffif .type _Z31__device_stub__Z8GPuEulerPfffifPfffif, @function _Z31__device_stub__Z8GPuEulerPfffifPfffif: .LFB2082: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movss %xmm0, 20(%rsp) movss %xmm1, 16(%rsp) movl %esi, 12(%rsp) movss %xmm2, 8(%rsp) movq %fs:40, %rax movq %rax, 136(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 20(%rsp), %rax movq %rax, 104(%rsp) leaq 16(%rsp), %rax movq %rax, 112(%rsp) leaq 12(%rsp), %rax movq %rax, 120(%rsp) leaq 8(%rsp), %rax movq %rax, 128(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 136(%rsp), %rax subq %fs:40, %rax jne .L8 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 168 pushq 40(%rsp) .cfi_def_cfa_offset 176 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z8GPuEulerPfffif(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2082: .size _Z31__device_stub__Z8GPuEulerPfffifPfffif, .-_Z31__device_stub__Z8GPuEulerPfffifPfffif .globl _Z8GPuEulerPfffif .type _Z8GPuEulerPfffif, @function _Z8GPuEulerPfffif: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z31__device_stub__Z8GPuEulerPfffifPfffif addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z8GPuEulerPfffif, .-_Z8GPuEulerPfffif .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "seccion 1.b\n" .LC8: .string "%f\n" .text .globl main .type main, @function main: .LFB2057: .cfi_startproc endbr64 pushq %r13 .cfi_def_cfa_offset 16 .cfi_offset 13, -16 pushq %r12 .cfi_def_cfa_offset 24 .cfi_offset 12, -24 pushq %rbp .cfi_def_cfa_offset 32 .cfi_offset 6, -32 pushq %rbx .cfi_def_cfa_offset 40 .cfi_offset 3, -40 subq $88, %rsp .cfi_def_cfa_offset 128 movq %fs:40, %rax movq %rax, 72(%rsp) xorl %eax, %eax leaq .LC0(%rip), %rsi movl $2, %edi call __printf_chk@PLT movl $-1, %r12d jmp .L14 .L12: leaq 32(%rsp), %rdi call cudaEventCreate@PLT leaq 40(%rsp), %rdi call cudaEventCreate@PLT movl $0, %esi movq 32(%rsp), %rdi call cudaEventRecord@PLT leaq 24(%rsp), %rdi movq %rbx, %rsi call cudaMalloc@PLT movl $256, 60(%rsp) movl $1, 64(%rsp) cvttss2sil 8(%rsp), %eax movl %eax, 48(%rsp) movl $1, 52(%rsp) movl $0, %r9d movl $0, %r8d movq 60(%rsp), %rdx movl $1, %ecx movq 48(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L18 .L13: movl $0, %esi movq 40(%rsp), %rdi call cudaEventRecord@PLT movq 40(%rsp), %rdi call cudaEventSynchronize@PLT leaq 20(%rsp), %rdi movq 40(%rsp), %rdx movq 32(%rsp), %rsi call cudaEventElapsedTime@PLT movl $2, %ecx movq %rbx, %rdx movq 24(%rsp), %rsi movq %rbp, %rdi call cudaMemcpy@PLT movq 24(%rsp), %rdi call cudaFree@PLT movq %rbp, %rdi call free@PLT movq 32(%rsp), %rdi call cudaEventDestroy@PLT movq 40(%rsp), %rdi call cudaEventDestroy@PLT pxor %xmm0, %xmm0 cvtss2sd 20(%rsp), %xmm0 leaq .LC8(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT subl $1, %r12d cmpl $-5, %r12d je .L19 .L14: pxor %xmm1, %xmm1 cvtsi2ssl %r12d, %xmm1 movss .LC1(%rip), %xmm0 call powf@PLT movss %xmm0, 12(%rsp) movss .LC1(%rip), %xmm4 divss %xmm0, %xmm4 movaps %xmm4, %xmm0 addss .LC2(%rip), %xmm0 cvttss2sil %xmm0, %r13d movslq %r13d, %rbx salq $2, %rbx movq %rbx, %rdi call malloc@PLT movq %rax, %rbp pxor %xmm0, %xmm0 cvtsi2ssl %r13d, %xmm0 mulss .LC3(%rip), %xmm0 movss %xmm0, 8(%rsp) movss .LC9(%rip), %xmm1 andps %xmm0, %xmm1 movss .LC4(%rip), %xmm2 ucomiss %xmm1, %xmm2 jbe .L12 cvttss2sil %xmm0, %eax pxor %xmm1, %xmm1 cvtsi2ssl %eax, %xmm1 movaps %xmm0, %xmm2 cmpnless %xmm1, %xmm2 movss .LC2(%rip), %xmm3 andps %xmm3, %xmm2 addss %xmm2, %xmm1 movss .LC9(%rip), %xmm2 andnps %xmm0, %xmm2 orps %xmm2, %xmm1 movss %xmm1, 8(%rsp) jmp .L12 .L18: movss 12(%rsp), %xmm2 movl %r13d, %esi movss .LC6(%rip), %xmm1 pxor %xmm0, %xmm0 movq 24(%rsp), %rdi call _Z31__device_stub__Z8GPuEulerPfffifPfffif jmp .L13 .L19: movq 72(%rsp), %rax subq %fs:40, %rax jne .L20 movl $0, %eax addq $88, %rsp .cfi_remember_state .cfi_def_cfa_offset 40 popq %rbx .cfi_def_cfa_offset 32 popq %rbp .cfi_def_cfa_offset 24 popq %r12 .cfi_def_cfa_offset 16 popq %r13 .cfi_def_cfa_offset 8 ret .L20: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1 .LC10: .string "_Z8GPuEulerPfffif" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2085: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC10(%rip), %rdx movq %rdx, %rcx leaq _Z8GPuEulerPfffif(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2085: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC1: .long 1092616192 .align 4 .LC2: .long 1065353216 .align 4 .LC3: .long 998244352 .align 4 .LC4: .long 1258291200 .align 4 .LC6: .long 1082130432 .align 4 .LC9: .long 2147483647 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include <time.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <cuda_runtime.h> __global__ void GPuEuler(float *y, float t_0, float y_0 ,int N, float delta) { int myID = threadIdx.x + blockDim.x * blockIdx.x; if (myID <N) { y[myID] = y_0; for(int i=0;i<myID;i++){ float j_del = delta * i+ t_0; y[myID] =y[myID]+ delta * (9*j_del*j_del-4*j_del+5); } } } int main(int argc, char const *argv[]) { printf("seccion 1.b\n"); int hilos1b = 256,n1b,bloque1b; float delta_t1b,tiempoGPU1b; float *dev_e1b,*hst_y; cudaEvent_t start1b, end1b; for(int i=1;i<5;i++) { delta_t1b=powf(10,-i); n1b=10/delta_t1b +1; hst_y = (float*) malloc(n1b*sizeof(float)); bloque1b = ceil((float) n1b /hilos1b); cudaEventCreate(&start1b); cudaEventCreate(&end1b); cudaEventRecord(start1b,0); cudaMalloc( (void**) &dev_e1b, n1b*sizeof(float)); GPuEuler<<<bloque1b,hilos1b>>>(dev_e1b,0,4,n1b,delta_t1b); cudaEventRecord(end1b,0); cudaEventSynchronize(end1b); cudaEventElapsedTime(&tiempoGPU1b,start1b,end1b); cudaMemcpy(hst_y,dev_e1b,n1b*sizeof(float),cudaMemcpyDeviceToHost); cudaFree(dev_e1b); free(hst_y); cudaEventDestroy(start1b); cudaEventDestroy(end1b); printf("%f\n",tiempoGPU1b); } return 0; }
#include <time.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <hip/hip_runtime.h> __global__ void GPuEuler(float *y, float t_0, float y_0 ,int N, float delta) { int myID = threadIdx.x + blockDim.x * blockIdx.x; if (myID <N) { y[myID] = y_0; for(int i=0;i<myID;i++){ float j_del = delta * i+ t_0; y[myID] =y[myID]+ delta * (9*j_del*j_del-4*j_del+5); } } } int main(int argc, char const *argv[]) { printf("seccion 1.b\n"); int hilos1b = 256,n1b,bloque1b; float delta_t1b,tiempoGPU1b; float *dev_e1b,*hst_y; hipEvent_t start1b, end1b; for(int i=1;i<5;i++) { delta_t1b=powf(10,-i); n1b=10/delta_t1b +1; hst_y = (float*) malloc(n1b*sizeof(float)); bloque1b = ceil((float) n1b /hilos1b); hipEventCreate(&start1b); hipEventCreate(&end1b); hipEventRecord(start1b,0); hipMalloc( (void**) &dev_e1b, n1b*sizeof(float)); GPuEuler<<<bloque1b,hilos1b>>>(dev_e1b,0,4,n1b,delta_t1b); hipEventRecord(end1b,0); hipEventSynchronize(end1b); hipEventElapsedTime(&tiempoGPU1b,start1b,end1b); hipMemcpy(hst_y,dev_e1b,n1b*sizeof(float),hipMemcpyDeviceToHost); hipFree(dev_e1b); free(hst_y); hipEventDestroy(start1b); hipEventDestroy(end1b); printf("%f\n",tiempoGPU1b); } return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <time.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <hip/hip_runtime.h> __global__ void GPuEuler(float *y, float t_0, float y_0 ,int N, float delta) { int myID = threadIdx.x + blockDim.x * blockIdx.x; if (myID <N) { y[myID] = y_0; for(int i=0;i<myID;i++){ float j_del = delta * i+ t_0; y[myID] =y[myID]+ delta * (9*j_del*j_del-4*j_del+5); } } } int main(int argc, char const *argv[]) { printf("seccion 1.b\n"); int hilos1b = 256,n1b,bloque1b; float delta_t1b,tiempoGPU1b; float *dev_e1b,*hst_y; hipEvent_t start1b, end1b; for(int i=1;i<5;i++) { delta_t1b=powf(10,-i); n1b=10/delta_t1b +1; hst_y = (float*) malloc(n1b*sizeof(float)); bloque1b = ceil((float) n1b /hilos1b); hipEventCreate(&start1b); hipEventCreate(&end1b); hipEventRecord(start1b,0); hipMalloc( (void**) &dev_e1b, n1b*sizeof(float)); GPuEuler<<<bloque1b,hilos1b>>>(dev_e1b,0,4,n1b,delta_t1b); hipEventRecord(end1b,0); hipEventSynchronize(end1b); hipEventElapsedTime(&tiempoGPU1b,start1b,end1b); hipMemcpy(hst_y,dev_e1b,n1b*sizeof(float),hipMemcpyDeviceToHost); hipFree(dev_e1b); free(hst_y); hipEventDestroy(start1b); hipEventDestroy(end1b); printf("%f\n",tiempoGPU1b); } return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z8GPuEulerPfffif .globl _Z8GPuEulerPfffif .p2align 8 .type _Z8GPuEulerPfffif,@function _Z8GPuEulerPfffif: s_clause 0x1 s_load_b32 s2, s[0:1], 0x24 s_load_b32 s3, s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_and_b32 s2, s2, 0xffff s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_mad_u64_u32 v[1:2], null, s15, s2, v[0:1] s_mov_b32 s2, exec_lo v_cmpx_gt_i32_e64 s3, v1 s_cbranch_execz .LBB0_5 s_clause 0x1 s_load_b64 s[2:3], s[0:1], 0x0 s_load_b32 s4, s[0:1], 0xc v_ashrrev_i32_e32 v2, 31, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[2:3], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v2, vcc_lo, s2, v2 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v3, vcc_lo, s3, v3, vcc_lo v_mov_b32_e32 v0, s4 v_cmp_lt_i32_e32 vcc_lo, 0, v1 s_mov_b32 s2, 0 global_store_b32 v[2:3], v0, off s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_5 global_load_b32 v0, v[2:3], off s_clause 0x1 s_load_b32 s3, s[0:1], 0x8 s_load_b32 s0, s[0:1], 0x14 s_mov_b32 s1, 0 .LBB0_3: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_cvt_f32_i32_e32 v4, s1 s_add_i32 s1, s1, 1 v_cmp_eq_u32_e32 vcc_lo, s1, v1 s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_fma_f32 v4, v4, s0, s3 s_or_b32 s2, vcc_lo, s2 v_mul_f32_e32 v5, 0x41100000, v4 v_mul_f32_e32 v6, -4.0, v4 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fmac_f32_e32 v6, v4, v5 v_add_f32_e32 v4, 0x40a00000, v6 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_1) v_fmac_f32_e32 v0, s0, v4 s_and_not1_b32 exec_lo, exec_lo, s2 s_cbranch_execnz .LBB0_3 s_or_b32 exec_lo, exec_lo, s2 global_store_b32 v[2:3], v0, off .LBB0_5: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z8GPuEulerPfffif .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 280 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 7 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z8GPuEulerPfffif, .Lfunc_end0-_Z8GPuEulerPfffif .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .offset: 8 .size: 4 .value_kind: by_value - .offset: 12 .size: 4 .value_kind: by_value - .offset: 16 .size: 4 .value_kind: by_value - .offset: 20 .size: 4 .value_kind: by_value - .offset: 24 .size: 4 .value_kind: hidden_block_count_x - .offset: 28 .size: 4 .value_kind: hidden_block_count_y - .offset: 32 .size: 4 .value_kind: hidden_block_count_z - .offset: 36 .size: 2 .value_kind: hidden_group_size_x - .offset: 38 .size: 2 .value_kind: hidden_group_size_y - .offset: 40 .size: 2 .value_kind: hidden_group_size_z - .offset: 42 .size: 2 .value_kind: hidden_remainder_x - .offset: 44 .size: 2 .value_kind: hidden_remainder_y - .offset: 46 .size: 2 .value_kind: hidden_remainder_z - .offset: 64 .size: 8 .value_kind: hidden_global_offset_x - .offset: 72 .size: 8 .value_kind: hidden_global_offset_y - .offset: 80 .size: 8 .value_kind: hidden_global_offset_z - .offset: 88 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 280 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z8GPuEulerPfffif .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z8GPuEulerPfffif.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 7 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <time.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <hip/hip_runtime.h> __global__ void GPuEuler(float *y, float t_0, float y_0 ,int N, float delta) { int myID = threadIdx.x + blockDim.x * blockIdx.x; if (myID <N) { y[myID] = y_0; for(int i=0;i<myID;i++){ float j_del = delta * i+ t_0; y[myID] =y[myID]+ delta * (9*j_del*j_del-4*j_del+5); } } } int main(int argc, char const *argv[]) { printf("seccion 1.b\n"); int hilos1b = 256,n1b,bloque1b; float delta_t1b,tiempoGPU1b; float *dev_e1b,*hst_y; hipEvent_t start1b, end1b; for(int i=1;i<5;i++) { delta_t1b=powf(10,-i); n1b=10/delta_t1b +1; hst_y = (float*) malloc(n1b*sizeof(float)); bloque1b = ceil((float) n1b /hilos1b); hipEventCreate(&start1b); hipEventCreate(&end1b); hipEventRecord(start1b,0); hipMalloc( (void**) &dev_e1b, n1b*sizeof(float)); GPuEuler<<<bloque1b,hilos1b>>>(dev_e1b,0,4,n1b,delta_t1b); hipEventRecord(end1b,0); hipEventSynchronize(end1b); hipEventElapsedTime(&tiempoGPU1b,start1b,end1b); hipMemcpy(hst_y,dev_e1b,n1b*sizeof(float),hipMemcpyDeviceToHost); hipFree(dev_e1b); free(hst_y); hipEventDestroy(start1b); hipEventDestroy(end1b); printf("%f\n",tiempoGPU1b); } return 0; }
.text .file "pregunta1b.hip" .globl _Z23__device_stub__GPuEulerPfffif # -- Begin function _Z23__device_stub__GPuEulerPfffif .p2align 4, 0x90 .type _Z23__device_stub__GPuEulerPfffif,@function _Z23__device_stub__GPuEulerPfffif: # @_Z23__device_stub__GPuEulerPfffif .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movss %xmm0, 20(%rsp) movss %xmm1, 16(%rsp) movl %esi, 12(%rsp) movss %xmm2, 8(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 20(%rsp), %rax movq %rax, 88(%rsp) leaq 16(%rsp), %rax movq %rax, 96(%rsp) leaq 12(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z8GPuEulerPfffif, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $136, %rsp .cfi_adjust_cfa_offset -136 retq .Lfunc_end0: .size _Z23__device_stub__GPuEulerPfffif, .Lfunc_end0-_Z23__device_stub__GPuEulerPfffif .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function main .LCPI1_0: .long 0x41200000 # float 10 .LCPI1_1: .long 0x3f800000 # float 1 .LCPI1_2: .long 0x3b800000 # float 0.00390625 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $184, %rsp .cfi_def_cfa_offset 240 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movabsq $4294967296, %rbx # imm = 0x100000000 movl $.Lstr, %edi callq puts@PLT movl $-1, %r15d leaq 256(%rbx), %r12 jmp .LBB1_1 .p2align 4, 0x90 .LBB1_3: # in Loop: Header=BB1_1 Depth=1 movq 8(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq 8(%rsp), %rdi callq hipEventSynchronize movq 16(%rsp), %rsi movq 8(%rsp), %rdx leaq 36(%rsp), %rdi callq hipEventElapsedTime movq 24(%rsp), %rsi movq %r13, %rdi movq %rbx, %rdx movl $2, %ecx callq hipMemcpy movq 24(%rsp), %rdi callq hipFree movq %r13, %rdi callq free movq 16(%rsp), %rdi callq hipEventDestroy movq 8(%rsp), %rdi callq hipEventDestroy movss 36(%rsp), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str.1, %edi movb $1, %al callq printf decl %r15d cmpl $-5, %r15d je .LBB1_4 .LBB1_1: # =>This Inner Loop Header: Depth=1 cvtsi2ss %r15d, %xmm1 movss .LCPI1_0(%rip), %xmm0 # xmm0 = mem[0],zero,zero,zero callq powf movss .LCPI1_0(%rip), %xmm1 # xmm1 = mem[0],zero,zero,zero movss %xmm0, 32(%rsp) # 4-byte Spill divss %xmm0, %xmm1 addss .LCPI1_1(%rip), %xmm1 movaps %xmm1, 160(%rsp) # 16-byte Spill cvttss2si %xmm1, %ebp movslq %ebp, %rbx shlq $2, %rbx movq %rbx, %rdi callq malloc movq %rax, %r13 cvttps2dq 160(%rsp), %xmm0 # 16-byte Folded Reload cvtdq2ps %xmm0, %xmm0 mulss .LCPI1_2(%rip), %xmm0 callq ceilf@PLT cvttss2si %xmm0, %r14d leaq 16(%rsp), %rdi callq hipEventCreate leaq 8(%rsp), %rdi callq hipEventCreate movq 16(%rsp), %rdi xorl %esi, %esi callq hipEventRecord leaq 24(%rsp), %rdi movq %rbx, %rsi callq hipMalloc movabsq $4294967296, %rax # imm = 0x100000000 orq %rax, %r14 movq %r14, %rdi movl $1, %esi movq %r12, %rdx movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_3 # %bb.2: # in Loop: Header=BB1_1 Depth=1 movq 24(%rsp), %rax movq %rax, 104(%rsp) movl $0, 52(%rsp) movl $1082130432, 48(%rsp) # imm = 0x40800000 movl %ebp, 44(%rsp) movss 32(%rsp), %xmm0 # 4-byte Reload # xmm0 = mem[0],zero,zero,zero movss %xmm0, 40(%rsp) leaq 104(%rsp), %rax movq %rax, 112(%rsp) leaq 52(%rsp), %rax movq %rax, 120(%rsp) leaq 48(%rsp), %rax movq %rax, 128(%rsp) leaq 44(%rsp), %rax movq %rax, 136(%rsp) leaq 40(%rsp), %rax movq %rax, 144(%rsp) leaq 88(%rsp), %rdi leaq 72(%rsp), %rsi leaq 64(%rsp), %rdx leaq 56(%rsp), %rcx callq __hipPopCallConfiguration movq 88(%rsp), %rsi movl 96(%rsp), %edx movq 72(%rsp), %rcx movl 80(%rsp), %r8d movl $_Z8GPuEulerPfffif, %edi leaq 112(%rsp), %r9 pushq 56(%rsp) .cfi_adjust_cfa_offset 8 pushq 72(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 jmp .LBB1_3 .LBB1_4: xorl %eax, %eax addq $184, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z8GPuEulerPfffif, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z8GPuEulerPfffif,@object # @_Z8GPuEulerPfffif .section .rodata,"a",@progbits .globl _Z8GPuEulerPfffif .p2align 3, 0x0 _Z8GPuEulerPfffif: .quad _Z23__device_stub__GPuEulerPfffif .size _Z8GPuEulerPfffif, 8 .type .L.str.1,@object # @.str.1 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.1: .asciz "%f\n" .size .L.str.1, 4 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z8GPuEulerPfffif" .size .L__unnamed_1, 18 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "seccion 1.b" .size .Lstr, 12 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z23__device_stub__GPuEulerPfffif .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z8GPuEulerPfffif .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z8GPuEulerPfffif .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R5, SR_TID.X ; /* 0x0000000000057919 */ /* 0x000e280000002100 */ /*0020*/ S2R R0, SR_CTAID.X ; /* 0x0000000000007919 */ /* 0x000e240000002500 */ /*0030*/ IMAD R5, R0, c[0x0][0x0], R5 ; /* 0x0000000000057a24 */ /* 0x001fca00078e0205 */ /*0040*/ ISETP.GE.AND P0, PT, R5, c[0x0][0x170], PT ; /* 0x00005c0005007a0c */ /* 0x000fda0003f06270 */ /*0050*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0060*/ HFMA2.MMA R2, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff027435 */ /* 0x000fe200000001ff */ /*0070*/ ISETP.GE.AND P0, PT, R5, 0x1, PT ; /* 0x000000010500780c */ /* 0x000fe20003f06270 */ /*0080*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe20000000a00 */ /*0090*/ MOV R7, c[0x0][0x16c] ; /* 0x00005b0000077a02 */ /* 0x000fce0000000f00 */ /*00a0*/ IMAD.WIDE R2, R5, R2, c[0x0][0x160] ; /* 0x0000580005027625 */ /* 0x000fca00078e0202 */ /*00b0*/ STG.E [R2.64], R7 ; /* 0x0000000702007986 */ /* 0x0001e2000c101904 */ /*00c0*/ @!P0 EXIT ; /* 0x000000000000894d */ /* 0x000fea0003800000 */ /*00d0*/ IADD3 R0, R5.reuse, -0x1, RZ ; /* 0xffffffff05007810 */ /* 0x040fe20007ffe0ff */ /*00e0*/ BSSY B0, 0x3c0 ; /* 0x000002d000007945 */ /* 0x000fe20003800000 */ /*00f0*/ LOP3.LUT R4, R5, 0x3, RZ, 0xc0, !PT ; /* 0x0000000305047812 */ /* 0x000fe400078ec0ff */ /*0100*/ ISETP.GE.U32.AND P1, PT, R0, 0x3, PT ; /* 0x000000030000780c */ /* 0x000fe40003f26070 */ /*0110*/ ISETP.NE.AND P0, PT, R4, RZ, PT ; /* 0x000000ff0400720c */ /* 0x000fe40003f05270 */ /*0120*/ MOV R7, c[0x0][0x16c] ; /* 0x00005b0000077a02 */ /* 0x001fe40000000f00 */ /*0130*/ MOV R0, RZ ; /* 0x000000ff00007202 */ /* 0x000fce0000000f00 */ /*0140*/ @!P1 BRA 0x3b0 ; /* 0x0000026000009947 */ /* 0x000fea0003800000 */ /*0150*/ HFMA2.MMA R0, -RZ, RZ, 0, 0 ; /* 0x00000000ff007435 */ /* 0x000fe200000001ff */ /*0160*/ IADD3 R5, R5, -R4, RZ ; /* 0x8000000405057210 */ /* 0x000fd20007ffe0ff */ /*0170*/ IADD3 R10, R0.reuse, 0x1, RZ ; /* 0x00000001000a7810 */ /* 0x040fe20007ffe0ff */ /*0180*/ I2F R6, R0 ; /* 0x0000000000067306 */ /* 0x0000620000201400 */ /*0190*/ IADD3 R13, R0.reuse, 0x2, RZ ; /* 0x00000002000d7810 */ /* 0x040fe40007ffe0ff */ /*01a0*/ IADD3 R15, R0.reuse, 0x3, RZ ; /* 0x00000003000f7810 */ /* 0x040fe40007ffe0ff */ /*01b0*/ MOV R16, c[0x0][0x174] ; /* 0x00005d0000107a02 */ /* 0x000fe40000000f00 */ /*01c0*/ IADD3 R5, R5, -0x4, RZ ; /* 0xfffffffc05057810 */ /* 0x000fe20007ffe0ff */ /*01d0*/ I2F R10, R10 ; /* 0x0000000a000a7306 */ /* 0x000ea20000201400 */ /*01e0*/ IADD3 R0, R0, 0x4, RZ ; /* 0x0000000400007810 */ /* 0x001fc40007ffe0ff */ /*01f0*/ ISETP.NE.AND P1, PT, R5, RZ, PT ; /* 0x000000ff0500720c */ /* 0x000fca0003f25270 */ /*0200*/ I2F R13, R13 ; /* 0x0000000d000d7306 */ /* 0x000e220000201400 */ /*0210*/ FFMA R6, R6, R16, c[0x0][0x168] ; /* 0x00005a0006067623 */ /* 0x002fc80000000010 */ /*0220*/ FMUL R9, R6.reuse, 9 ; /* 0x4110000006097820 */ /* 0x040fe40000400000 */ /*0230*/ FMUL R8, R6, 4 ; /* 0x4080000006087820 */ /* 0x000fe20000400000 */ /*0240*/ I2F R15, R15 ; /* 0x0000000f000f7306 */ /* 0x000e620000201400 */ /*0250*/ FFMA R11, R10, R16, c[0x0][0x168] ; /* 0x00005a000a0b7623 */ /* 0x004fe40000000010 */ /*0260*/ FFMA R8, R6, R9, -R8 ; /* 0x0000000906087223 */ /* 0x000fe40000000808 */ /*0270*/ FMUL R12, R11.reuse, 9 ; /* 0x411000000b0c7820 */ /* 0x040fe40000400000 */ /*0280*/ FMUL R10, R11, 4 ; /* 0x408000000b0a7820 */ /* 0x000fc40000400000 */ /*0290*/ FFMA R14, R13, R16, c[0x0][0x168] ; /* 0x00005a000d0e7623 */ /* 0x001fe40000000010 */ /*02a0*/ FADD R8, R8, 5 ; /* 0x40a0000008087421 */ /* 0x000fe40000000000 */ /*02b0*/ FFMA R10, R11, R12, -R10 ; /* 0x0000000c0b0a7223 */ /* 0x000fe4000000080a */ /*02c0*/ FMUL R9, R14.reuse, 9 ; /* 0x411000000e097820 */ /* 0x040fe40000400000 */ /*02d0*/ FFMA R13, R15, R16, c[0x0][0x168] ; /* 0x00005a000f0d7623 */ /* 0x002fe40000000010 */ /*02e0*/ FMUL R6, R14, 4 ; /* 0x408000000e067820 */ /* 0x000fc40000400000 */ /*02f0*/ FFMA R8, R8, c[0x0][0x174], R7 ; /* 0x00005d0008087a23 */ /* 0x000fe40000000007 */ /*0300*/ FMUL R12, R13.reuse, 9 ; /* 0x411000000d0c7820 */ /* 0x040fe40000400000 */ /*0310*/ FMUL R11, R13.reuse, 4 ; /* 0x408000000d0b7820 */ /* 0x040fe40000400000 */ /*0320*/ FADD R7, R10, 5 ; /* 0x40a000000a077421 */ /* 0x000fe40000000000 */ /*0330*/ FFMA R6, R14, R9, -R6 ; /* 0x000000090e067223 */ /* 0x000fe40000000806 */ /*0340*/ FFMA R11, R13, R12, -R11 ; /* 0x0000000c0d0b7223 */ /* 0x000fc4000000080b */ /*0350*/ FFMA R7, R7, c[0x0][0x174], R8 ; /* 0x00005d0007077a23 */ /* 0x000fe40000000008 */ /*0360*/ FADD R6, R6, 5 ; /* 0x40a0000006067421 */ /* 0x000fe40000000000 */ /*0370*/ FADD R8, R11, 5 ; /* 0x40a000000b087421 */ /* 0x000fe40000000000 */ /*0380*/ FFMA R7, R6, c[0x0][0x174], R7 ; /* 0x00005d0006077a23 */ /* 0x000fc80000000007 */ /*0390*/ FFMA R7, R8, c[0x0][0x174], R7 ; /* 0x00005d0008077a23 */ /* 0x000fe20000000007 */ /*03a0*/ @P1 BRA 0x170 ; /* 0xfffffdc000001947 */ /* 0x000fea000383ffff */ /*03b0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*03c0*/ BSSY B0, 0x4b0 ; /* 0x000000e000007945 */ /* 0x000fe20003800000 */ /*03d0*/ @!P0 BRA 0x4a0 ; /* 0x000000c000008947 */ /* 0x000fea0003800000 */ /*03e0*/ MOV R8, c[0x0][0x174] ; /* 0x00005d0000087a02 */ /* 0x000fe40000000f00 */ /*03f0*/ I2F R5, R0 ; /* 0x0000000000057306 */ /* 0x0000620000201400 */ /*0400*/ IADD3 R4, R4, -0x1, RZ ; /* 0xffffffff04047810 */ /* 0x000fc80007ffe0ff */ /*0410*/ ISETP.NE.AND P0, PT, R4, RZ, PT ; /* 0x000000ff0400720c */ /* 0x000fe40003f05270 */ /*0420*/ IADD3 R0, R0, 0x1, RZ ; /* 0x0000000100007810 */ /* 0x001fe20007ffe0ff */ /*0430*/ FFMA R6, R5, R8, c[0x0][0x168] ; /* 0x00005a0005067623 */ /* 0x002fc80000000008 */ /*0440*/ FMUL R9, R6.reuse, 9 ; /* 0x4110000006097820 */ /* 0x040fe40000400000 */ /*0450*/ FMUL R5, R6, -4 ; /* 0xc080000006057820 */ /* 0x000fc80000400000 */ /*0460*/ FFMA R5, R6, R9, R5 ; /* 0x0000000906057223 */ /* 0x000fc80000000005 */ /*0470*/ FADD R6, R5, 5 ; /* 0x40a0000005067421 */ /* 0x000fc80000000000 */ /*0480*/ FFMA R7, R6, c[0x0][0x174], R7 ; /* 0x00005d0006077a23 */ /* 0x000fe20000000007 */ /*0490*/ @P0 BRA 0x3f0 ; /* 0xffffff5000000947 */ /* 0x000fea000383ffff */ /*04a0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*04b0*/ STG.E [R2.64], R7 ; /* 0x0000000702007986 */ /* 0x000fe2000c101904 */ /*04c0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*04d0*/ BRA 0x4d0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*04e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*04f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0500*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0510*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0520*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0530*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0540*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0550*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0560*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0570*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z8GPuEulerPfffif .globl _Z8GPuEulerPfffif .p2align 8 .type _Z8GPuEulerPfffif,@function _Z8GPuEulerPfffif: s_clause 0x1 s_load_b32 s2, s[0:1], 0x24 s_load_b32 s3, s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_and_b32 s2, s2, 0xffff s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_mad_u64_u32 v[1:2], null, s15, s2, v[0:1] s_mov_b32 s2, exec_lo v_cmpx_gt_i32_e64 s3, v1 s_cbranch_execz .LBB0_5 s_clause 0x1 s_load_b64 s[2:3], s[0:1], 0x0 s_load_b32 s4, s[0:1], 0xc v_ashrrev_i32_e32 v2, 31, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[2:3], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v2, vcc_lo, s2, v2 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v3, vcc_lo, s3, v3, vcc_lo v_mov_b32_e32 v0, s4 v_cmp_lt_i32_e32 vcc_lo, 0, v1 s_mov_b32 s2, 0 global_store_b32 v[2:3], v0, off s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_5 global_load_b32 v0, v[2:3], off s_clause 0x1 s_load_b32 s3, s[0:1], 0x8 s_load_b32 s0, s[0:1], 0x14 s_mov_b32 s1, 0 .LBB0_3: s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_cvt_f32_i32_e32 v4, s1 s_add_i32 s1, s1, 1 v_cmp_eq_u32_e32 vcc_lo, s1, v1 s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_fma_f32 v4, v4, s0, s3 s_or_b32 s2, vcc_lo, s2 v_mul_f32_e32 v5, 0x41100000, v4 v_mul_f32_e32 v6, -4.0, v4 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fmac_f32_e32 v6, v4, v5 v_add_f32_e32 v4, 0x40a00000, v6 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_1) v_fmac_f32_e32 v0, s0, v4 s_and_not1_b32 exec_lo, exec_lo, s2 s_cbranch_execnz .LBB0_3 s_or_b32 exec_lo, exec_lo, s2 global_store_b32 v[2:3], v0, off .LBB0_5: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z8GPuEulerPfffif .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 280 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 7 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z8GPuEulerPfffif, .Lfunc_end0-_Z8GPuEulerPfffif .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .offset: 8 .size: 4 .value_kind: by_value - .offset: 12 .size: 4 .value_kind: by_value - .offset: 16 .size: 4 .value_kind: by_value - .offset: 20 .size: 4 .value_kind: by_value - .offset: 24 .size: 4 .value_kind: hidden_block_count_x - .offset: 28 .size: 4 .value_kind: hidden_block_count_y - .offset: 32 .size: 4 .value_kind: hidden_block_count_z - .offset: 36 .size: 2 .value_kind: hidden_group_size_x - .offset: 38 .size: 2 .value_kind: hidden_group_size_y - .offset: 40 .size: 2 .value_kind: hidden_group_size_z - .offset: 42 .size: 2 .value_kind: hidden_remainder_x - .offset: 44 .size: 2 .value_kind: hidden_remainder_y - .offset: 46 .size: 2 .value_kind: hidden_remainder_z - .offset: 64 .size: 8 .value_kind: hidden_global_offset_x - .offset: 72 .size: 8 .value_kind: hidden_global_offset_y - .offset: 80 .size: 8 .value_kind: hidden_global_offset_z - .offset: 88 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 280 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z8GPuEulerPfffif .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z8GPuEulerPfffif.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 7 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_001b53e4_00000000-6_pregunta1b.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2060: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2060: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z31__device_stub__Z8GPuEulerPfffifPfffif .type _Z31__device_stub__Z8GPuEulerPfffifPfffif, @function _Z31__device_stub__Z8GPuEulerPfffifPfffif: .LFB2082: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movss %xmm0, 20(%rsp) movss %xmm1, 16(%rsp) movl %esi, 12(%rsp) movss %xmm2, 8(%rsp) movq %fs:40, %rax movq %rax, 136(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 20(%rsp), %rax movq %rax, 104(%rsp) leaq 16(%rsp), %rax movq %rax, 112(%rsp) leaq 12(%rsp), %rax movq %rax, 120(%rsp) leaq 8(%rsp), %rax movq %rax, 128(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 136(%rsp), %rax subq %fs:40, %rax jne .L8 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 168 pushq 40(%rsp) .cfi_def_cfa_offset 176 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z8GPuEulerPfffif(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2082: .size _Z31__device_stub__Z8GPuEulerPfffifPfffif, .-_Z31__device_stub__Z8GPuEulerPfffifPfffif .globl _Z8GPuEulerPfffif .type _Z8GPuEulerPfffif, @function _Z8GPuEulerPfffif: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z31__device_stub__Z8GPuEulerPfffifPfffif addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z8GPuEulerPfffif, .-_Z8GPuEulerPfffif .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "seccion 1.b\n" .LC8: .string "%f\n" .text .globl main .type main, @function main: .LFB2057: .cfi_startproc endbr64 pushq %r13 .cfi_def_cfa_offset 16 .cfi_offset 13, -16 pushq %r12 .cfi_def_cfa_offset 24 .cfi_offset 12, -24 pushq %rbp .cfi_def_cfa_offset 32 .cfi_offset 6, -32 pushq %rbx .cfi_def_cfa_offset 40 .cfi_offset 3, -40 subq $88, %rsp .cfi_def_cfa_offset 128 movq %fs:40, %rax movq %rax, 72(%rsp) xorl %eax, %eax leaq .LC0(%rip), %rsi movl $2, %edi call __printf_chk@PLT movl $-1, %r12d jmp .L14 .L12: leaq 32(%rsp), %rdi call cudaEventCreate@PLT leaq 40(%rsp), %rdi call cudaEventCreate@PLT movl $0, %esi movq 32(%rsp), %rdi call cudaEventRecord@PLT leaq 24(%rsp), %rdi movq %rbx, %rsi call cudaMalloc@PLT movl $256, 60(%rsp) movl $1, 64(%rsp) cvttss2sil 8(%rsp), %eax movl %eax, 48(%rsp) movl $1, 52(%rsp) movl $0, %r9d movl $0, %r8d movq 60(%rsp), %rdx movl $1, %ecx movq 48(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L18 .L13: movl $0, %esi movq 40(%rsp), %rdi call cudaEventRecord@PLT movq 40(%rsp), %rdi call cudaEventSynchronize@PLT leaq 20(%rsp), %rdi movq 40(%rsp), %rdx movq 32(%rsp), %rsi call cudaEventElapsedTime@PLT movl $2, %ecx movq %rbx, %rdx movq 24(%rsp), %rsi movq %rbp, %rdi call cudaMemcpy@PLT movq 24(%rsp), %rdi call cudaFree@PLT movq %rbp, %rdi call free@PLT movq 32(%rsp), %rdi call cudaEventDestroy@PLT movq 40(%rsp), %rdi call cudaEventDestroy@PLT pxor %xmm0, %xmm0 cvtss2sd 20(%rsp), %xmm0 leaq .LC8(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT subl $1, %r12d cmpl $-5, %r12d je .L19 .L14: pxor %xmm1, %xmm1 cvtsi2ssl %r12d, %xmm1 movss .LC1(%rip), %xmm0 call powf@PLT movss %xmm0, 12(%rsp) movss .LC1(%rip), %xmm4 divss %xmm0, %xmm4 movaps %xmm4, %xmm0 addss .LC2(%rip), %xmm0 cvttss2sil %xmm0, %r13d movslq %r13d, %rbx salq $2, %rbx movq %rbx, %rdi call malloc@PLT movq %rax, %rbp pxor %xmm0, %xmm0 cvtsi2ssl %r13d, %xmm0 mulss .LC3(%rip), %xmm0 movss %xmm0, 8(%rsp) movss .LC9(%rip), %xmm1 andps %xmm0, %xmm1 movss .LC4(%rip), %xmm2 ucomiss %xmm1, %xmm2 jbe .L12 cvttss2sil %xmm0, %eax pxor %xmm1, %xmm1 cvtsi2ssl %eax, %xmm1 movaps %xmm0, %xmm2 cmpnless %xmm1, %xmm2 movss .LC2(%rip), %xmm3 andps %xmm3, %xmm2 addss %xmm2, %xmm1 movss .LC9(%rip), %xmm2 andnps %xmm0, %xmm2 orps %xmm2, %xmm1 movss %xmm1, 8(%rsp) jmp .L12 .L18: movss 12(%rsp), %xmm2 movl %r13d, %esi movss .LC6(%rip), %xmm1 pxor %xmm0, %xmm0 movq 24(%rsp), %rdi call _Z31__device_stub__Z8GPuEulerPfffifPfffif jmp .L13 .L19: movq 72(%rsp), %rax subq %fs:40, %rax jne .L20 movl $0, %eax addq $88, %rsp .cfi_remember_state .cfi_def_cfa_offset 40 popq %rbx .cfi_def_cfa_offset 32 popq %rbp .cfi_def_cfa_offset 24 popq %r12 .cfi_def_cfa_offset 16 popq %r13 .cfi_def_cfa_offset 8 ret .L20: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1 .LC10: .string "_Z8GPuEulerPfffif" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2085: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC10(%rip), %rdx movq %rdx, %rcx leaq _Z8GPuEulerPfffif(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2085: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC1: .long 1092616192 .align 4 .LC2: .long 1065353216 .align 4 .LC3: .long 998244352 .align 4 .LC4: .long 1258291200 .align 4 .LC6: .long 1082130432 .align 4 .LC9: .long 2147483647 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "pregunta1b.hip" .globl _Z23__device_stub__GPuEulerPfffif # -- Begin function _Z23__device_stub__GPuEulerPfffif .p2align 4, 0x90 .type _Z23__device_stub__GPuEulerPfffif,@function _Z23__device_stub__GPuEulerPfffif: # @_Z23__device_stub__GPuEulerPfffif .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movss %xmm0, 20(%rsp) movss %xmm1, 16(%rsp) movl %esi, 12(%rsp) movss %xmm2, 8(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 20(%rsp), %rax movq %rax, 88(%rsp) leaq 16(%rsp), %rax movq %rax, 96(%rsp) leaq 12(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z8GPuEulerPfffif, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $136, %rsp .cfi_adjust_cfa_offset -136 retq .Lfunc_end0: .size _Z23__device_stub__GPuEulerPfffif, .Lfunc_end0-_Z23__device_stub__GPuEulerPfffif .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function main .LCPI1_0: .long 0x41200000 # float 10 .LCPI1_1: .long 0x3f800000 # float 1 .LCPI1_2: .long 0x3b800000 # float 0.00390625 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $184, %rsp .cfi_def_cfa_offset 240 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movabsq $4294967296, %rbx # imm = 0x100000000 movl $.Lstr, %edi callq puts@PLT movl $-1, %r15d leaq 256(%rbx), %r12 jmp .LBB1_1 .p2align 4, 0x90 .LBB1_3: # in Loop: Header=BB1_1 Depth=1 movq 8(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq 8(%rsp), %rdi callq hipEventSynchronize movq 16(%rsp), %rsi movq 8(%rsp), %rdx leaq 36(%rsp), %rdi callq hipEventElapsedTime movq 24(%rsp), %rsi movq %r13, %rdi movq %rbx, %rdx movl $2, %ecx callq hipMemcpy movq 24(%rsp), %rdi callq hipFree movq %r13, %rdi callq free movq 16(%rsp), %rdi callq hipEventDestroy movq 8(%rsp), %rdi callq hipEventDestroy movss 36(%rsp), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str.1, %edi movb $1, %al callq printf decl %r15d cmpl $-5, %r15d je .LBB1_4 .LBB1_1: # =>This Inner Loop Header: Depth=1 cvtsi2ss %r15d, %xmm1 movss .LCPI1_0(%rip), %xmm0 # xmm0 = mem[0],zero,zero,zero callq powf movss .LCPI1_0(%rip), %xmm1 # xmm1 = mem[0],zero,zero,zero movss %xmm0, 32(%rsp) # 4-byte Spill divss %xmm0, %xmm1 addss .LCPI1_1(%rip), %xmm1 movaps %xmm1, 160(%rsp) # 16-byte Spill cvttss2si %xmm1, %ebp movslq %ebp, %rbx shlq $2, %rbx movq %rbx, %rdi callq malloc movq %rax, %r13 cvttps2dq 160(%rsp), %xmm0 # 16-byte Folded Reload cvtdq2ps %xmm0, %xmm0 mulss .LCPI1_2(%rip), %xmm0 callq ceilf@PLT cvttss2si %xmm0, %r14d leaq 16(%rsp), %rdi callq hipEventCreate leaq 8(%rsp), %rdi callq hipEventCreate movq 16(%rsp), %rdi xorl %esi, %esi callq hipEventRecord leaq 24(%rsp), %rdi movq %rbx, %rsi callq hipMalloc movabsq $4294967296, %rax # imm = 0x100000000 orq %rax, %r14 movq %r14, %rdi movl $1, %esi movq %r12, %rdx movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_3 # %bb.2: # in Loop: Header=BB1_1 Depth=1 movq 24(%rsp), %rax movq %rax, 104(%rsp) movl $0, 52(%rsp) movl $1082130432, 48(%rsp) # imm = 0x40800000 movl %ebp, 44(%rsp) movss 32(%rsp), %xmm0 # 4-byte Reload # xmm0 = mem[0],zero,zero,zero movss %xmm0, 40(%rsp) leaq 104(%rsp), %rax movq %rax, 112(%rsp) leaq 52(%rsp), %rax movq %rax, 120(%rsp) leaq 48(%rsp), %rax movq %rax, 128(%rsp) leaq 44(%rsp), %rax movq %rax, 136(%rsp) leaq 40(%rsp), %rax movq %rax, 144(%rsp) leaq 88(%rsp), %rdi leaq 72(%rsp), %rsi leaq 64(%rsp), %rdx leaq 56(%rsp), %rcx callq __hipPopCallConfiguration movq 88(%rsp), %rsi movl 96(%rsp), %edx movq 72(%rsp), %rcx movl 80(%rsp), %r8d movl $_Z8GPuEulerPfffif, %edi leaq 112(%rsp), %r9 pushq 56(%rsp) .cfi_adjust_cfa_offset 8 pushq 72(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 jmp .LBB1_3 .LBB1_4: xorl %eax, %eax addq $184, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z8GPuEulerPfffif, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z8GPuEulerPfffif,@object # @_Z8GPuEulerPfffif .section .rodata,"a",@progbits .globl _Z8GPuEulerPfffif .p2align 3, 0x0 _Z8GPuEulerPfffif: .quad _Z23__device_stub__GPuEulerPfffif .size _Z8GPuEulerPfffif, 8 .type .L.str.1,@object # @.str.1 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.1: .asciz "%f\n" .size .L.str.1, 4 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z8GPuEulerPfffif" .size .L__unnamed_1, 18 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "seccion 1.b" .size .Lstr, 12 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z23__device_stub__GPuEulerPfffif .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z8GPuEulerPfffif .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
#include <stdio.h> // Cuda supports printf in kernels for // hardware with compute compatibility >= 2.0 __global__ void helloworld() { // CUDA runtime uses device overloading or printf in kernels printf("Hello world!\n"); } int main(void) { helloworld<<<1,1>>>(); return 0; }
code for sm_80 Function : _Z10helloworldv .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ IMAD.MOV.U32 R1, RZ, RZ, c[0x0][0x28] ; /* 0x00000a00ff017624 */ /* 0x000fe400078e00ff */ /*0010*/ MOV R0, 0x0 ; /* 0x0000000000007802 */ /* 0x000fe20000000f00 */ /*0020*/ IMAD.MOV.U32 R4, RZ, RZ, c[0x4][0x8] ; /* 0x01000200ff047624 */ /* 0x000fe200078e00ff */ /*0030*/ CS2R R6, SRZ ; /* 0x0000000000067805 */ /* 0x000fe2000001ff00 */ /*0040*/ IMAD.MOV.U32 R5, RZ, RZ, c[0x4][0xc] ; /* 0x01000300ff057624 */ /* 0x000fe200078e00ff */ /*0050*/ LDC.64 R2, c[0x4][R0] ; /* 0x0100000000027b82 */ /* 0x00006c0000000a00 */ /*0060*/ LEPC R8 ; /* 0x000000000008734e */ /* 0x000fe40000000000 */ /*0070*/ MOV R11, 0xe0 ; /* 0x000000e0000b7802 */ /* 0x000fe40000000f00 */ /*0080*/ MOV R20, 0x60 ; /* 0x0000006000147802 */ /* 0x000fe40000000f00 */ /*0090*/ MOV R21, 0x0 ; /* 0x0000000000157802 */ /* 0x000fe40000000f00 */ /*00a0*/ MOV R0, 0x0 ; /* 0x0000000000007802 */ /* 0x001fc40000000f00 */ /*00b0*/ IADD3 R20, P0, P1, -R20, R11, R8 ; /* 0x0000000b14147210 */ /* 0x000fc8000791e108 */ /*00c0*/ IADD3.X R21, ~R0, R21, R9, P0, P1 ; /* 0x0000001500157210 */ /* 0x000fc800007e2509 */ /*00d0*/ CALL.ABS.NOINC R2 ; /* 0x0000000002007343 */ /* 0x002fea0003c00000 */ /*00e0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00f0*/ BRA 0xf0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0100*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0110*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0120*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0130*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0140*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0150*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0160*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0170*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include <stdio.h> // Cuda supports printf in kernels for // hardware with compute compatibility >= 2.0 __global__ void helloworld() { // CUDA runtime uses device overloading or printf in kernels printf("Hello world!\n"); } int main(void) { helloworld<<<1,1>>>(); return 0; }
.file "tmpxft_0014b04c_00000000-6_helloworld.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2060: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2060: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z29__device_stub__Z10helloworldvv .type _Z29__device_stub__Z10helloworldvv, @function _Z29__device_stub__Z10helloworldvv: .LFB2082: .cfi_startproc endbr64 subq $88, %rsp .cfi_def_cfa_offset 96 movq %fs:40, %rax movq %rax, 72(%rsp) xorl %eax, %eax movl $1, 16(%rsp) movl $1, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) leaq 8(%rsp), %rcx movq %rsp, %rdx leaq 28(%rsp), %rsi leaq 16(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 72(%rsp), %rax subq %fs:40, %rax jne .L8 addq $88, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 8(%rsp) .cfi_def_cfa_offset 104 pushq 8(%rsp) .cfi_def_cfa_offset 112 leaq 80(%rsp), %r9 movq 44(%rsp), %rcx movl 52(%rsp), %r8d movq 32(%rsp), %rsi movl 40(%rsp), %edx leaq _Z10helloworldv(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 96 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2082: .size _Z29__device_stub__Z10helloworldvv, .-_Z29__device_stub__Z10helloworldvv .globl _Z10helloworldv .type _Z10helloworldv, @function _Z10helloworldv: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z10helloworldvv addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z10helloworldv, .-_Z10helloworldv .globl main .type main, @function main: .LFB2057: .cfi_startproc endbr64 subq $40, %rsp .cfi_def_cfa_offset 48 movl $1, 20(%rsp) movl $1, 24(%rsp) movl $1, 8(%rsp) movl $1, 12(%rsp) movl $0, %r9d movl $0, %r8d movq 20(%rsp), %rdx movl $1, %ecx movq 8(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L14 .L12: movl $0, %eax addq $40, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L14: .cfi_restore_state call _Z29__device_stub__Z10helloworldvv jmp .L12 .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z10helloworldv" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2085: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z10helloworldv(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2085: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include <stdio.h> // Cuda supports printf in kernels for // hardware with compute compatibility >= 2.0 __global__ void helloworld() { // CUDA runtime uses device overloading or printf in kernels printf("Hello world!\n"); } int main(void) { helloworld<<<1,1>>>(); return 0; }
#include <hip/hip_runtime.h> #include <stdio.h> // Cuda supports printf in kernels for // hardware with compute compatibility >= 2.0 __global__ void helloworld() { // CUDA runtime uses device overloading or printf in kernels printf("Hello world!\n"); } int main(void) { helloworld<<<1,1>>>(); return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> #include <stdio.h> // Cuda supports printf in kernels for // hardware with compute compatibility >= 2.0 __global__ void helloworld() { // CUDA runtime uses device overloading or printf in kernels printf("Hello world!\n"); } int main(void) { helloworld<<<1,1>>>(); return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z10helloworldv .globl _Z10helloworldv .p2align 8 .type _Z10helloworldv,@function _Z10helloworldv: s_load_b64 s[2:3], s[0:1], 0x50 v_mbcnt_lo_u32_b32 v20, -1, 0 v_mov_b32_e32 v6, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_dual_mov_b32 v7, 0 :: v_dual_mov_b32 v4, v20 v_readfirstlane_b32 s0, v4 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_eq_u32_e64 s0, s0, v4 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_6 v_mov_b32_e32 v0, 0 s_mov_b32 s4, exec_lo s_waitcnt lgkmcnt(0) global_load_b64 v[8:9], v0, s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv s_clause 0x1 global_load_b64 v[1:2], v0, s[2:3] offset:40 global_load_b64 v[5:6], v0, s[2:3] s_waitcnt vmcnt(1) v_and_b32_e32 v1, v1, v8 v_and_b32_e32 v2, v2, v9 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_hi_u32 v3, v1, 24 v_mul_lo_u32 v2, v2, 24 v_mul_lo_u32 v1, v1, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v2, v3, v2 s_waitcnt vmcnt(0) v_add_co_u32 v1, vcc_lo, v5, v1 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v2, vcc_lo, v6, v2, vcc_lo global_load_b64 v[6:7], v[1:2], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[6:7], v0, v[6:9], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmpx_ne_u64_e64 v[6:7], v[8:9] s_cbranch_execz .LBB0_5 s_mov_b32 s5, 0 .p2align 6 .LBB0_3: s_sleep 1 s_clause 0x1 global_load_b64 v[1:2], v0, s[2:3] offset:40 global_load_b64 v[10:11], v0, s[2:3] v_dual_mov_b32 v9, v7 :: v_dual_mov_b32 v8, v6 s_waitcnt vmcnt(1) s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_and_b32_e32 v1, v1, v8 v_and_b32_e32 v7, v2, v9 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u64_u32 v[5:6], null, v1, 24, v[10:11] v_mov_b32_e32 v1, v6 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u64_u32 v[2:3], null, v7, 24, v[1:2] v_mov_b32_e32 v6, v2 global_load_b64 v[6:7], v[5:6], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[6:7], v0, v[6:9], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmp_eq_u64_e32 vcc_lo, v[6:7], v[8:9] s_or_b32 s5, vcc_lo, s5 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s5 s_cbranch_execnz .LBB0_3 s_or_b32 exec_lo, exec_lo, s5 .LBB0_5: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s4 .LBB0_6: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s1 v_mov_b32_e32 v5, 0 v_readfirstlane_b32 s4, v6 v_readfirstlane_b32 s5, v7 s_mov_b32 s8, exec_lo s_waitcnt lgkmcnt(0) s_clause 0x1 global_load_b64 v[8:9], v5, s[2:3] offset:40 global_load_b128 v[0:3], v5, s[2:3] s_waitcnt vmcnt(1) v_readfirstlane_b32 s6, v8 v_readfirstlane_b32 s7, v9 s_delay_alu instid0(VALU_DEP_1) s_and_b64 s[6:7], s[4:5], s[6:7] s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_8 v_dual_mov_b32 v6, s8 :: v_dual_mov_b32 v7, 0 s_mul_i32 s8, s7, 24 s_mul_hi_u32 s9, s6, 24 v_dual_mov_b32 v8, 2 :: v_dual_mov_b32 v9, 1 s_add_i32 s9, s9, s8 s_mul_i32 s8, s6, 24 s_waitcnt vmcnt(0) v_add_co_u32 v10, vcc_lo, v0, s8 v_add_co_ci_u32_e32 v11, vcc_lo, s9, v1, vcc_lo global_store_b128 v[10:11], v[6:9], off offset:8 .LBB0_8: s_or_b32 exec_lo, exec_lo, s1 s_lshl_b64 s[8:9], s[6:7], 12 v_lshlrev_b64 v[4:5], 6, v[4:5] s_waitcnt vmcnt(0) v_add_co_u32 v2, vcc_lo, v2, s8 v_add_co_ci_u32_e32 v7, vcc_lo, s9, v3, vcc_lo v_mov_b32_e32 v3, 0 s_mov_b32 s8, 0 s_delay_alu instid0(VALU_DEP_3) v_add_co_u32 v6, vcc_lo, v2, v4 v_mov_b32_e32 v2, 33 s_mov_b32 s9, s8 s_mov_b32 s10, s8 s_mov_b32 s11, s8 v_add_co_ci_u32_e32 v7, vcc_lo, v7, v5, vcc_lo v_mov_b32_e32 v4, v3 v_dual_mov_b32 v5, v3 :: v_dual_mov_b32 v8, s8 v_dual_mov_b32 v9, s9 :: v_dual_mov_b32 v10, s10 v_mov_b32_e32 v11, s11 s_clause 0x3 global_store_b128 v[6:7], v[2:5], off global_store_b128 v[6:7], v[8:11], off offset:16 global_store_b128 v[6:7], v[8:11], off offset:32 global_store_b128 v[6:7], v[8:11], off offset:48 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_16 v_dual_mov_b32 v10, 0 :: v_dual_mov_b32 v11, s4 v_mov_b32_e32 v12, s5 s_clause 0x1 global_load_b64 v[13:14], v10, s[2:3] offset:32 glc global_load_b64 v[2:3], v10, s[2:3] offset:40 s_waitcnt vmcnt(0) v_readfirstlane_b32 s8, v2 v_readfirstlane_b32 s9, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_b64 s[8:9], s[8:9], s[4:5] s_mul_i32 s9, s9, 24 s_mul_hi_u32 s10, s8, 24 s_mul_i32 s8, s8, 24 s_add_i32 s10, s10, s9 v_add_co_u32 v8, vcc_lo, v0, s8 v_add_co_ci_u32_e32 v9, vcc_lo, s10, v1, vcc_lo s_mov_b32 s8, exec_lo global_store_b64 v[8:9], v[13:14], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[4:5], v10, v[11:14], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmpx_ne_u64_e64 v[4:5], v[13:14] s_cbranch_execz .LBB0_12 s_mov_b32 s9, 0 .LBB0_11: v_dual_mov_b32 v2, s4 :: v_dual_mov_b32 v3, s5 s_sleep 1 global_store_b64 v[8:9], v[4:5], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v10, v[2:5], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[2:3], v[4:5] v_dual_mov_b32 v5, v3 :: v_dual_mov_b32 v4, v2 s_or_b32 s9, vcc_lo, s9 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s9 s_cbranch_execnz .LBB0_11 .LBB0_12: s_or_b32 exec_lo, exec_lo, s8 v_mov_b32_e32 v2, 0 s_mov_b32 s9, exec_lo s_mov_b32 s8, exec_lo v_mbcnt_lo_u32_b32 v4, s9, 0 global_load_b64 v[2:3], v2, s[2:3] offset:16 v_cmpx_eq_u32_e32 0, v4 s_cbranch_execz .LBB0_14 s_bcnt1_i32_b32 s9, s9 s_delay_alu instid0(SALU_CYCLE_1) v_dual_mov_b32 v5, 0 :: v_dual_mov_b32 v4, s9 s_waitcnt vmcnt(0) global_atomic_add_u64 v[2:3], v[4:5], off offset:8 .LBB0_14: s_or_b32 exec_lo, exec_lo, s8 s_waitcnt vmcnt(0) global_load_b64 v[4:5], v[2:3], off offset:16 s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, 0, v[4:5] s_cbranch_vccnz .LBB0_16 global_load_b32 v2, v[2:3], off offset:24 v_mov_b32_e32 v3, 0 s_waitcnt vmcnt(0) v_readfirstlane_b32 s8, v2 s_waitcnt_vscnt null, 0x0 global_store_b64 v[4:5], v[2:3], off s_and_b32 m0, s8, 0xff s_sendmsg sendmsg(MSG_INTERRUPT) .LBB0_16: s_or_b32 exec_lo, exec_lo, s1 s_mul_i32 s1, s7, 24 s_mul_hi_u32 s7, s6, 24 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) s_add_i32 s7, s7, s1 s_mul_i32 s1, s6, 24 v_add_co_u32 v0, vcc_lo, v0, s1 v_add_co_ci_u32_e32 v1, vcc_lo, s7, v1, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v0, 20 v_add_co_ci_u32_e32 v1, vcc_lo, 0, v1, vcc_lo s_branch .LBB0_20 .p2align 6 .LBB0_17: s_or_b32 exec_lo, exec_lo, s1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_readfirstlane_b32 s1, v2 s_cmp_eq_u32 s1, 0 s_cbranch_scc1 .LBB0_19 s_sleep 1 s_cbranch_execnz .LBB0_20 s_branch .LBB0_22 .p2align 6 .LBB0_19: s_branch .LBB0_22 .LBB0_20: v_mov_b32_e32 v2, 1 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_17 global_load_b32 v2, v[0:1], off glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_and_b32_e32 v2, 1, v2 s_branch .LBB0_17 .LBB0_22: global_load_b64 v[22:23], v[6:7], off s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_26 v_mov_b32_e32 v6, 0 s_clause 0x2 global_load_b64 v[2:3], v6, s[2:3] offset:40 global_load_b64 v[7:8], v6, s[2:3] offset:24 glc global_load_b64 v[4:5], v6, s[2:3] s_waitcnt vmcnt(2) v_add_co_u32 v9, vcc_lo, v2, 1 v_add_co_ci_u32_e32 v10, vcc_lo, 0, v3, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v9, s4 v_add_co_ci_u32_e32 v1, vcc_lo, s5, v10, vcc_lo s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_cmp_eq_u64_e32 vcc_lo, 0, v[0:1] v_dual_cndmask_b32 v1, v1, v10 :: v_dual_cndmask_b32 v0, v0, v9 v_and_b32_e32 v3, v1, v3 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_and_b32_e32 v2, v0, v2 v_mul_lo_u32 v3, v3, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_mul_hi_u32 v9, v2, 24 v_mul_lo_u32 v2, v2, 24 v_add_nc_u32_e32 v3, v9, v3 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_co_u32 v4, vcc_lo, v4, v2 v_mov_b32_e32 v2, v7 v_add_co_ci_u32_e32 v5, vcc_lo, v5, v3, vcc_lo v_mov_b32_e32 v3, v8 global_store_b64 v[4:5], v[7:8], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v6, v[0:3], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_ne_u64_e32 vcc_lo, v[2:3], v[7:8] s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_26 s_mov_b32 s0, 0 .LBB0_25: s_sleep 1 global_store_b64 v[4:5], v[2:3], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[7:8], v6, v[0:3], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[7:8], v[2:3] v_dual_mov_b32 v2, v7 :: v_dual_mov_b32 v3, v8 s_or_b32 s0, vcc_lo, s0 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s0 s_cbranch_execnz .LBB0_25 .LBB0_26: s_or_b32 exec_lo, exec_lo, s1 s_getpc_b64 s[4:5] s_add_u32 s4, s4, .str@rel32@lo+4 s_addc_u32 s5, s5, .str@rel32@hi+12 s_mov_b32 s0, -1 s_cmp_lg_u64 s[4:5], 0 s_cbranch_scc0 .LBB0_105 s_waitcnt vmcnt(0) v_dual_mov_b32 v1, v23 :: v_dual_and_b32 v0, -3, v22 v_mov_b32_e32 v25, 0 s_mov_b64 s[6:7], 14 s_branch .LBB0_29 .LBB0_28: s_or_b32 exec_lo, exec_lo, s1 s_sub_u32 s6, s6, s8 s_subb_u32 s7, s7, s9 s_add_u32 s4, s4, s8 s_addc_u32 s5, s5, s9 s_cmp_lg_u64 s[6:7], 0 s_cbranch_scc0 .LBB0_104 .LBB0_29: v_cmp_lt_u64_e64 s0, s[6:7], 56 s_delay_alu instid0(VALU_DEP_1) s_and_b32 s0, s0, exec_lo s_cselect_b32 s8, s6, 56 s_cselect_b32 s9, s7, 0 s_cmp_gt_u32 s8, 7 s_mov_b32 s0, -1 s_cbranch_scc1 .LBB0_34 v_mov_b32_e32 v2, 0 v_mov_b32_e32 v3, 0 s_cmp_eq_u32 s8, 0 s_cbranch_scc1 .LBB0_33 s_lshl_b64 s[0:1], s[8:9], 3 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], s[4:5] .LBB0_32: global_load_u8 v4, v25, s[12:13] s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v4 s_delay_alu instid0(VALU_DEP_1) v_lshlrev_b64 v[4:5], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_add_u32 s12, s12, 1 s_addc_u32 s13, s13, 0 s_cmp_lg_u32 s0, s10 v_or_b32_e32 v2, v4, v2 v_or_b32_e32 v3, v5, v3 s_cbranch_scc1 .LBB0_32 .LBB0_33: s_mov_b32 s0, 0 s_mov_b32 s15, 0 .LBB0_34: s_and_not1_b32 vcc_lo, exec_lo, s0 s_mov_b64 s[0:1], s[4:5] s_cbranch_vccnz .LBB0_36 global_load_b64 v[2:3], v25, s[4:5] s_add_i32 s15, s8, -8 s_add_u32 s0, s4, 8 s_addc_u32 s1, s5, 0 .LBB0_36: s_cmp_gt_u32 s15, 7 s_cbranch_scc1 .LBB0_41 v_mov_b32_e32 v4, 0 v_mov_b32_e32 v5, 0 s_cmp_eq_u32 s15, 0 s_cbranch_scc1 .LBB0_40 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_39: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v6, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v6 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[6:7], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s15, s12 v_or_b32_e32 v4, v6, v4 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v5, v7, v5 s_cbranch_scc1 .LBB0_39 .LBB0_40: s_mov_b32 s14, 0 s_cbranch_execz .LBB0_42 s_branch .LBB0_43 .LBB0_41: .LBB0_42: global_load_b64 v[4:5], v25, s[0:1] s_add_i32 s14, s15, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_43: s_cmp_gt_u32 s14, 7 s_cbranch_scc1 .LBB0_48 v_mov_b32_e32 v6, 0 v_mov_b32_e32 v7, 0 s_cmp_eq_u32 s14, 0 s_cbranch_scc1 .LBB0_47 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_46: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v8, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v8 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[8:9], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s14, s12 v_or_b32_e32 v6, v8, v6 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v7, v9, v7 s_cbranch_scc1 .LBB0_46 .LBB0_47: s_mov_b32 s15, 0 s_cbranch_execz .LBB0_49 s_branch .LBB0_50 .LBB0_48: .LBB0_49: global_load_b64 v[6:7], v25, s[0:1] s_add_i32 s15, s14, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_50: s_cmp_gt_u32 s15, 7 s_cbranch_scc1 .LBB0_55 v_mov_b32_e32 v8, 0 v_mov_b32_e32 v9, 0 s_cmp_eq_u32 s15, 0 s_cbranch_scc1 .LBB0_54 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_53: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v10, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v10 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[10:11], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s15, s12 v_or_b32_e32 v8, v10, v8 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v9, v11, v9 s_cbranch_scc1 .LBB0_53 .LBB0_54: s_mov_b32 s14, 0 s_cbranch_execz .LBB0_56 s_branch .LBB0_57 .LBB0_55: .LBB0_56: global_load_b64 v[8:9], v25, s[0:1] s_add_i32 s14, s15, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_57: s_cmp_gt_u32 s14, 7 s_cbranch_scc1 .LBB0_62 v_mov_b32_e32 v10, 0 v_mov_b32_e32 v11, 0 s_cmp_eq_u32 s14, 0 s_cbranch_scc1 .LBB0_61 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_60: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v12, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v12 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[12:13], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s14, s12 v_or_b32_e32 v10, v12, v10 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v11, v13, v11 s_cbranch_scc1 .LBB0_60 .LBB0_61: s_mov_b32 s15, 0 s_cbranch_execz .LBB0_63 s_branch .LBB0_64 .LBB0_62: .LBB0_63: global_load_b64 v[10:11], v25, s[0:1] s_add_i32 s15, s14, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_64: s_cmp_gt_u32 s15, 7 s_cbranch_scc1 .LBB0_69 v_mov_b32_e32 v12, 0 v_mov_b32_e32 v13, 0 s_cmp_eq_u32 s15, 0 s_cbranch_scc1 .LBB0_68 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_67: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v14, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v14 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[14:15], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s15, s12 v_or_b32_e32 v12, v14, v12 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v13, v15, v13 s_cbranch_scc1 .LBB0_67 .LBB0_68: s_mov_b32 s14, 0 s_cbranch_execz .LBB0_70 s_branch .LBB0_71 .LBB0_69: .LBB0_70: global_load_b64 v[12:13], v25, s[0:1] s_add_i32 s14, s15, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_71: s_cmp_gt_u32 s14, 7 s_cbranch_scc1 .LBB0_76 v_mov_b32_e32 v14, 0 v_mov_b32_e32 v15, 0 s_cmp_eq_u32 s14, 0 s_cbranch_scc1 .LBB0_75 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], s[0:1] .LBB0_74: global_load_u8 v16, v25, s[12:13] s_add_i32 s14, s14, -1 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v16 s_delay_alu instid0(VALU_DEP_1) v_lshlrev_b64 v[16:17], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_add_u32 s12, s12, 1 s_addc_u32 s13, s13, 0 s_cmp_lg_u32 s14, 0 v_or_b32_e32 v14, v16, v14 v_or_b32_e32 v15, v17, v15 s_cbranch_scc1 .LBB0_74 .LBB0_75: s_cbranch_execz .LBB0_77 s_branch .LBB0_78 .LBB0_76: .LBB0_77: global_load_b64 v[14:15], v25, s[0:1] .LBB0_78: v_mov_b32_e32 v24, v20 v_mov_b32_e32 v26, 0 v_mov_b32_e32 v27, 0 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_readfirstlane_b32 s0, v24 v_cmp_eq_u32_e64 s0, s0, v24 s_delay_alu instid0(VALU_DEP_1) s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_84 global_load_b64 v[18:19], v25, s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv s_clause 0x1 global_load_b64 v[16:17], v25, s[2:3] offset:40 global_load_b64 v[26:27], v25, s[2:3] s_mov_b32 s10, exec_lo s_waitcnt vmcnt(1) v_and_b32_e32 v17, v17, v19 v_and_b32_e32 v16, v16, v18 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_lo_u32 v17, v17, 24 v_mul_hi_u32 v21, v16, 24 v_mul_lo_u32 v16, v16, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v17, v21, v17 s_waitcnt vmcnt(0) v_add_co_u32 v16, vcc_lo, v26, v16 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v17, vcc_lo, v27, v17, vcc_lo global_load_b64 v[16:17], v[16:17], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[26:27], v25, v[16:19], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmpx_ne_u64_e64 v[26:27], v[18:19] s_cbranch_execz .LBB0_83 s_mov_b32 s11, 0 .p2align 6 .LBB0_81: s_sleep 1 s_clause 0x1 global_load_b64 v[16:17], v25, s[2:3] offset:40 global_load_b64 v[28:29], v25, s[2:3] v_dual_mov_b32 v18, v26 :: v_dual_mov_b32 v19, v27 s_waitcnt vmcnt(1) s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_and_b32_e32 v16, v16, v18 s_waitcnt vmcnt(0) v_mad_u64_u32 v[26:27], null, v16, 24, v[28:29] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_dual_mov_b32 v16, v27 :: v_dual_and_b32 v17, v17, v19 v_mad_u64_u32 v[27:28], null, v17, 24, v[16:17] global_load_b64 v[16:17], v[26:27], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[26:27], v25, v[16:19], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmp_eq_u64_e32 vcc_lo, v[26:27], v[18:19] s_or_b32 s11, vcc_lo, s11 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s11 s_cbranch_execnz .LBB0_81 s_or_b32 exec_lo, exec_lo, s11 .LBB0_83: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s10 .LBB0_84: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s1 s_clause 0x1 global_load_b64 v[28:29], v25, s[2:3] offset:40 global_load_b128 v[16:19], v25, s[2:3] v_readfirstlane_b32 s10, v26 v_readfirstlane_b32 s11, v27 s_mov_b32 s14, exec_lo s_waitcnt vmcnt(1) v_readfirstlane_b32 s12, v28 v_readfirstlane_b32 s13, v29 s_delay_alu instid0(VALU_DEP_1) s_and_b64 s[12:13], s[10:11], s[12:13] s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_86 v_dual_mov_b32 v26, s14 :: v_dual_mov_b32 v27, 0 s_mul_i32 s14, s13, 24 s_mul_hi_u32 s15, s12, 24 v_dual_mov_b32 v28, 2 :: v_dual_mov_b32 v29, 1 s_add_i32 s15, s15, s14 s_mul_i32 s14, s12, 24 s_waitcnt vmcnt(0) v_add_co_u32 v30, vcc_lo, v16, s14 v_add_co_ci_u32_e32 v31, vcc_lo, s15, v17, vcc_lo global_store_b128 v[30:31], v[26:29], off offset:8 .LBB0_86: s_or_b32 exec_lo, exec_lo, s1 v_cmp_gt_u64_e64 vcc_lo, s[6:7], 56 v_or_b32_e32 v21, 2, v0 s_lshl_b64 s[14:15], s[12:13], 12 v_lshlrev_b64 v[26:27], 6, v[24:25] s_lshl_b32 s1, s8, 2 s_delay_alu instid0(SALU_CYCLE_1) s_add_i32 s1, s1, 28 v_cndmask_b32_e32 v0, v21, v0, vcc_lo s_waitcnt vmcnt(0) v_add_co_u32 v18, vcc_lo, v18, s14 v_add_co_ci_u32_e32 v19, vcc_lo, s15, v19, vcc_lo s_and_b32 s1, s1, 0x1e0 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_co_u32 v18, vcc_lo, v18, v26 v_and_or_b32 v0, v0, 0xffffff1f, s1 v_add_co_ci_u32_e32 v19, vcc_lo, v19, v27, vcc_lo s_clause 0x3 global_store_b128 v[18:19], v[0:3], off global_store_b128 v[18:19], v[4:7], off offset:16 global_store_b128 v[18:19], v[8:11], off offset:32 global_store_b128 v[18:19], v[12:15], off offset:48 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_94 s_clause 0x1 global_load_b64 v[8:9], v25, s[2:3] offset:32 glc global_load_b64 v[0:1], v25, s[2:3] offset:40 v_dual_mov_b32 v6, s10 :: v_dual_mov_b32 v7, s11 s_waitcnt vmcnt(0) v_readfirstlane_b32 s14, v0 v_readfirstlane_b32 s15, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_b64 s[14:15], s[14:15], s[10:11] s_mul_i32 s15, s15, 24 s_mul_hi_u32 s16, s14, 24 s_mul_i32 s14, s14, 24 s_add_i32 s16, s16, s15 v_add_co_u32 v4, vcc_lo, v16, s14 v_add_co_ci_u32_e32 v5, vcc_lo, s16, v17, vcc_lo s_mov_b32 s14, exec_lo global_store_b64 v[4:5], v[8:9], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v25, v[6:9], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmpx_ne_u64_e64 v[2:3], v[8:9] s_cbranch_execz .LBB0_90 s_mov_b32 s15, 0 .LBB0_89: v_dual_mov_b32 v0, s10 :: v_dual_mov_b32 v1, s11 s_sleep 1 global_store_b64 v[4:5], v[2:3], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[0:1], v25, v[0:3], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[2:3] v_dual_mov_b32 v3, v1 :: v_dual_mov_b32 v2, v0 s_or_b32 s15, vcc_lo, s15 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s15 s_cbranch_execnz .LBB0_89 .LBB0_90: s_or_b32 exec_lo, exec_lo, s14 global_load_b64 v[0:1], v25, s[2:3] offset:16 s_mov_b32 s15, exec_lo s_mov_b32 s14, exec_lo v_mbcnt_lo_u32_b32 v2, s15, 0 s_delay_alu instid0(VALU_DEP_1) v_cmpx_eq_u32_e32 0, v2 s_cbranch_execz .LBB0_92 s_bcnt1_i32_b32 s15, s15 s_delay_alu instid0(SALU_CYCLE_1) v_dual_mov_b32 v3, 0 :: v_dual_mov_b32 v2, s15 s_waitcnt vmcnt(0) global_atomic_add_u64 v[0:1], v[2:3], off offset:8 .LBB0_92: s_or_b32 exec_lo, exec_lo, s14 s_waitcnt vmcnt(0) global_load_b64 v[2:3], v[0:1], off offset:16 s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, 0, v[2:3] s_cbranch_vccnz .LBB0_94 global_load_b32 v24, v[0:1], off offset:24 s_waitcnt vmcnt(0) v_readfirstlane_b32 s14, v24 s_waitcnt_vscnt null, 0x0 global_store_b64 v[2:3], v[24:25], off s_and_b32 m0, s14, 0xff s_sendmsg sendmsg(MSG_INTERRUPT) .LBB0_94: s_or_b32 exec_lo, exec_lo, s1 s_mul_i32 s1, s13, 24 s_mul_hi_u32 s13, s12, 24 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) s_add_i32 s13, s13, s1 s_mul_i32 s1, s12, 24 v_add_co_u32 v0, vcc_lo, v16, s1 v_add_co_ci_u32_e32 v1, vcc_lo, s13, v17, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v0, 20 v_add_co_ci_u32_e32 v1, vcc_lo, 0, v1, vcc_lo s_branch .LBB0_98 .p2align 6 .LBB0_95: s_or_b32 exec_lo, exec_lo, s1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_readfirstlane_b32 s1, v2 s_cmp_eq_u32 s1, 0 s_cbranch_scc1 .LBB0_97 s_sleep 1 s_cbranch_execnz .LBB0_98 s_branch .LBB0_100 .p2align 6 .LBB0_97: s_branch .LBB0_100 .LBB0_98: v_mov_b32_e32 v2, 1 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_95 global_load_b32 v2, v[0:1], off glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_and_b32_e32 v2, 1, v2 s_branch .LBB0_95 .LBB0_100: global_load_b64 v[0:1], v[18:19], off s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_28 s_clause 0x2 global_load_b64 v[4:5], v25, s[2:3] offset:40 global_load_b64 v[8:9], v25, s[2:3] offset:24 glc global_load_b64 v[6:7], v25, s[2:3] s_waitcnt vmcnt(2) v_add_co_u32 v10, vcc_lo, v4, 1 v_add_co_ci_u32_e32 v11, vcc_lo, 0, v5, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v2, vcc_lo, v10, s10 v_add_co_ci_u32_e32 v3, vcc_lo, s11, v11, vcc_lo s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_cmp_eq_u64_e32 vcc_lo, 0, v[2:3] v_dual_cndmask_b32 v3, v3, v11 :: v_dual_cndmask_b32 v2, v2, v10 v_and_b32_e32 v5, v3, v5 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_and_b32_e32 v4, v2, v4 v_mul_hi_u32 v10, v4, 24 v_mul_lo_u32 v4, v4, 24 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_add_co_u32 v6, vcc_lo, v6, v4 v_mov_b32_e32 v4, v8 v_mul_lo_u32 v5, v5, 24 v_add_nc_u32_e32 v5, v10, v5 s_delay_alu instid0(VALU_DEP_1) v_add_co_ci_u32_e32 v7, vcc_lo, v7, v5, vcc_lo v_mov_b32_e32 v5, v9 global_store_b64 v[6:7], v[8:9], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[4:5], v25, v[2:5], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_ne_u64_e32 vcc_lo, v[4:5], v[8:9] s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_28 s_mov_b32 s0, 0 .LBB0_103: s_sleep 1 global_store_b64 v[6:7], v[4:5], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[8:9], v25, v[2:5], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[8:9], v[4:5] v_dual_mov_b32 v4, v8 :: v_dual_mov_b32 v5, v9 s_or_b32 s0, vcc_lo, s0 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s0 s_cbranch_execnz .LBB0_103 s_branch .LBB0_28 .LBB0_104: s_mov_b32 s0, 0 .LBB0_105: s_delay_alu instid0(SALU_CYCLE_1) s_and_b32 vcc_lo, exec_lo, s0 s_cbranch_vccz .LBB0_132 v_readfirstlane_b32 s0, v20 v_mov_b32_e32 v4, 0 v_mov_b32_e32 v5, 0 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_eq_u32_e64 s0, s0, v20 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_112 s_waitcnt vmcnt(0) v_mov_b32_e32 v0, 0 s_mov_b32 s4, exec_lo global_load_b64 v[6:7], v0, s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv s_clause 0x1 global_load_b64 v[1:2], v0, s[2:3] offset:40 global_load_b64 v[3:4], v0, s[2:3] s_waitcnt vmcnt(1) v_and_b32_e32 v1, v1, v6 v_and_b32_e32 v2, v2, v7 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_hi_u32 v5, v1, 24 v_mul_lo_u32 v2, v2, 24 v_mul_lo_u32 v1, v1, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v2, v5, v2 s_waitcnt vmcnt(0) v_add_co_u32 v1, vcc_lo, v3, v1 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v2, vcc_lo, v4, v2, vcc_lo global_load_b64 v[4:5], v[1:2], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[4:5], v0, v[4:7], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmpx_ne_u64_e64 v[4:5], v[6:7] s_cbranch_execz .LBB0_111 s_mov_b32 s5, 0 .p2align 6 .LBB0_109: s_sleep 1 s_clause 0x1 global_load_b64 v[1:2], v0, s[2:3] offset:40 global_load_b64 v[8:9], v0, s[2:3] v_dual_mov_b32 v7, v5 :: v_dual_mov_b32 v6, v4 s_waitcnt vmcnt(1) s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_and_b32_e32 v1, v1, v6 s_waitcnt vmcnt(0) v_mad_u64_u32 v[3:4], null, v1, 24, v[8:9] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_dual_mov_b32 v1, v4 :: v_dual_and_b32 v2, v2, v7 v_mad_u64_u32 v[4:5], null, v2, 24, v[1:2] global_load_b64 v[4:5], v[3:4], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[4:5], v0, v[4:7], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmp_eq_u64_e32 vcc_lo, v[4:5], v[6:7] s_or_b32 s5, vcc_lo, s5 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s5 s_cbranch_execnz .LBB0_109 s_or_b32 exec_lo, exec_lo, s5 .LBB0_111: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s4 .LBB0_112: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s1 v_mov_b32_e32 v21, 0 v_readfirstlane_b32 s4, v4 v_readfirstlane_b32 s5, v5 s_mov_b32 s8, exec_lo s_clause 0x1 global_load_b64 v[6:7], v21, s[2:3] offset:40 global_load_b128 v[0:3], v21, s[2:3] s_waitcnt vmcnt(1) v_readfirstlane_b32 s6, v6 v_readfirstlane_b32 s7, v7 s_delay_alu instid0(VALU_DEP_1) s_and_b64 s[6:7], s[4:5], s[6:7] s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_114 v_dual_mov_b32 v4, s8 :: v_dual_mov_b32 v5, 0 s_mul_i32 s8, s7, 24 s_mul_hi_u32 s9, s6, 24 v_dual_mov_b32 v6, 2 :: v_dual_mov_b32 v7, 1 s_add_i32 s9, s9, s8 s_mul_i32 s8, s6, 24 s_waitcnt vmcnt(0) v_add_co_u32 v8, vcc_lo, v0, s8 v_add_co_ci_u32_e32 v9, vcc_lo, s9, v1, vcc_lo global_store_b128 v[8:9], v[4:7], off offset:8 .LBB0_114: s_or_b32 exec_lo, exec_lo, s1 s_lshl_b64 s[8:9], s[6:7], 12 v_and_or_b32 v22, v22, 0xffffff1d, 34 s_waitcnt vmcnt(0) v_add_co_u32 v4, vcc_lo, v2, s8 v_add_co_ci_u32_e32 v5, vcc_lo, s9, v3, vcc_lo v_lshlrev_b64 v[2:3], 6, v[20:21] s_mov_b32 s8, 0 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) s_mov_b32 s9, s8 s_mov_b32 s10, s8 s_mov_b32 s11, s8 v_add_co_u32 v8, vcc_lo, v4, v2 v_mov_b32_e32 v6, 0 v_add_co_ci_u32_e32 v9, vcc_lo, v5, v3, vcc_lo v_dual_mov_b32 v2, s8 :: v_dual_mov_b32 v5, s11 v_dual_mov_b32 v3, s9 :: v_dual_mov_b32 v4, s10 s_delay_alu instid0(VALU_DEP_4) v_mov_b32_e32 v7, v6 s_clause 0x4 global_store_b64 v[8:9], v[22:23], off global_store_b128 v[8:9], v[2:5], off offset:8 global_store_b128 v[8:9], v[2:5], off offset:24 global_store_b128 v[8:9], v[2:5], off offset:40 global_store_b64 v[8:9], v[6:7], off offset:56 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_122 v_dual_mov_b32 v8, 0 :: v_dual_mov_b32 v9, s4 v_mov_b32_e32 v10, s5 s_clause 0x1 global_load_b64 v[11:12], v8, s[2:3] offset:32 glc global_load_b64 v[2:3], v8, s[2:3] offset:40 s_waitcnt vmcnt(0) v_readfirstlane_b32 s8, v2 v_readfirstlane_b32 s9, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_b64 s[8:9], s[8:9], s[4:5] s_mul_i32 s9, s9, 24 s_mul_hi_u32 s10, s8, 24 s_mul_i32 s8, s8, 24 s_add_i32 s10, s10, s9 v_add_co_u32 v6, vcc_lo, v0, s8 v_add_co_ci_u32_e32 v7, vcc_lo, s10, v1, vcc_lo s_mov_b32 s8, exec_lo global_store_b64 v[6:7], v[11:12], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[4:5], v8, v[9:12], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmpx_ne_u64_e64 v[4:5], v[11:12] s_cbranch_execz .LBB0_118 s_mov_b32 s9, 0 .LBB0_117: v_dual_mov_b32 v2, s4 :: v_dual_mov_b32 v3, s5 s_sleep 1 global_store_b64 v[6:7], v[4:5], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v8, v[2:5], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[2:3], v[4:5] v_dual_mov_b32 v5, v3 :: v_dual_mov_b32 v4, v2 s_or_b32 s9, vcc_lo, s9 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s9 s_cbranch_execnz .LBB0_117 .LBB0_118: s_or_b32 exec_lo, exec_lo, s8 v_mov_b32_e32 v2, 0 s_mov_b32 s9, exec_lo s_mov_b32 s8, exec_lo v_mbcnt_lo_u32_b32 v4, s9, 0 global_load_b64 v[2:3], v2, s[2:3] offset:16 v_cmpx_eq_u32_e32 0, v4 s_cbranch_execz .LBB0_120 s_bcnt1_i32_b32 s9, s9 s_delay_alu instid0(SALU_CYCLE_1) v_dual_mov_b32 v5, 0 :: v_dual_mov_b32 v4, s9 s_waitcnt vmcnt(0) global_atomic_add_u64 v[2:3], v[4:5], off offset:8 .LBB0_120: s_or_b32 exec_lo, exec_lo, s8 s_waitcnt vmcnt(0) global_load_b64 v[4:5], v[2:3], off offset:16 s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, 0, v[4:5] s_cbranch_vccnz .LBB0_122 global_load_b32 v2, v[2:3], off offset:24 v_mov_b32_e32 v3, 0 s_waitcnt vmcnt(0) v_readfirstlane_b32 s8, v2 s_waitcnt_vscnt null, 0x0 global_store_b64 v[4:5], v[2:3], off s_and_b32 m0, s8, 0xff s_sendmsg sendmsg(MSG_INTERRUPT) .LBB0_122: s_or_b32 exec_lo, exec_lo, s1 s_mul_i32 s1, s7, 24 s_mul_hi_u32 s7, s6, 24 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) s_add_i32 s7, s7, s1 s_mul_i32 s1, s6, 24 v_add_co_u32 v0, vcc_lo, v0, s1 v_add_co_ci_u32_e32 v1, vcc_lo, s7, v1, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v0, 20 v_add_co_ci_u32_e32 v1, vcc_lo, 0, v1, vcc_lo s_branch .LBB0_126 .p2align 6 .LBB0_123: s_or_b32 exec_lo, exec_lo, s1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_readfirstlane_b32 s1, v2 s_cmp_eq_u32 s1, 0 s_cbranch_scc1 .LBB0_125 s_sleep 1 s_cbranch_execnz .LBB0_126 s_branch .LBB0_128 .p2align 6 .LBB0_125: s_branch .LBB0_128 .LBB0_126: v_mov_b32_e32 v2, 1 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_123 global_load_b32 v2, v[0:1], off glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_and_b32_e32 v2, 1, v2 s_branch .LBB0_123 .LBB0_128: s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_132 v_mov_b32_e32 v6, 0 s_clause 0x2 global_load_b64 v[2:3], v6, s[2:3] offset:40 global_load_b64 v[7:8], v6, s[2:3] offset:24 glc global_load_b64 v[4:5], v6, s[2:3] s_waitcnt vmcnt(2) v_add_co_u32 v9, vcc_lo, v2, 1 v_add_co_ci_u32_e32 v10, vcc_lo, 0, v3, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v9, s4 v_add_co_ci_u32_e32 v1, vcc_lo, s5, v10, vcc_lo s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_cmp_eq_u64_e32 vcc_lo, 0, v[0:1] v_dual_cndmask_b32 v1, v1, v10 :: v_dual_cndmask_b32 v0, v0, v9 v_and_b32_e32 v3, v1, v3 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_and_b32_e32 v2, v0, v2 v_mul_lo_u32 v3, v3, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_mul_hi_u32 v9, v2, 24 v_mul_lo_u32 v2, v2, 24 v_add_nc_u32_e32 v3, v9, v3 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_co_u32 v4, vcc_lo, v4, v2 v_mov_b32_e32 v2, v7 v_add_co_ci_u32_e32 v5, vcc_lo, v5, v3, vcc_lo v_mov_b32_e32 v3, v8 global_store_b64 v[4:5], v[7:8], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v6, v[0:3], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_ne_u64_e32 vcc_lo, v[2:3], v[7:8] s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_132 s_mov_b32 s0, 0 .LBB0_131: s_sleep 1 global_store_b64 v[4:5], v[2:3], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[7:8], v6, v[0:3], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[7:8], v[2:3] v_dual_mov_b32 v2, v7 :: v_dual_mov_b32 v3, v8 s_or_b32 s0, vcc_lo, s0 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s0 s_cbranch_execnz .LBB0_131 .LBB0_132: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z10helloworldv .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 256 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 32 .amdhsa_next_free_sgpr 18 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z10helloworldv, .Lfunc_end0-_Z10helloworldv .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type .str,@object .section .rodata.str1.1,"aMS",@progbits,1 .str: .asciz "Hello world!\n" .size .str, 14 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .offset: 0 .size: 4 .value_kind: hidden_block_count_x - .offset: 4 .size: 4 .value_kind: hidden_block_count_y - .offset: 8 .size: 4 .value_kind: hidden_block_count_z - .offset: 12 .size: 2 .value_kind: hidden_group_size_x - .offset: 14 .size: 2 .value_kind: hidden_group_size_y - .offset: 16 .size: 2 .value_kind: hidden_group_size_z - .offset: 18 .size: 2 .value_kind: hidden_remainder_x - .offset: 20 .size: 2 .value_kind: hidden_remainder_y - .offset: 22 .size: 2 .value_kind: hidden_remainder_z - .offset: 40 .size: 8 .value_kind: hidden_global_offset_x - .offset: 48 .size: 8 .value_kind: hidden_global_offset_y - .offset: 56 .size: 8 .value_kind: hidden_global_offset_z - .offset: 64 .size: 2 .value_kind: hidden_grid_dims - .offset: 80 .size: 8 .value_kind: hidden_hostcall_buffer .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 256 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z10helloworldv .private_segment_fixed_size: 0 .sgpr_count: 20 .sgpr_spill_count: 0 .symbol: _Z10helloworldv.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 32 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <hip/hip_runtime.h> #include <stdio.h> // Cuda supports printf in kernels for // hardware with compute compatibility >= 2.0 __global__ void helloworld() { // CUDA runtime uses device overloading or printf in kernels printf("Hello world!\n"); } int main(void) { helloworld<<<1,1>>>(); return 0; }
.text .file "helloworld.hip" .globl _Z25__device_stub__helloworldv # -- Begin function _Z25__device_stub__helloworldv .p2align 4, 0x90 .type _Z25__device_stub__helloworldv,@function _Z25__device_stub__helloworldv: # @_Z25__device_stub__helloworldv .cfi_startproc # %bb.0: subq $56, %rsp .cfi_def_cfa_offset 64 leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 48(%rsp), %r9 movl $_Z10helloworldv, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $72, %rsp .cfi_adjust_cfa_offset -72 retq .Lfunc_end0: .size _Z25__device_stub__helloworldv, .Lfunc_end0-_Z25__device_stub__helloworldv .cfi_endproc # -- End function .globl main # -- Begin function main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: subq $56, %rsp .cfi_def_cfa_offset 64 movabsq $4294967297, %rdi # imm = 0x100000001 movl $1, %esi movq %rdi, %rdx movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_2 # %bb.1: leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 48(%rsp), %r9 movl $_Z10helloworldv, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_2: xorl %eax, %eax addq $56, %rsp .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z10helloworldv, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z10helloworldv,@object # @_Z10helloworldv .section .rodata,"a",@progbits .globl _Z10helloworldv .p2align 3, 0x0 _Z10helloworldv: .quad _Z25__device_stub__helloworldv .size _Z10helloworldv, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z10helloworldv" .size .L__unnamed_1, 16 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z25__device_stub__helloworldv .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z10helloworldv .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z10helloworldv .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ IMAD.MOV.U32 R1, RZ, RZ, c[0x0][0x28] ; /* 0x00000a00ff017624 */ /* 0x000fe400078e00ff */ /*0010*/ MOV R0, 0x0 ; /* 0x0000000000007802 */ /* 0x000fe20000000f00 */ /*0020*/ IMAD.MOV.U32 R4, RZ, RZ, c[0x4][0x8] ; /* 0x01000200ff047624 */ /* 0x000fe200078e00ff */ /*0030*/ CS2R R6, SRZ ; /* 0x0000000000067805 */ /* 0x000fe2000001ff00 */ /*0040*/ IMAD.MOV.U32 R5, RZ, RZ, c[0x4][0xc] ; /* 0x01000300ff057624 */ /* 0x000fe200078e00ff */ /*0050*/ LDC.64 R2, c[0x4][R0] ; /* 0x0100000000027b82 */ /* 0x00006c0000000a00 */ /*0060*/ LEPC R8 ; /* 0x000000000008734e */ /* 0x000fe40000000000 */ /*0070*/ MOV R11, 0xe0 ; /* 0x000000e0000b7802 */ /* 0x000fe40000000f00 */ /*0080*/ MOV R20, 0x60 ; /* 0x0000006000147802 */ /* 0x000fe40000000f00 */ /*0090*/ MOV R21, 0x0 ; /* 0x0000000000157802 */ /* 0x000fe40000000f00 */ /*00a0*/ MOV R0, 0x0 ; /* 0x0000000000007802 */ /* 0x001fc40000000f00 */ /*00b0*/ IADD3 R20, P0, P1, -R20, R11, R8 ; /* 0x0000000b14147210 */ /* 0x000fc8000791e108 */ /*00c0*/ IADD3.X R21, ~R0, R21, R9, P0, P1 ; /* 0x0000001500157210 */ /* 0x000fc800007e2509 */ /*00d0*/ CALL.ABS.NOINC R2 ; /* 0x0000000002007343 */ /* 0x002fea0003c00000 */ /*00e0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00f0*/ BRA 0xf0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0100*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0110*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0120*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0130*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0140*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0150*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0160*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0170*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z10helloworldv .globl _Z10helloworldv .p2align 8 .type _Z10helloworldv,@function _Z10helloworldv: s_load_b64 s[2:3], s[0:1], 0x50 v_mbcnt_lo_u32_b32 v20, -1, 0 v_mov_b32_e32 v6, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_dual_mov_b32 v7, 0 :: v_dual_mov_b32 v4, v20 v_readfirstlane_b32 s0, v4 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_eq_u32_e64 s0, s0, v4 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_6 v_mov_b32_e32 v0, 0 s_mov_b32 s4, exec_lo s_waitcnt lgkmcnt(0) global_load_b64 v[8:9], v0, s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv s_clause 0x1 global_load_b64 v[1:2], v0, s[2:3] offset:40 global_load_b64 v[5:6], v0, s[2:3] s_waitcnt vmcnt(1) v_and_b32_e32 v1, v1, v8 v_and_b32_e32 v2, v2, v9 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_hi_u32 v3, v1, 24 v_mul_lo_u32 v2, v2, 24 v_mul_lo_u32 v1, v1, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v2, v3, v2 s_waitcnt vmcnt(0) v_add_co_u32 v1, vcc_lo, v5, v1 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v2, vcc_lo, v6, v2, vcc_lo global_load_b64 v[6:7], v[1:2], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[6:7], v0, v[6:9], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmpx_ne_u64_e64 v[6:7], v[8:9] s_cbranch_execz .LBB0_5 s_mov_b32 s5, 0 .p2align 6 .LBB0_3: s_sleep 1 s_clause 0x1 global_load_b64 v[1:2], v0, s[2:3] offset:40 global_load_b64 v[10:11], v0, s[2:3] v_dual_mov_b32 v9, v7 :: v_dual_mov_b32 v8, v6 s_waitcnt vmcnt(1) s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_and_b32_e32 v1, v1, v8 v_and_b32_e32 v7, v2, v9 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u64_u32 v[5:6], null, v1, 24, v[10:11] v_mov_b32_e32 v1, v6 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u64_u32 v[2:3], null, v7, 24, v[1:2] v_mov_b32_e32 v6, v2 global_load_b64 v[6:7], v[5:6], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[6:7], v0, v[6:9], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmp_eq_u64_e32 vcc_lo, v[6:7], v[8:9] s_or_b32 s5, vcc_lo, s5 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s5 s_cbranch_execnz .LBB0_3 s_or_b32 exec_lo, exec_lo, s5 .LBB0_5: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s4 .LBB0_6: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s1 v_mov_b32_e32 v5, 0 v_readfirstlane_b32 s4, v6 v_readfirstlane_b32 s5, v7 s_mov_b32 s8, exec_lo s_waitcnt lgkmcnt(0) s_clause 0x1 global_load_b64 v[8:9], v5, s[2:3] offset:40 global_load_b128 v[0:3], v5, s[2:3] s_waitcnt vmcnt(1) v_readfirstlane_b32 s6, v8 v_readfirstlane_b32 s7, v9 s_delay_alu instid0(VALU_DEP_1) s_and_b64 s[6:7], s[4:5], s[6:7] s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_8 v_dual_mov_b32 v6, s8 :: v_dual_mov_b32 v7, 0 s_mul_i32 s8, s7, 24 s_mul_hi_u32 s9, s6, 24 v_dual_mov_b32 v8, 2 :: v_dual_mov_b32 v9, 1 s_add_i32 s9, s9, s8 s_mul_i32 s8, s6, 24 s_waitcnt vmcnt(0) v_add_co_u32 v10, vcc_lo, v0, s8 v_add_co_ci_u32_e32 v11, vcc_lo, s9, v1, vcc_lo global_store_b128 v[10:11], v[6:9], off offset:8 .LBB0_8: s_or_b32 exec_lo, exec_lo, s1 s_lshl_b64 s[8:9], s[6:7], 12 v_lshlrev_b64 v[4:5], 6, v[4:5] s_waitcnt vmcnt(0) v_add_co_u32 v2, vcc_lo, v2, s8 v_add_co_ci_u32_e32 v7, vcc_lo, s9, v3, vcc_lo v_mov_b32_e32 v3, 0 s_mov_b32 s8, 0 s_delay_alu instid0(VALU_DEP_3) v_add_co_u32 v6, vcc_lo, v2, v4 v_mov_b32_e32 v2, 33 s_mov_b32 s9, s8 s_mov_b32 s10, s8 s_mov_b32 s11, s8 v_add_co_ci_u32_e32 v7, vcc_lo, v7, v5, vcc_lo v_mov_b32_e32 v4, v3 v_dual_mov_b32 v5, v3 :: v_dual_mov_b32 v8, s8 v_dual_mov_b32 v9, s9 :: v_dual_mov_b32 v10, s10 v_mov_b32_e32 v11, s11 s_clause 0x3 global_store_b128 v[6:7], v[2:5], off global_store_b128 v[6:7], v[8:11], off offset:16 global_store_b128 v[6:7], v[8:11], off offset:32 global_store_b128 v[6:7], v[8:11], off offset:48 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_16 v_dual_mov_b32 v10, 0 :: v_dual_mov_b32 v11, s4 v_mov_b32_e32 v12, s5 s_clause 0x1 global_load_b64 v[13:14], v10, s[2:3] offset:32 glc global_load_b64 v[2:3], v10, s[2:3] offset:40 s_waitcnt vmcnt(0) v_readfirstlane_b32 s8, v2 v_readfirstlane_b32 s9, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_b64 s[8:9], s[8:9], s[4:5] s_mul_i32 s9, s9, 24 s_mul_hi_u32 s10, s8, 24 s_mul_i32 s8, s8, 24 s_add_i32 s10, s10, s9 v_add_co_u32 v8, vcc_lo, v0, s8 v_add_co_ci_u32_e32 v9, vcc_lo, s10, v1, vcc_lo s_mov_b32 s8, exec_lo global_store_b64 v[8:9], v[13:14], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[4:5], v10, v[11:14], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmpx_ne_u64_e64 v[4:5], v[13:14] s_cbranch_execz .LBB0_12 s_mov_b32 s9, 0 .LBB0_11: v_dual_mov_b32 v2, s4 :: v_dual_mov_b32 v3, s5 s_sleep 1 global_store_b64 v[8:9], v[4:5], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v10, v[2:5], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[2:3], v[4:5] v_dual_mov_b32 v5, v3 :: v_dual_mov_b32 v4, v2 s_or_b32 s9, vcc_lo, s9 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s9 s_cbranch_execnz .LBB0_11 .LBB0_12: s_or_b32 exec_lo, exec_lo, s8 v_mov_b32_e32 v2, 0 s_mov_b32 s9, exec_lo s_mov_b32 s8, exec_lo v_mbcnt_lo_u32_b32 v4, s9, 0 global_load_b64 v[2:3], v2, s[2:3] offset:16 v_cmpx_eq_u32_e32 0, v4 s_cbranch_execz .LBB0_14 s_bcnt1_i32_b32 s9, s9 s_delay_alu instid0(SALU_CYCLE_1) v_dual_mov_b32 v5, 0 :: v_dual_mov_b32 v4, s9 s_waitcnt vmcnt(0) global_atomic_add_u64 v[2:3], v[4:5], off offset:8 .LBB0_14: s_or_b32 exec_lo, exec_lo, s8 s_waitcnt vmcnt(0) global_load_b64 v[4:5], v[2:3], off offset:16 s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, 0, v[4:5] s_cbranch_vccnz .LBB0_16 global_load_b32 v2, v[2:3], off offset:24 v_mov_b32_e32 v3, 0 s_waitcnt vmcnt(0) v_readfirstlane_b32 s8, v2 s_waitcnt_vscnt null, 0x0 global_store_b64 v[4:5], v[2:3], off s_and_b32 m0, s8, 0xff s_sendmsg sendmsg(MSG_INTERRUPT) .LBB0_16: s_or_b32 exec_lo, exec_lo, s1 s_mul_i32 s1, s7, 24 s_mul_hi_u32 s7, s6, 24 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) s_add_i32 s7, s7, s1 s_mul_i32 s1, s6, 24 v_add_co_u32 v0, vcc_lo, v0, s1 v_add_co_ci_u32_e32 v1, vcc_lo, s7, v1, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v0, 20 v_add_co_ci_u32_e32 v1, vcc_lo, 0, v1, vcc_lo s_branch .LBB0_20 .p2align 6 .LBB0_17: s_or_b32 exec_lo, exec_lo, s1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_readfirstlane_b32 s1, v2 s_cmp_eq_u32 s1, 0 s_cbranch_scc1 .LBB0_19 s_sleep 1 s_cbranch_execnz .LBB0_20 s_branch .LBB0_22 .p2align 6 .LBB0_19: s_branch .LBB0_22 .LBB0_20: v_mov_b32_e32 v2, 1 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_17 global_load_b32 v2, v[0:1], off glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_and_b32_e32 v2, 1, v2 s_branch .LBB0_17 .LBB0_22: global_load_b64 v[22:23], v[6:7], off s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_26 v_mov_b32_e32 v6, 0 s_clause 0x2 global_load_b64 v[2:3], v6, s[2:3] offset:40 global_load_b64 v[7:8], v6, s[2:3] offset:24 glc global_load_b64 v[4:5], v6, s[2:3] s_waitcnt vmcnt(2) v_add_co_u32 v9, vcc_lo, v2, 1 v_add_co_ci_u32_e32 v10, vcc_lo, 0, v3, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v9, s4 v_add_co_ci_u32_e32 v1, vcc_lo, s5, v10, vcc_lo s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_cmp_eq_u64_e32 vcc_lo, 0, v[0:1] v_dual_cndmask_b32 v1, v1, v10 :: v_dual_cndmask_b32 v0, v0, v9 v_and_b32_e32 v3, v1, v3 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_and_b32_e32 v2, v0, v2 v_mul_lo_u32 v3, v3, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_mul_hi_u32 v9, v2, 24 v_mul_lo_u32 v2, v2, 24 v_add_nc_u32_e32 v3, v9, v3 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_co_u32 v4, vcc_lo, v4, v2 v_mov_b32_e32 v2, v7 v_add_co_ci_u32_e32 v5, vcc_lo, v5, v3, vcc_lo v_mov_b32_e32 v3, v8 global_store_b64 v[4:5], v[7:8], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v6, v[0:3], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_ne_u64_e32 vcc_lo, v[2:3], v[7:8] s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_26 s_mov_b32 s0, 0 .LBB0_25: s_sleep 1 global_store_b64 v[4:5], v[2:3], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[7:8], v6, v[0:3], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[7:8], v[2:3] v_dual_mov_b32 v2, v7 :: v_dual_mov_b32 v3, v8 s_or_b32 s0, vcc_lo, s0 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s0 s_cbranch_execnz .LBB0_25 .LBB0_26: s_or_b32 exec_lo, exec_lo, s1 s_getpc_b64 s[4:5] s_add_u32 s4, s4, .str@rel32@lo+4 s_addc_u32 s5, s5, .str@rel32@hi+12 s_mov_b32 s0, -1 s_cmp_lg_u64 s[4:5], 0 s_cbranch_scc0 .LBB0_105 s_waitcnt vmcnt(0) v_dual_mov_b32 v1, v23 :: v_dual_and_b32 v0, -3, v22 v_mov_b32_e32 v25, 0 s_mov_b64 s[6:7], 14 s_branch .LBB0_29 .LBB0_28: s_or_b32 exec_lo, exec_lo, s1 s_sub_u32 s6, s6, s8 s_subb_u32 s7, s7, s9 s_add_u32 s4, s4, s8 s_addc_u32 s5, s5, s9 s_cmp_lg_u64 s[6:7], 0 s_cbranch_scc0 .LBB0_104 .LBB0_29: v_cmp_lt_u64_e64 s0, s[6:7], 56 s_delay_alu instid0(VALU_DEP_1) s_and_b32 s0, s0, exec_lo s_cselect_b32 s8, s6, 56 s_cselect_b32 s9, s7, 0 s_cmp_gt_u32 s8, 7 s_mov_b32 s0, -1 s_cbranch_scc1 .LBB0_34 v_mov_b32_e32 v2, 0 v_mov_b32_e32 v3, 0 s_cmp_eq_u32 s8, 0 s_cbranch_scc1 .LBB0_33 s_lshl_b64 s[0:1], s[8:9], 3 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], s[4:5] .LBB0_32: global_load_u8 v4, v25, s[12:13] s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v4 s_delay_alu instid0(VALU_DEP_1) v_lshlrev_b64 v[4:5], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_add_u32 s12, s12, 1 s_addc_u32 s13, s13, 0 s_cmp_lg_u32 s0, s10 v_or_b32_e32 v2, v4, v2 v_or_b32_e32 v3, v5, v3 s_cbranch_scc1 .LBB0_32 .LBB0_33: s_mov_b32 s0, 0 s_mov_b32 s15, 0 .LBB0_34: s_and_not1_b32 vcc_lo, exec_lo, s0 s_mov_b64 s[0:1], s[4:5] s_cbranch_vccnz .LBB0_36 global_load_b64 v[2:3], v25, s[4:5] s_add_i32 s15, s8, -8 s_add_u32 s0, s4, 8 s_addc_u32 s1, s5, 0 .LBB0_36: s_cmp_gt_u32 s15, 7 s_cbranch_scc1 .LBB0_41 v_mov_b32_e32 v4, 0 v_mov_b32_e32 v5, 0 s_cmp_eq_u32 s15, 0 s_cbranch_scc1 .LBB0_40 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_39: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v6, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v6 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[6:7], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s15, s12 v_or_b32_e32 v4, v6, v4 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v5, v7, v5 s_cbranch_scc1 .LBB0_39 .LBB0_40: s_mov_b32 s14, 0 s_cbranch_execz .LBB0_42 s_branch .LBB0_43 .LBB0_41: .LBB0_42: global_load_b64 v[4:5], v25, s[0:1] s_add_i32 s14, s15, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_43: s_cmp_gt_u32 s14, 7 s_cbranch_scc1 .LBB0_48 v_mov_b32_e32 v6, 0 v_mov_b32_e32 v7, 0 s_cmp_eq_u32 s14, 0 s_cbranch_scc1 .LBB0_47 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_46: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v8, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v8 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[8:9], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s14, s12 v_or_b32_e32 v6, v8, v6 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v7, v9, v7 s_cbranch_scc1 .LBB0_46 .LBB0_47: s_mov_b32 s15, 0 s_cbranch_execz .LBB0_49 s_branch .LBB0_50 .LBB0_48: .LBB0_49: global_load_b64 v[6:7], v25, s[0:1] s_add_i32 s15, s14, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_50: s_cmp_gt_u32 s15, 7 s_cbranch_scc1 .LBB0_55 v_mov_b32_e32 v8, 0 v_mov_b32_e32 v9, 0 s_cmp_eq_u32 s15, 0 s_cbranch_scc1 .LBB0_54 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_53: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v10, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v10 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[10:11], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s15, s12 v_or_b32_e32 v8, v10, v8 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v9, v11, v9 s_cbranch_scc1 .LBB0_53 .LBB0_54: s_mov_b32 s14, 0 s_cbranch_execz .LBB0_56 s_branch .LBB0_57 .LBB0_55: .LBB0_56: global_load_b64 v[8:9], v25, s[0:1] s_add_i32 s14, s15, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_57: s_cmp_gt_u32 s14, 7 s_cbranch_scc1 .LBB0_62 v_mov_b32_e32 v10, 0 v_mov_b32_e32 v11, 0 s_cmp_eq_u32 s14, 0 s_cbranch_scc1 .LBB0_61 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_60: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v12, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v12 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[12:13], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s14, s12 v_or_b32_e32 v10, v12, v10 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v11, v13, v11 s_cbranch_scc1 .LBB0_60 .LBB0_61: s_mov_b32 s15, 0 s_cbranch_execz .LBB0_63 s_branch .LBB0_64 .LBB0_62: .LBB0_63: global_load_b64 v[10:11], v25, s[0:1] s_add_i32 s15, s14, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_64: s_cmp_gt_u32 s15, 7 s_cbranch_scc1 .LBB0_69 v_mov_b32_e32 v12, 0 v_mov_b32_e32 v13, 0 s_cmp_eq_u32 s15, 0 s_cbranch_scc1 .LBB0_68 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], 0 .LBB0_67: s_delay_alu instid0(SALU_CYCLE_1) s_add_u32 s16, s0, s12 s_addc_u32 s17, s1, s13 s_add_u32 s12, s12, 1 global_load_u8 v14, v25, s[16:17] s_addc_u32 s13, s13, 0 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v14 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_lshlrev_b64 v[14:15], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_cmp_lg_u32 s15, s12 v_or_b32_e32 v12, v14, v12 s_delay_alu instid0(VALU_DEP_2) v_or_b32_e32 v13, v15, v13 s_cbranch_scc1 .LBB0_67 .LBB0_68: s_mov_b32 s14, 0 s_cbranch_execz .LBB0_70 s_branch .LBB0_71 .LBB0_69: .LBB0_70: global_load_b64 v[12:13], v25, s[0:1] s_add_i32 s14, s15, -8 s_add_u32 s0, s0, 8 s_addc_u32 s1, s1, 0 .LBB0_71: s_cmp_gt_u32 s14, 7 s_cbranch_scc1 .LBB0_76 v_mov_b32_e32 v14, 0 v_mov_b32_e32 v15, 0 s_cmp_eq_u32 s14, 0 s_cbranch_scc1 .LBB0_75 s_mov_b64 s[10:11], 0 s_mov_b64 s[12:13], s[0:1] .LBB0_74: global_load_u8 v16, v25, s[12:13] s_add_i32 s14, s14, -1 s_waitcnt vmcnt(0) v_and_b32_e32 v24, 0xffff, v16 s_delay_alu instid0(VALU_DEP_1) v_lshlrev_b64 v[16:17], s10, v[24:25] s_add_u32 s10, s10, 8 s_addc_u32 s11, s11, 0 s_add_u32 s12, s12, 1 s_addc_u32 s13, s13, 0 s_cmp_lg_u32 s14, 0 v_or_b32_e32 v14, v16, v14 v_or_b32_e32 v15, v17, v15 s_cbranch_scc1 .LBB0_74 .LBB0_75: s_cbranch_execz .LBB0_77 s_branch .LBB0_78 .LBB0_76: .LBB0_77: global_load_b64 v[14:15], v25, s[0:1] .LBB0_78: v_mov_b32_e32 v24, v20 v_mov_b32_e32 v26, 0 v_mov_b32_e32 v27, 0 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_readfirstlane_b32 s0, v24 v_cmp_eq_u32_e64 s0, s0, v24 s_delay_alu instid0(VALU_DEP_1) s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_84 global_load_b64 v[18:19], v25, s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv s_clause 0x1 global_load_b64 v[16:17], v25, s[2:3] offset:40 global_load_b64 v[26:27], v25, s[2:3] s_mov_b32 s10, exec_lo s_waitcnt vmcnt(1) v_and_b32_e32 v17, v17, v19 v_and_b32_e32 v16, v16, v18 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_lo_u32 v17, v17, 24 v_mul_hi_u32 v21, v16, 24 v_mul_lo_u32 v16, v16, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v17, v21, v17 s_waitcnt vmcnt(0) v_add_co_u32 v16, vcc_lo, v26, v16 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v17, vcc_lo, v27, v17, vcc_lo global_load_b64 v[16:17], v[16:17], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[26:27], v25, v[16:19], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmpx_ne_u64_e64 v[26:27], v[18:19] s_cbranch_execz .LBB0_83 s_mov_b32 s11, 0 .p2align 6 .LBB0_81: s_sleep 1 s_clause 0x1 global_load_b64 v[16:17], v25, s[2:3] offset:40 global_load_b64 v[28:29], v25, s[2:3] v_dual_mov_b32 v18, v26 :: v_dual_mov_b32 v19, v27 s_waitcnt vmcnt(1) s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_and_b32_e32 v16, v16, v18 s_waitcnt vmcnt(0) v_mad_u64_u32 v[26:27], null, v16, 24, v[28:29] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_dual_mov_b32 v16, v27 :: v_dual_and_b32 v17, v17, v19 v_mad_u64_u32 v[27:28], null, v17, 24, v[16:17] global_load_b64 v[16:17], v[26:27], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[26:27], v25, v[16:19], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmp_eq_u64_e32 vcc_lo, v[26:27], v[18:19] s_or_b32 s11, vcc_lo, s11 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s11 s_cbranch_execnz .LBB0_81 s_or_b32 exec_lo, exec_lo, s11 .LBB0_83: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s10 .LBB0_84: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s1 s_clause 0x1 global_load_b64 v[28:29], v25, s[2:3] offset:40 global_load_b128 v[16:19], v25, s[2:3] v_readfirstlane_b32 s10, v26 v_readfirstlane_b32 s11, v27 s_mov_b32 s14, exec_lo s_waitcnt vmcnt(1) v_readfirstlane_b32 s12, v28 v_readfirstlane_b32 s13, v29 s_delay_alu instid0(VALU_DEP_1) s_and_b64 s[12:13], s[10:11], s[12:13] s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_86 v_dual_mov_b32 v26, s14 :: v_dual_mov_b32 v27, 0 s_mul_i32 s14, s13, 24 s_mul_hi_u32 s15, s12, 24 v_dual_mov_b32 v28, 2 :: v_dual_mov_b32 v29, 1 s_add_i32 s15, s15, s14 s_mul_i32 s14, s12, 24 s_waitcnt vmcnt(0) v_add_co_u32 v30, vcc_lo, v16, s14 v_add_co_ci_u32_e32 v31, vcc_lo, s15, v17, vcc_lo global_store_b128 v[30:31], v[26:29], off offset:8 .LBB0_86: s_or_b32 exec_lo, exec_lo, s1 v_cmp_gt_u64_e64 vcc_lo, s[6:7], 56 v_or_b32_e32 v21, 2, v0 s_lshl_b64 s[14:15], s[12:13], 12 v_lshlrev_b64 v[26:27], 6, v[24:25] s_lshl_b32 s1, s8, 2 s_delay_alu instid0(SALU_CYCLE_1) s_add_i32 s1, s1, 28 v_cndmask_b32_e32 v0, v21, v0, vcc_lo s_waitcnt vmcnt(0) v_add_co_u32 v18, vcc_lo, v18, s14 v_add_co_ci_u32_e32 v19, vcc_lo, s15, v19, vcc_lo s_and_b32 s1, s1, 0x1e0 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_co_u32 v18, vcc_lo, v18, v26 v_and_or_b32 v0, v0, 0xffffff1f, s1 v_add_co_ci_u32_e32 v19, vcc_lo, v19, v27, vcc_lo s_clause 0x3 global_store_b128 v[18:19], v[0:3], off global_store_b128 v[18:19], v[4:7], off offset:16 global_store_b128 v[18:19], v[8:11], off offset:32 global_store_b128 v[18:19], v[12:15], off offset:48 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_94 s_clause 0x1 global_load_b64 v[8:9], v25, s[2:3] offset:32 glc global_load_b64 v[0:1], v25, s[2:3] offset:40 v_dual_mov_b32 v6, s10 :: v_dual_mov_b32 v7, s11 s_waitcnt vmcnt(0) v_readfirstlane_b32 s14, v0 v_readfirstlane_b32 s15, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_b64 s[14:15], s[14:15], s[10:11] s_mul_i32 s15, s15, 24 s_mul_hi_u32 s16, s14, 24 s_mul_i32 s14, s14, 24 s_add_i32 s16, s16, s15 v_add_co_u32 v4, vcc_lo, v16, s14 v_add_co_ci_u32_e32 v5, vcc_lo, s16, v17, vcc_lo s_mov_b32 s14, exec_lo global_store_b64 v[4:5], v[8:9], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v25, v[6:9], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmpx_ne_u64_e64 v[2:3], v[8:9] s_cbranch_execz .LBB0_90 s_mov_b32 s15, 0 .LBB0_89: v_dual_mov_b32 v0, s10 :: v_dual_mov_b32 v1, s11 s_sleep 1 global_store_b64 v[4:5], v[2:3], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[0:1], v25, v[0:3], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[2:3] v_dual_mov_b32 v3, v1 :: v_dual_mov_b32 v2, v0 s_or_b32 s15, vcc_lo, s15 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s15 s_cbranch_execnz .LBB0_89 .LBB0_90: s_or_b32 exec_lo, exec_lo, s14 global_load_b64 v[0:1], v25, s[2:3] offset:16 s_mov_b32 s15, exec_lo s_mov_b32 s14, exec_lo v_mbcnt_lo_u32_b32 v2, s15, 0 s_delay_alu instid0(VALU_DEP_1) v_cmpx_eq_u32_e32 0, v2 s_cbranch_execz .LBB0_92 s_bcnt1_i32_b32 s15, s15 s_delay_alu instid0(SALU_CYCLE_1) v_dual_mov_b32 v3, 0 :: v_dual_mov_b32 v2, s15 s_waitcnt vmcnt(0) global_atomic_add_u64 v[0:1], v[2:3], off offset:8 .LBB0_92: s_or_b32 exec_lo, exec_lo, s14 s_waitcnt vmcnt(0) global_load_b64 v[2:3], v[0:1], off offset:16 s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, 0, v[2:3] s_cbranch_vccnz .LBB0_94 global_load_b32 v24, v[0:1], off offset:24 s_waitcnt vmcnt(0) v_readfirstlane_b32 s14, v24 s_waitcnt_vscnt null, 0x0 global_store_b64 v[2:3], v[24:25], off s_and_b32 m0, s14, 0xff s_sendmsg sendmsg(MSG_INTERRUPT) .LBB0_94: s_or_b32 exec_lo, exec_lo, s1 s_mul_i32 s1, s13, 24 s_mul_hi_u32 s13, s12, 24 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) s_add_i32 s13, s13, s1 s_mul_i32 s1, s12, 24 v_add_co_u32 v0, vcc_lo, v16, s1 v_add_co_ci_u32_e32 v1, vcc_lo, s13, v17, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v0, 20 v_add_co_ci_u32_e32 v1, vcc_lo, 0, v1, vcc_lo s_branch .LBB0_98 .p2align 6 .LBB0_95: s_or_b32 exec_lo, exec_lo, s1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_readfirstlane_b32 s1, v2 s_cmp_eq_u32 s1, 0 s_cbranch_scc1 .LBB0_97 s_sleep 1 s_cbranch_execnz .LBB0_98 s_branch .LBB0_100 .p2align 6 .LBB0_97: s_branch .LBB0_100 .LBB0_98: v_mov_b32_e32 v2, 1 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_95 global_load_b32 v2, v[0:1], off glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_and_b32_e32 v2, 1, v2 s_branch .LBB0_95 .LBB0_100: global_load_b64 v[0:1], v[18:19], off s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_28 s_clause 0x2 global_load_b64 v[4:5], v25, s[2:3] offset:40 global_load_b64 v[8:9], v25, s[2:3] offset:24 glc global_load_b64 v[6:7], v25, s[2:3] s_waitcnt vmcnt(2) v_add_co_u32 v10, vcc_lo, v4, 1 v_add_co_ci_u32_e32 v11, vcc_lo, 0, v5, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v2, vcc_lo, v10, s10 v_add_co_ci_u32_e32 v3, vcc_lo, s11, v11, vcc_lo s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_cmp_eq_u64_e32 vcc_lo, 0, v[2:3] v_dual_cndmask_b32 v3, v3, v11 :: v_dual_cndmask_b32 v2, v2, v10 v_and_b32_e32 v5, v3, v5 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_and_b32_e32 v4, v2, v4 v_mul_hi_u32 v10, v4, 24 v_mul_lo_u32 v4, v4, 24 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_add_co_u32 v6, vcc_lo, v6, v4 v_mov_b32_e32 v4, v8 v_mul_lo_u32 v5, v5, 24 v_add_nc_u32_e32 v5, v10, v5 s_delay_alu instid0(VALU_DEP_1) v_add_co_ci_u32_e32 v7, vcc_lo, v7, v5, vcc_lo v_mov_b32_e32 v5, v9 global_store_b64 v[6:7], v[8:9], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[4:5], v25, v[2:5], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_ne_u64_e32 vcc_lo, v[4:5], v[8:9] s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_28 s_mov_b32 s0, 0 .LBB0_103: s_sleep 1 global_store_b64 v[6:7], v[4:5], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[8:9], v25, v[2:5], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[8:9], v[4:5] v_dual_mov_b32 v4, v8 :: v_dual_mov_b32 v5, v9 s_or_b32 s0, vcc_lo, s0 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s0 s_cbranch_execnz .LBB0_103 s_branch .LBB0_28 .LBB0_104: s_mov_b32 s0, 0 .LBB0_105: s_delay_alu instid0(SALU_CYCLE_1) s_and_b32 vcc_lo, exec_lo, s0 s_cbranch_vccz .LBB0_132 v_readfirstlane_b32 s0, v20 v_mov_b32_e32 v4, 0 v_mov_b32_e32 v5, 0 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_eq_u32_e64 s0, s0, v20 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_112 s_waitcnt vmcnt(0) v_mov_b32_e32 v0, 0 s_mov_b32 s4, exec_lo global_load_b64 v[6:7], v0, s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv s_clause 0x1 global_load_b64 v[1:2], v0, s[2:3] offset:40 global_load_b64 v[3:4], v0, s[2:3] s_waitcnt vmcnt(1) v_and_b32_e32 v1, v1, v6 v_and_b32_e32 v2, v2, v7 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_hi_u32 v5, v1, 24 v_mul_lo_u32 v2, v2, 24 v_mul_lo_u32 v1, v1, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v2, v5, v2 s_waitcnt vmcnt(0) v_add_co_u32 v1, vcc_lo, v3, v1 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v2, vcc_lo, v4, v2, vcc_lo global_load_b64 v[4:5], v[1:2], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[4:5], v0, v[4:7], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmpx_ne_u64_e64 v[4:5], v[6:7] s_cbranch_execz .LBB0_111 s_mov_b32 s5, 0 .p2align 6 .LBB0_109: s_sleep 1 s_clause 0x1 global_load_b64 v[1:2], v0, s[2:3] offset:40 global_load_b64 v[8:9], v0, s[2:3] v_dual_mov_b32 v7, v5 :: v_dual_mov_b32 v6, v4 s_waitcnt vmcnt(1) s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_and_b32_e32 v1, v1, v6 s_waitcnt vmcnt(0) v_mad_u64_u32 v[3:4], null, v1, 24, v[8:9] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_dual_mov_b32 v1, v4 :: v_dual_and_b32 v2, v2, v7 v_mad_u64_u32 v[4:5], null, v2, 24, v[1:2] global_load_b64 v[4:5], v[3:4], off glc s_waitcnt vmcnt(0) global_atomic_cmpswap_b64 v[4:5], v0, v[4:7], s[2:3] offset:24 glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_cmp_eq_u64_e32 vcc_lo, v[4:5], v[6:7] s_or_b32 s5, vcc_lo, s5 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s5 s_cbranch_execnz .LBB0_109 s_or_b32 exec_lo, exec_lo, s5 .LBB0_111: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s4 .LBB0_112: s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 exec_lo, exec_lo, s1 v_mov_b32_e32 v21, 0 v_readfirstlane_b32 s4, v4 v_readfirstlane_b32 s5, v5 s_mov_b32 s8, exec_lo s_clause 0x1 global_load_b64 v[6:7], v21, s[2:3] offset:40 global_load_b128 v[0:3], v21, s[2:3] s_waitcnt vmcnt(1) v_readfirstlane_b32 s6, v6 v_readfirstlane_b32 s7, v7 s_delay_alu instid0(VALU_DEP_1) s_and_b64 s[6:7], s[4:5], s[6:7] s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_114 v_dual_mov_b32 v4, s8 :: v_dual_mov_b32 v5, 0 s_mul_i32 s8, s7, 24 s_mul_hi_u32 s9, s6, 24 v_dual_mov_b32 v6, 2 :: v_dual_mov_b32 v7, 1 s_add_i32 s9, s9, s8 s_mul_i32 s8, s6, 24 s_waitcnt vmcnt(0) v_add_co_u32 v8, vcc_lo, v0, s8 v_add_co_ci_u32_e32 v9, vcc_lo, s9, v1, vcc_lo global_store_b128 v[8:9], v[4:7], off offset:8 .LBB0_114: s_or_b32 exec_lo, exec_lo, s1 s_lshl_b64 s[8:9], s[6:7], 12 v_and_or_b32 v22, v22, 0xffffff1d, 34 s_waitcnt vmcnt(0) v_add_co_u32 v4, vcc_lo, v2, s8 v_add_co_ci_u32_e32 v5, vcc_lo, s9, v3, vcc_lo v_lshlrev_b64 v[2:3], 6, v[20:21] s_mov_b32 s8, 0 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) s_mov_b32 s9, s8 s_mov_b32 s10, s8 s_mov_b32 s11, s8 v_add_co_u32 v8, vcc_lo, v4, v2 v_mov_b32_e32 v6, 0 v_add_co_ci_u32_e32 v9, vcc_lo, v5, v3, vcc_lo v_dual_mov_b32 v2, s8 :: v_dual_mov_b32 v5, s11 v_dual_mov_b32 v3, s9 :: v_dual_mov_b32 v4, s10 s_delay_alu instid0(VALU_DEP_4) v_mov_b32_e32 v7, v6 s_clause 0x4 global_store_b64 v[8:9], v[22:23], off global_store_b128 v[8:9], v[2:5], off offset:8 global_store_b128 v[8:9], v[2:5], off offset:24 global_store_b128 v[8:9], v[2:5], off offset:40 global_store_b64 v[8:9], v[6:7], off offset:56 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_122 v_dual_mov_b32 v8, 0 :: v_dual_mov_b32 v9, s4 v_mov_b32_e32 v10, s5 s_clause 0x1 global_load_b64 v[11:12], v8, s[2:3] offset:32 glc global_load_b64 v[2:3], v8, s[2:3] offset:40 s_waitcnt vmcnt(0) v_readfirstlane_b32 s8, v2 v_readfirstlane_b32 s9, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_b64 s[8:9], s[8:9], s[4:5] s_mul_i32 s9, s9, 24 s_mul_hi_u32 s10, s8, 24 s_mul_i32 s8, s8, 24 s_add_i32 s10, s10, s9 v_add_co_u32 v6, vcc_lo, v0, s8 v_add_co_ci_u32_e32 v7, vcc_lo, s10, v1, vcc_lo s_mov_b32 s8, exec_lo global_store_b64 v[6:7], v[11:12], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[4:5], v8, v[9:12], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmpx_ne_u64_e64 v[4:5], v[11:12] s_cbranch_execz .LBB0_118 s_mov_b32 s9, 0 .LBB0_117: v_dual_mov_b32 v2, s4 :: v_dual_mov_b32 v3, s5 s_sleep 1 global_store_b64 v[6:7], v[4:5], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v8, v[2:5], s[2:3] offset:32 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[2:3], v[4:5] v_dual_mov_b32 v5, v3 :: v_dual_mov_b32 v4, v2 s_or_b32 s9, vcc_lo, s9 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s9 s_cbranch_execnz .LBB0_117 .LBB0_118: s_or_b32 exec_lo, exec_lo, s8 v_mov_b32_e32 v2, 0 s_mov_b32 s9, exec_lo s_mov_b32 s8, exec_lo v_mbcnt_lo_u32_b32 v4, s9, 0 global_load_b64 v[2:3], v2, s[2:3] offset:16 v_cmpx_eq_u32_e32 0, v4 s_cbranch_execz .LBB0_120 s_bcnt1_i32_b32 s9, s9 s_delay_alu instid0(SALU_CYCLE_1) v_dual_mov_b32 v5, 0 :: v_dual_mov_b32 v4, s9 s_waitcnt vmcnt(0) global_atomic_add_u64 v[2:3], v[4:5], off offset:8 .LBB0_120: s_or_b32 exec_lo, exec_lo, s8 s_waitcnt vmcnt(0) global_load_b64 v[4:5], v[2:3], off offset:16 s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, 0, v[4:5] s_cbranch_vccnz .LBB0_122 global_load_b32 v2, v[2:3], off offset:24 v_mov_b32_e32 v3, 0 s_waitcnt vmcnt(0) v_readfirstlane_b32 s8, v2 s_waitcnt_vscnt null, 0x0 global_store_b64 v[4:5], v[2:3], off s_and_b32 m0, s8, 0xff s_sendmsg sendmsg(MSG_INTERRUPT) .LBB0_122: s_or_b32 exec_lo, exec_lo, s1 s_mul_i32 s1, s7, 24 s_mul_hi_u32 s7, s6, 24 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) s_add_i32 s7, s7, s1 s_mul_i32 s1, s6, 24 v_add_co_u32 v0, vcc_lo, v0, s1 v_add_co_ci_u32_e32 v1, vcc_lo, s7, v1, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v0, 20 v_add_co_ci_u32_e32 v1, vcc_lo, 0, v1, vcc_lo s_branch .LBB0_126 .p2align 6 .LBB0_123: s_or_b32 exec_lo, exec_lo, s1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_readfirstlane_b32 s1, v2 s_cmp_eq_u32 s1, 0 s_cbranch_scc1 .LBB0_125 s_sleep 1 s_cbranch_execnz .LBB0_126 s_branch .LBB0_128 .p2align 6 .LBB0_125: s_branch .LBB0_128 .LBB0_126: v_mov_b32_e32 v2, 1 s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_123 global_load_b32 v2, v[0:1], off glc s_waitcnt vmcnt(0) buffer_gl1_inv buffer_gl0_inv v_and_b32_e32 v2, 1, v2 s_branch .LBB0_123 .LBB0_128: s_and_saveexec_b32 s1, s0 s_cbranch_execz .LBB0_132 v_mov_b32_e32 v6, 0 s_clause 0x2 global_load_b64 v[2:3], v6, s[2:3] offset:40 global_load_b64 v[7:8], v6, s[2:3] offset:24 glc global_load_b64 v[4:5], v6, s[2:3] s_waitcnt vmcnt(2) v_add_co_u32 v9, vcc_lo, v2, 1 v_add_co_ci_u32_e32 v10, vcc_lo, 0, v3, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, v9, s4 v_add_co_ci_u32_e32 v1, vcc_lo, s5, v10, vcc_lo s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_cmp_eq_u64_e32 vcc_lo, 0, v[0:1] v_dual_cndmask_b32 v1, v1, v10 :: v_dual_cndmask_b32 v0, v0, v9 v_and_b32_e32 v3, v1, v3 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_and_b32_e32 v2, v0, v2 v_mul_lo_u32 v3, v3, 24 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_mul_hi_u32 v9, v2, 24 v_mul_lo_u32 v2, v2, 24 v_add_nc_u32_e32 v3, v9, v3 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_co_u32 v4, vcc_lo, v4, v2 v_mov_b32_e32 v2, v7 v_add_co_ci_u32_e32 v5, vcc_lo, v5, v3, vcc_lo v_mov_b32_e32 v3, v8 global_store_b64 v[4:5], v[7:8], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[2:3], v6, v[0:3], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_ne_u64_e32 vcc_lo, v[2:3], v[7:8] s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_132 s_mov_b32 s0, 0 .LBB0_131: s_sleep 1 global_store_b64 v[4:5], v[2:3], off s_waitcnt_vscnt null, 0x0 global_atomic_cmpswap_b64 v[7:8], v6, v[0:3], s[2:3] offset:24 glc s_waitcnt vmcnt(0) v_cmp_eq_u64_e32 vcc_lo, v[7:8], v[2:3] v_dual_mov_b32 v2, v7 :: v_dual_mov_b32 v3, v8 s_or_b32 s0, vcc_lo, s0 s_delay_alu instid0(SALU_CYCLE_1) s_and_not1_b32 exec_lo, exec_lo, s0 s_cbranch_execnz .LBB0_131 .LBB0_132: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z10helloworldv .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 256 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 32 .amdhsa_next_free_sgpr 18 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z10helloworldv, .Lfunc_end0-_Z10helloworldv .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type .str,@object .section .rodata.str1.1,"aMS",@progbits,1 .str: .asciz "Hello world!\n" .size .str, 14 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .offset: 0 .size: 4 .value_kind: hidden_block_count_x - .offset: 4 .size: 4 .value_kind: hidden_block_count_y - .offset: 8 .size: 4 .value_kind: hidden_block_count_z - .offset: 12 .size: 2 .value_kind: hidden_group_size_x - .offset: 14 .size: 2 .value_kind: hidden_group_size_y - .offset: 16 .size: 2 .value_kind: hidden_group_size_z - .offset: 18 .size: 2 .value_kind: hidden_remainder_x - .offset: 20 .size: 2 .value_kind: hidden_remainder_y - .offset: 22 .size: 2 .value_kind: hidden_remainder_z - .offset: 40 .size: 8 .value_kind: hidden_global_offset_x - .offset: 48 .size: 8 .value_kind: hidden_global_offset_y - .offset: 56 .size: 8 .value_kind: hidden_global_offset_z - .offset: 64 .size: 2 .value_kind: hidden_grid_dims - .offset: 80 .size: 8 .value_kind: hidden_hostcall_buffer .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 256 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z10helloworldv .private_segment_fixed_size: 0 .sgpr_count: 20 .sgpr_spill_count: 0 .symbol: _Z10helloworldv.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 32 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_0014b04c_00000000-6_helloworld.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2060: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2060: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z29__device_stub__Z10helloworldvv .type _Z29__device_stub__Z10helloworldvv, @function _Z29__device_stub__Z10helloworldvv: .LFB2082: .cfi_startproc endbr64 subq $88, %rsp .cfi_def_cfa_offset 96 movq %fs:40, %rax movq %rax, 72(%rsp) xorl %eax, %eax movl $1, 16(%rsp) movl $1, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) leaq 8(%rsp), %rcx movq %rsp, %rdx leaq 28(%rsp), %rsi leaq 16(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 72(%rsp), %rax subq %fs:40, %rax jne .L8 addq $88, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 8(%rsp) .cfi_def_cfa_offset 104 pushq 8(%rsp) .cfi_def_cfa_offset 112 leaq 80(%rsp), %r9 movq 44(%rsp), %rcx movl 52(%rsp), %r8d movq 32(%rsp), %rsi movl 40(%rsp), %edx leaq _Z10helloworldv(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 96 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2082: .size _Z29__device_stub__Z10helloworldvv, .-_Z29__device_stub__Z10helloworldvv .globl _Z10helloworldv .type _Z10helloworldv, @function _Z10helloworldv: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z10helloworldvv addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z10helloworldv, .-_Z10helloworldv .globl main .type main, @function main: .LFB2057: .cfi_startproc endbr64 subq $40, %rsp .cfi_def_cfa_offset 48 movl $1, 20(%rsp) movl $1, 24(%rsp) movl $1, 8(%rsp) movl $1, 12(%rsp) movl $0, %r9d movl $0, %r8d movq 20(%rsp), %rdx movl $1, %ecx movq 8(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L14 .L12: movl $0, %eax addq $40, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L14: .cfi_restore_state call _Z29__device_stub__Z10helloworldvv jmp .L12 .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z10helloworldv" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2085: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z10helloworldv(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2085: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "helloworld.hip" .globl _Z25__device_stub__helloworldv # -- Begin function _Z25__device_stub__helloworldv .p2align 4, 0x90 .type _Z25__device_stub__helloworldv,@function _Z25__device_stub__helloworldv: # @_Z25__device_stub__helloworldv .cfi_startproc # %bb.0: subq $56, %rsp .cfi_def_cfa_offset 64 leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 48(%rsp), %r9 movl $_Z10helloworldv, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $72, %rsp .cfi_adjust_cfa_offset -72 retq .Lfunc_end0: .size _Z25__device_stub__helloworldv, .Lfunc_end0-_Z25__device_stub__helloworldv .cfi_endproc # -- End function .globl main # -- Begin function main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: subq $56, %rsp .cfi_def_cfa_offset 64 movabsq $4294967297, %rdi # imm = 0x100000001 movl $1, %esi movq %rdi, %rdx movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_2 # %bb.1: leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 48(%rsp), %r9 movl $_Z10helloworldv, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_2: xorl %eax, %eax addq $56, %rsp .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z10helloworldv, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z10helloworldv,@object # @_Z10helloworldv .section .rodata,"a",@progbits .globl _Z10helloworldv .p2align 3, 0x0 _Z10helloworldv: .quad _Z25__device_stub__helloworldv .size _Z10helloworldv, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z10helloworldv" .size .L__unnamed_1, 16 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z25__device_stub__helloworldv .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z10helloworldv .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
/* Small CUDA exercise to try to improve efficiency by using two separate streams to set up a staged copying and execution (when instead of one large copy, followed by one large kernel computation, one does it in many small chunks, with copying and computing done in parallel in two streams, after first chunk was copied to the device). For the purpose of this exercise, ignore the copying of the results from the device to host at the end; only do the staged copy and execute for the copying of the initial data to the device + the kernel. We use cudaMallocHost to allocate arrays on host in pinned memory, which is both results in faster copying to/from GPU (compared to malloc), and also a CUDA requirement for copying running concurrently with a kernel. Make sure that the "Result:" value printed by the code is (almost) identical in both original and modified versions of the code. If not, you have a bug! Hints: You will have to use the following CUDA functions: - cudaStreamCreate - cudaMemcpyAsync - cudaStreamDestroy - cudaDeviceSynchronize * You will have to set up a for loop for multiple chunks copying and kernel execution; * Number of chunks should be a variable (or macro parameter); for simplicity, make NMAX dividable by the number of chunks; * In cudaMemcpyAsync the first two arguments should be "&d_A[ind], &h_A[ind]", not "d_A, h_A", ind being the starting index for the current chunk to copy; * You'll have to pass two more arguments to the kernel - ind and number of threads per chunk; * Nblocks will be different - shoul be computed per chunk. * You'll have to modify the kernel slightly; At the end, you should get the timings (based on 10 runs, NMAX=1000000, BLOCK_SIZE=128) similar to this: NCHUNKS t, ms - 2.76 - the original (non-staged) version of the code 1 2.72 - result is similar to the non-staged code 2 2.08 - even with only 2 chunks, we already see 33% speedup 4 1.79 5 1.75 10 1.72 - seems to be the best timing, 60% faster than the original code 20 1.87 - as NCHUNKS increases, the results get worse. Why? 100 3.53 - for too many NCHUNKS results can get even worse than in non-staged code To compile: nvcc -arch=sm_20 -O2 staged.cu -o staged The best/average timings: ../best_time.sh ./staged */ #include <sys/time.h> #include <ctype.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <time.h> #include <unistd.h> // Number of times to run the test (for better timings accuracy): #define NTESTS 100 // Number of threads in one block (possible range is 32...1024): #define BLOCK_SIZE 128 // Total number of threads (total number of elements to process in the kernel): #define NMAX 1000000 /* Subtract the `struct timeval' values X and Y, storing the result in RESULT. Return 1 if the difference is negative, otherwise 0. */ // It messes up with y! int timeval_subtract (double *result, struct timeval *x, struct timeval *y) { struct timeval result0; /* Perform the carry for the later subtraction by updating y. */ if (x->tv_usec < y->tv_usec) { int nsec = (y->tv_usec - x->tv_usec) / 1000000 + 1; y->tv_usec -= 1000000 * nsec; y->tv_sec += nsec; } if (x->tv_usec - y->tv_usec > 1000000) { int nsec = (y->tv_usec - x->tv_usec) / 1000000; y->tv_usec += 1000000 * nsec; y->tv_sec -= nsec; } /* Compute the time remaining to wait. tv_usec is certainly positive. */ result0.tv_sec = x->tv_sec - y->tv_sec; result0.tv_usec = x->tv_usec - y->tv_usec; *result = ((double)result0.tv_usec)/1e6 + (double)result0.tv_sec; /* Return 1 if result is negative. */ return x->tv_sec < y->tv_sec; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ // The kernel: __global__ void MyKernel (double *d_A, double *d_B) { double x, y, z; int i = threadIdx.x + blockDim.x * blockIdx.x; if (i >= NMAX) return; // Some meaningless cpu-intensive computation: x = pow(d_A[i], 2.71); y = pow(d_A[i], 0.35); z = 2*x + 5*y; d_B[i] = x + y + z + x*y + x/y + y/z; return; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ int main (int argc,char **argv) { struct timeval tdr0, tdr1, tdr, tdr01; double restime, restime0, restime1; int devid, devcount, error, Max_gridsize; double *h_A, *h_B, *d_A, *d_B; /* find number of device in current "context" */ cudaGetDevice(&devid); /* find how many devices are available */ if (cudaGetDeviceCount(&devcount) || devcount==0) { printf ("No CUDA devices!\n"); exit (1); } else { cudaDeviceProp deviceProp; cudaGetDeviceProperties (&deviceProp, devid); printf ("Device count, devid: %d %d\n", devcount, devid); printf ("Device: %s\n", deviceProp.name); printf("[deviceProp.major.deviceProp.minor] = [%d.%d]\n\n", deviceProp.major, deviceProp.minor); Max_gridsize = deviceProp.maxGridSize[0]; } // Loop to run the timing test multiple times: int kk; for (kk=0; kk<NTESTS; kk++) { // Using cudaMallocHost (intead of malloc) to accelerate data copying: // Initial data array on host: if (error = cudaMallocHost (&h_A, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } // Results array on host: if (error = cudaMallocHost (&h_B, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } // ALlocating arrays on GPU: if (error = cudaMalloc (&d_A, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } if (error = cudaMalloc (&d_B, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } // Initializing the input array: for (int i=0; i<NMAX; i++) h_A[i] = (double)rand()/(double)RAND_MAX; // Number of blocks of threads: int Nblocks = (NMAX+BLOCK_SIZE-1) / BLOCK_SIZE; if (Nblocks > Max_gridsize) { printf ("Nblocks > Max_gridsize! %d %d\n", Nblocks, Max_gridsize); exit (1); } if (error = cudaDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr0, NULL); //-------------------------------------------------------------------------------- // Copying the data to device (we time it): if (error = cudaMemcpy (d_A, h_A, NMAX*sizeof(double), cudaMemcpyHostToDevice)) { printf ("Error %d\n", error); exit (error); } // Intermediate timing, to measure timings separately for copying and kernel execution // (Should be removed in the solution code) if (error = cudaDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr01, NULL); // The kernel call: MyKernel <<<Nblocks, BLOCK_SIZE>>> (d_A, d_B); //-------------------------------------------------------------------------------- if (error = cudaDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr1, NULL); tdr = tdr0; timeval_subtract (&restime, &tdr1, &tdr); // Copying the result back to host (we don't time it): if (error = cudaMemcpy (h_B, d_B, NMAX*sizeof(double), cudaMemcpyDeviceToHost)) { printf ("Error %d\n", error); exit (error); } if (error = cudaDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } // Adding up the results, for accuracy/correctness testing: double result = 0.0; for (int i=0; i<NMAX; i++) { result += h_B[i]; } tdr = tdr0; timeval_subtract (&restime0, &tdr01, &tdr); tdr = tdr01; timeval_subtract (&restime1, &tdr1, &tdr); printf ("Individual timings: %e %e\n", restime0, restime1); printf ("Result: %e\n\n", result); printf ("Time: %e\n", restime); cudaFreeHost (h_A); cudaFreeHost (h_B); cudaFree (d_A); cudaFree (d_B); } // kk loop return 0; }
.file "tmpxft_000f5ee8_00000000-6_staged.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2074: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2074: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z16timeval_subtractPdP7timevalS1_ .type _Z16timeval_subtractPdP7timevalS1_, @function _Z16timeval_subtractPdP7timevalS1_: .LFB2070: .cfi_startproc endbr64 movq %rdx, %rcx movq 8(%rsi), %rax movq 8(%rdx), %r8 cmpq %r8, %rax jge .L4 movq %r8, %r9 subq %rax, %r9 movabsq $4835703278458516699, %rdx movq %r9, %rax imulq %rdx sarq $18, %rdx sarq $63, %r9 subq %r9, %rdx addl $1, %edx imull $1000000, %edx, %eax cltq subq %rax, %r8 movq %r8, 8(%rcx) movslq %edx, %rdx addq %rdx, (%rcx) .L4: movq 8(%rsi), %rax movq 8(%rcx), %r8 movq %rax, %rdx subq %r8, %rdx cmpq $1000000, %rdx jle .L5 movq %r8, %r9 subq %rax, %r9 movabsq $4835703278458516699, %rdx movq %r9, %rax imulq %rdx sarq $18, %rdx sarq $63, %r9 subq %r9, %rdx imull $1000000, %edx, %eax cltq addq %r8, %rax movq %rax, 8(%rcx) movslq %edx, %rdx subq %rdx, (%rcx) .L5: movq 8(%rsi), %rax subq 8(%rcx), %rax pxor %xmm0, %xmm0 cvtsi2sdq %rax, %xmm0 divsd .LC0(%rip), %xmm0 movq (%rsi), %rax subq (%rcx), %rax pxor %xmm1, %xmm1 cvtsi2sdq %rax, %xmm1 addsd %xmm1, %xmm0 movsd %xmm0, (%rdi) movq (%rcx), %rax cmpq %rax, (%rsi) setl %al movzbl %al, %eax ret .cfi_endproc .LFE2070: .size _Z16timeval_subtractPdP7timevalS1_, .-_Z16timeval_subtractPdP7timevalS1_ .globl _Z29__device_stub__Z8MyKernelPdS_PdS_ .type _Z29__device_stub__Z8MyKernelPdS_PdS_, @function _Z29__device_stub__Z8MyKernelPdS_PdS_: .LFB2096: .cfi_startproc endbr64 subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 8(%rsp) movq %rsi, (%rsp) movq %fs:40, %rax movq %rax, 104(%rsp) xorl %eax, %eax leaq 8(%rsp), %rax movq %rax, 80(%rsp) movq %rsp, %rax movq %rax, 88(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) leaq 24(%rsp), %rcx leaq 16(%rsp), %rdx leaq 44(%rsp), %rsi leaq 32(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L10 .L6: movq 104(%rsp), %rax subq %fs:40, %rax jne .L11 addq $120, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L10: .cfi_restore_state pushq 24(%rsp) .cfi_def_cfa_offset 136 pushq 24(%rsp) .cfi_def_cfa_offset 144 leaq 96(%rsp), %r9 movq 60(%rsp), %rcx movl 68(%rsp), %r8d movq 48(%rsp), %rsi movl 56(%rsp), %edx leaq _Z8MyKernelPdS_(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 128 jmp .L6 .L11: call __stack_chk_fail@PLT .cfi_endproc .LFE2096: .size _Z29__device_stub__Z8MyKernelPdS_PdS_, .-_Z29__device_stub__Z8MyKernelPdS_PdS_ .globl _Z8MyKernelPdS_ .type _Z8MyKernelPdS_, @function _Z8MyKernelPdS_: .LFB2097: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z8MyKernelPdS_PdS_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2097: .size _Z8MyKernelPdS_, .-_Z8MyKernelPdS_ .section .rodata.str1.1,"aMS",@progbits,1 .LC2: .string "Device count, devid: %d %d\n" .LC3: .string "Device: %s\n" .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC4: .string "[deviceProp.major.deviceProp.minor] = [%d.%d]\n\n" .section .rodata.str1.1 .LC5: .string "No CUDA devices!\n" .LC6: .string "Error %d\n" .section .rodata.str1.8 .align 8 .LC8: .string "Nblocks > Max_gridsize! %d %d\n" .section .rodata.str1.1 .LC9: .string "Individual timings: %e %e\n" .LC10: .string "Result: %e\n\n" .LC11: .string "Time: %e\n" .text .globl main .type main, @function main: .LFB2071: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $1224, %rsp .cfi_def_cfa_offset 1280 movq %fs:40, %rax movq %rax, 1208(%rsp) xorl %eax, %eax leaq 24(%rsp), %rdi call cudaGetDevice@PLT leaq 28(%rsp), %rdi call cudaGetDeviceCount@PLT testl %eax, %eax jne .L15 cmpl $0, 28(%rsp) je .L15 leaq 176(%rsp), %rbx movl 24(%rsp), %esi movq %rbx, %rdi call cudaGetDeviceProperties_v2@PLT movl 24(%rsp), %ecx movl 28(%rsp), %edx leaq .LC2(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq %rbx, %rdx leaq .LC3(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl 540(%rsp), %ecx movl 536(%rsp), %edx leaq .LC4(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl 512(%rsp), %r12d movl $100, %ebx leaq .LC9(%rip), %r14 leaq .LC10(%rip), %r13 leaq .LC11(%rip), %r15 jmp .L16 .L15: leaq .LC5(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl $1, %edi call exit@PLT .L40: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L41: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L42: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L43: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L44: movl %r12d, %ecx movl $7813, %edx leaq .LC8(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl $1, %edi call exit@PLT .L45: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L46: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L47: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L25: call cudaDeviceSynchronize@PLT movl %eax, %ebp testl %eax, %eax jne .L36 leaq 128(%rsp), %rbp movl $0, %esi movq %rbp, %rdi call gettimeofday@PLT movdqa 112(%rsp), %xmm3 movaps %xmm3, 144(%rsp) leaq 144(%rsp), %rdx leaq 32(%rsp), %rdi movq %rbp, %rsi call _Z16timeval_subtractPdP7timevalS1_ movl $2, %ecx movl $8000000, %edx movq 80(%rsp), %rsi movq 64(%rsp), %rdi call cudaMemcpy@PLT movl %eax, %ebp testl %eax, %eax jne .L37 call cudaDeviceSynchronize@PLT movl %eax, %ebp testl %eax, %eax jne .L38 movq 64(%rsp), %rax leaq 8000000(%rax), %rdx movq $0x000000000, 8(%rsp) .L29: movsd 8(%rsp), %xmm2 addsd (%rax), %xmm2 movsd %xmm2, 8(%rsp) addq $8, %rax cmpq %rdx, %rax jne .L29 movdqa 112(%rsp), %xmm4 movaps %xmm4, 144(%rsp) leaq 144(%rsp), %rbp leaq 160(%rsp), %rsi leaq 40(%rsp), %rdi movq %rbp, %rdx call _Z16timeval_subtractPdP7timevalS1_ movdqa 160(%rsp), %xmm5 movaps %xmm5, 144(%rsp) leaq 128(%rsp), %rsi leaq 48(%rsp), %rdi movq %rbp, %rdx call _Z16timeval_subtractPdP7timevalS1_ movsd 48(%rsp), %xmm1 movsd 40(%rsp), %xmm0 movq %r14, %rsi movl $2, %edi movl $2, %eax call __printf_chk@PLT movsd 8(%rsp), %xmm0 movq %r13, %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movsd 32(%rsp), %xmm0 movq %r15, %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movq 56(%rsp), %rdi call cudaFreeHost@PLT movq 64(%rsp), %rdi call cudaFreeHost@PLT movq 72(%rsp), %rdi call cudaFree@PLT movq 80(%rsp), %rdi call cudaFree@PLT subl $1, %ebx je .L39 .L16: leaq 56(%rsp), %rdi movl $0, %edx movl $8000000, %esi call cudaHostAlloc@PLT movl %eax, %ebp testl %eax, %eax jne .L40 leaq 64(%rsp), %rdi movl $0, %edx movl $8000000, %esi call cudaHostAlloc@PLT movl %eax, %ebp testl %eax, %eax jne .L41 leaq 72(%rsp), %rdi movl $8000000, %esi call cudaMalloc@PLT movl %eax, %ebp testl %eax, %eax jne .L42 leaq 80(%rsp), %rdi movl $8000000, %esi call cudaMalloc@PLT movl %eax, %ebp testl %eax, %eax jne .L43 movl $0, %ebp .L20: call rand@PLT pxor %xmm0, %xmm0 cvtsi2sdl %eax, %xmm0 divsd .LC7(%rip), %xmm0 movq 56(%rsp), %rax movsd %xmm0, (%rax,%rbp) addq $8, %rbp cmpq $8000000, %rbp jne .L20 cmpl $7812, %r12d jle .L44 call cudaDeviceSynchronize@PLT movl %eax, %ebp testl %eax, %eax jne .L45 leaq 112(%rsp), %rdi movl $0, %esi call gettimeofday@PLT movl $1, %ecx movl $8000000, %edx movq 56(%rsp), %rsi movq 72(%rsp), %rdi call cudaMemcpy@PLT movl %eax, %ebp testl %eax, %eax jne .L46 call cudaDeviceSynchronize@PLT movl %eax, %ebp testl %eax, %eax jne .L47 leaq 160(%rsp), %rdi movl $0, %esi call gettimeofday@PLT movl $128, 100(%rsp) movl $1, 104(%rsp) movl $1, 108(%rsp) movl $7813, 88(%rsp) movl $1, 92(%rsp) movl $1, 96(%rsp) movl $0, %r9d movl $0, %r8d movq 100(%rsp), %rdx movl $1, %ecx movq 88(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax jne .L25 movq 80(%rsp), %rsi movq 72(%rsp), %rdi call _Z29__device_stub__Z8MyKernelPdS_PdS_ jmp .L25 .L36: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L37: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L38: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L39: movq 1208(%rsp), %rax subq %fs:40, %rax jne .L48 movl $0, %eax addq $1224, %rsp .cfi_remember_state .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .L48: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE2071: .size main, .-main .section .rodata.str1.1 .LC12: .string "_Z8MyKernelPdS_" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2099: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC12(%rip), %rdx movq %rdx, %rcx leaq _Z8MyKernelPdS_(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2099: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst8,"aM",@progbits,8 .align 8 .LC0: .long 0 .long 1093567616 .align 8 .LC7: .long -4194304 .long 1105199103 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
/* Small CUDA exercise to try to improve efficiency by using two separate streams to set up a staged copying and execution (when instead of one large copy, followed by one large kernel computation, one does it in many small chunks, with copying and computing done in parallel in two streams, after first chunk was copied to the device). For the purpose of this exercise, ignore the copying of the results from the device to host at the end; only do the staged copy and execute for the copying of the initial data to the device + the kernel. We use cudaMallocHost to allocate arrays on host in pinned memory, which is both results in faster copying to/from GPU (compared to malloc), and also a CUDA requirement for copying running concurrently with a kernel. Make sure that the "Result:" value printed by the code is (almost) identical in both original and modified versions of the code. If not, you have a bug! Hints: You will have to use the following CUDA functions: - cudaStreamCreate - cudaMemcpyAsync - cudaStreamDestroy - cudaDeviceSynchronize * You will have to set up a for loop for multiple chunks copying and kernel execution; * Number of chunks should be a variable (or macro parameter); for simplicity, make NMAX dividable by the number of chunks; * In cudaMemcpyAsync the first two arguments should be "&d_A[ind], &h_A[ind]", not "d_A, h_A", ind being the starting index for the current chunk to copy; * You'll have to pass two more arguments to the kernel - ind and number of threads per chunk; * Nblocks will be different - shoul be computed per chunk. * You'll have to modify the kernel slightly; At the end, you should get the timings (based on 10 runs, NMAX=1000000, BLOCK_SIZE=128) similar to this: NCHUNKS t, ms - 2.76 - the original (non-staged) version of the code 1 2.72 - result is similar to the non-staged code 2 2.08 - even with only 2 chunks, we already see 33% speedup 4 1.79 5 1.75 10 1.72 - seems to be the best timing, 60% faster than the original code 20 1.87 - as NCHUNKS increases, the results get worse. Why? 100 3.53 - for too many NCHUNKS results can get even worse than in non-staged code To compile: nvcc -arch=sm_20 -O2 staged.cu -o staged The best/average timings: ../best_time.sh ./staged */ #include <sys/time.h> #include <ctype.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <time.h> #include <unistd.h> // Number of times to run the test (for better timings accuracy): #define NTESTS 100 // Number of threads in one block (possible range is 32...1024): #define BLOCK_SIZE 128 // Total number of threads (total number of elements to process in the kernel): #define NMAX 1000000 /* Subtract the `struct timeval' values X and Y, storing the result in RESULT. Return 1 if the difference is negative, otherwise 0. */ // It messes up with y! int timeval_subtract (double *result, struct timeval *x, struct timeval *y) { struct timeval result0; /* Perform the carry for the later subtraction by updating y. */ if (x->tv_usec < y->tv_usec) { int nsec = (y->tv_usec - x->tv_usec) / 1000000 + 1; y->tv_usec -= 1000000 * nsec; y->tv_sec += nsec; } if (x->tv_usec - y->tv_usec > 1000000) { int nsec = (y->tv_usec - x->tv_usec) / 1000000; y->tv_usec += 1000000 * nsec; y->tv_sec -= nsec; } /* Compute the time remaining to wait. tv_usec is certainly positive. */ result0.tv_sec = x->tv_sec - y->tv_sec; result0.tv_usec = x->tv_usec - y->tv_usec; *result = ((double)result0.tv_usec)/1e6 + (double)result0.tv_sec; /* Return 1 if result is negative. */ return x->tv_sec < y->tv_sec; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ // The kernel: __global__ void MyKernel (double *d_A, double *d_B) { double x, y, z; int i = threadIdx.x + blockDim.x * blockIdx.x; if (i >= NMAX) return; // Some meaningless cpu-intensive computation: x = pow(d_A[i], 2.71); y = pow(d_A[i], 0.35); z = 2*x + 5*y; d_B[i] = x + y + z + x*y + x/y + y/z; return; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ int main (int argc,char **argv) { struct timeval tdr0, tdr1, tdr, tdr01; double restime, restime0, restime1; int devid, devcount, error, Max_gridsize; double *h_A, *h_B, *d_A, *d_B; /* find number of device in current "context" */ cudaGetDevice(&devid); /* find how many devices are available */ if (cudaGetDeviceCount(&devcount) || devcount==0) { printf ("No CUDA devices!\n"); exit (1); } else { cudaDeviceProp deviceProp; cudaGetDeviceProperties (&deviceProp, devid); printf ("Device count, devid: %d %d\n", devcount, devid); printf ("Device: %s\n", deviceProp.name); printf("[deviceProp.major.deviceProp.minor] = [%d.%d]\n\n", deviceProp.major, deviceProp.minor); Max_gridsize = deviceProp.maxGridSize[0]; } // Loop to run the timing test multiple times: int kk; for (kk=0; kk<NTESTS; kk++) { // Using cudaMallocHost (intead of malloc) to accelerate data copying: // Initial data array on host: if (error = cudaMallocHost (&h_A, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } // Results array on host: if (error = cudaMallocHost (&h_B, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } // ALlocating arrays on GPU: if (error = cudaMalloc (&d_A, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } if (error = cudaMalloc (&d_B, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } // Initializing the input array: for (int i=0; i<NMAX; i++) h_A[i] = (double)rand()/(double)RAND_MAX; // Number of blocks of threads: int Nblocks = (NMAX+BLOCK_SIZE-1) / BLOCK_SIZE; if (Nblocks > Max_gridsize) { printf ("Nblocks > Max_gridsize! %d %d\n", Nblocks, Max_gridsize); exit (1); } if (error = cudaDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr0, NULL); //-------------------------------------------------------------------------------- // Copying the data to device (we time it): if (error = cudaMemcpy (d_A, h_A, NMAX*sizeof(double), cudaMemcpyHostToDevice)) { printf ("Error %d\n", error); exit (error); } // Intermediate timing, to measure timings separately for copying and kernel execution // (Should be removed in the solution code) if (error = cudaDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr01, NULL); // The kernel call: MyKernel <<<Nblocks, BLOCK_SIZE>>> (d_A, d_B); //-------------------------------------------------------------------------------- if (error = cudaDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr1, NULL); tdr = tdr0; timeval_subtract (&restime, &tdr1, &tdr); // Copying the result back to host (we don't time it): if (error = cudaMemcpy (h_B, d_B, NMAX*sizeof(double), cudaMemcpyDeviceToHost)) { printf ("Error %d\n", error); exit (error); } if (error = cudaDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } // Adding up the results, for accuracy/correctness testing: double result = 0.0; for (int i=0; i<NMAX; i++) { result += h_B[i]; } tdr = tdr0; timeval_subtract (&restime0, &tdr01, &tdr); tdr = tdr01; timeval_subtract (&restime1, &tdr1, &tdr); printf ("Individual timings: %e %e\n", restime0, restime1); printf ("Result: %e\n\n", result); printf ("Time: %e\n", restime); cudaFreeHost (h_A); cudaFreeHost (h_B); cudaFree (d_A); cudaFree (d_B); } // kk loop return 0; }
/* Small CUDA exercise to try to improve efficiency by using two separate streams to set up a staged copying and execution (when instead of one large copy, followed by one large kernel computation, one does it in many small chunks, with copying and computing done in parallel in two streams, after first chunk was copied to the device). For the purpose of this exercise, ignore the copying of the results from the device to host at the end; only do the staged copy and execute for the copying of the initial data to the device + the kernel. We use cudaMallocHost to allocate arrays on host in pinned memory, which is both results in faster copying to/from GPU (compared to malloc), and also a CUDA requirement for copying running concurrently with a kernel. Make sure that the "Result:" value printed by the code is (almost) identical in both original and modified versions of the code. If not, you have a bug! Hints: You will have to use the following CUDA functions: - cudaStreamCreate - cudaMemcpyAsync - cudaStreamDestroy - cudaDeviceSynchronize * You will have to set up a for loop for multiple chunks copying and kernel execution; * Number of chunks should be a variable (or macro parameter); for simplicity, make NMAX dividable by the number of chunks; * In cudaMemcpyAsync the first two arguments should be "&d_A[ind], &h_A[ind]", not "d_A, h_A", ind being the starting index for the current chunk to copy; * You'll have to pass two more arguments to the kernel - ind and number of threads per chunk; * Nblocks will be different - shoul be computed per chunk. * You'll have to modify the kernel slightly; At the end, you should get the timings (based on 10 runs, NMAX=1000000, BLOCK_SIZE=128) similar to this: NCHUNKS t, ms - 2.76 - the original (non-staged) version of the code 1 2.72 - result is similar to the non-staged code 2 2.08 - even with only 2 chunks, we already see 33% speedup 4 1.79 5 1.75 10 1.72 - seems to be the best timing, 60% faster than the original code 20 1.87 - as NCHUNKS increases, the results get worse. Why? 100 3.53 - for too many NCHUNKS results can get even worse than in non-staged code To compile: nvcc -arch=sm_20 -O2 staged.cu -o staged The best/average timings: ../best_time.sh ./staged */ #include <hip/hip_runtime.h> #include <sys/time.h> #include <ctype.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <time.h> #include <unistd.h> // Number of times to run the test (for better timings accuracy): #define NTESTS 100 // Number of threads in one block (possible range is 32...1024): #define BLOCK_SIZE 128 // Total number of threads (total number of elements to process in the kernel): #define NMAX 1000000 /* Subtract the `struct timeval' values X and Y, storing the result in RESULT. Return 1 if the difference is negative, otherwise 0. */ // It messes up with y! int timeval_subtract (double *result, struct timeval *x, struct timeval *y) { struct timeval result0; /* Perform the carry for the later subtraction by updating y. */ if (x->tv_usec < y->tv_usec) { int nsec = (y->tv_usec - x->tv_usec) / 1000000 + 1; y->tv_usec -= 1000000 * nsec; y->tv_sec += nsec; } if (x->tv_usec - y->tv_usec > 1000000) { int nsec = (y->tv_usec - x->tv_usec) / 1000000; y->tv_usec += 1000000 * nsec; y->tv_sec -= nsec; } /* Compute the time remaining to wait. tv_usec is certainly positive. */ result0.tv_sec = x->tv_sec - y->tv_sec; result0.tv_usec = x->tv_usec - y->tv_usec; *result = ((double)result0.tv_usec)/1e6 + (double)result0.tv_sec; /* Return 1 if result is negative. */ return x->tv_sec < y->tv_sec; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ // The kernel: __global__ void MyKernel (double *d_A, double *d_B) { double x, y, z; int i = threadIdx.x + blockDim.x * blockIdx.x; if (i >= NMAX) return; // Some meaningless cpu-intensive computation: x = pow(d_A[i], 2.71); y = pow(d_A[i], 0.35); z = 2*x + 5*y; d_B[i] = x + y + z + x*y + x/y + y/z; return; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ int main (int argc,char **argv) { struct timeval tdr0, tdr1, tdr, tdr01; double restime, restime0, restime1; int devid, devcount, error, Max_gridsize; double *h_A, *h_B, *d_A, *d_B; /* find number of device in current "context" */ hipGetDevice(&devid); /* find how many devices are available */ if (hipGetDeviceCount(&devcount) || devcount==0) { printf ("No CUDA devices!\n"); exit (1); } else { hipDeviceProp_t deviceProp; hipGetDeviceProperties (&deviceProp, devid); printf ("Device count, devid: %d %d\n", devcount, devid); printf ("Device: %s\n", deviceProp.name); printf("[deviceProp.major.deviceProp.minor] = [%d.%d]\n\n", deviceProp.major, deviceProp.minor); Max_gridsize = deviceProp.maxGridSize[0]; } // Loop to run the timing test multiple times: int kk; for (kk=0; kk<NTESTS; kk++) { // Using cudaMallocHost (intead of malloc) to accelerate data copying: // Initial data array on host: if (error = hipHostMalloc (&h_A, NMAX*sizeof(double), hipHostMallocDefault)) { printf ("Error %d\n", error); exit (error); } // Results array on host: if (error = hipHostMalloc (&h_B, NMAX*sizeof(double), hipHostMallocDefault)) { printf ("Error %d\n", error); exit (error); } // ALlocating arrays on GPU: if (error = hipMalloc (&d_A, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } if (error = hipMalloc (&d_B, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } // Initializing the input array: for (int i=0; i<NMAX; i++) h_A[i] = (double)rand()/(double)RAND_MAX; // Number of blocks of threads: int Nblocks = (NMAX+BLOCK_SIZE-1) / BLOCK_SIZE; if (Nblocks > Max_gridsize) { printf ("Nblocks > Max_gridsize! %d %d\n", Nblocks, Max_gridsize); exit (1); } if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr0, NULL); //-------------------------------------------------------------------------------- // Copying the data to device (we time it): if (error = hipMemcpy (d_A, h_A, NMAX*sizeof(double), hipMemcpyHostToDevice)) { printf ("Error %d\n", error); exit (error); } // Intermediate timing, to measure timings separately for copying and kernel execution // (Should be removed in the solution code) if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr01, NULL); // The kernel call: MyKernel <<<Nblocks, BLOCK_SIZE>>> (d_A, d_B); //-------------------------------------------------------------------------------- if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr1, NULL); tdr = tdr0; timeval_subtract (&restime, &tdr1, &tdr); // Copying the result back to host (we don't time it): if (error = hipMemcpy (h_B, d_B, NMAX*sizeof(double), hipMemcpyDeviceToHost)) { printf ("Error %d\n", error); exit (error); } if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } // Adding up the results, for accuracy/correctness testing: double result = 0.0; for (int i=0; i<NMAX; i++) { result += h_B[i]; } tdr = tdr0; timeval_subtract (&restime0, &tdr01, &tdr); tdr = tdr01; timeval_subtract (&restime1, &tdr1, &tdr); printf ("Individual timings: %e %e\n", restime0, restime1); printf ("Result: %e\n\n", result); printf ("Time: %e\n", restime); hipHostFree (h_A); hipHostFree (h_B); hipFree (d_A); hipFree (d_B); } // kk loop return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
/* Small CUDA exercise to try to improve efficiency by using two separate streams to set up a staged copying and execution (when instead of one large copy, followed by one large kernel computation, one does it in many small chunks, with copying and computing done in parallel in two streams, after first chunk was copied to the device). For the purpose of this exercise, ignore the copying of the results from the device to host at the end; only do the staged copy and execute for the copying of the initial data to the device + the kernel. We use cudaMallocHost to allocate arrays on host in pinned memory, which is both results in faster copying to/from GPU (compared to malloc), and also a CUDA requirement for copying running concurrently with a kernel. Make sure that the "Result:" value printed by the code is (almost) identical in both original and modified versions of the code. If not, you have a bug! Hints: You will have to use the following CUDA functions: - cudaStreamCreate - cudaMemcpyAsync - cudaStreamDestroy - cudaDeviceSynchronize * You will have to set up a for loop for multiple chunks copying and kernel execution; * Number of chunks should be a variable (or macro parameter); for simplicity, make NMAX dividable by the number of chunks; * In cudaMemcpyAsync the first two arguments should be "&d_A[ind], &h_A[ind]", not "d_A, h_A", ind being the starting index for the current chunk to copy; * You'll have to pass two more arguments to the kernel - ind and number of threads per chunk; * Nblocks will be different - shoul be computed per chunk. * You'll have to modify the kernel slightly; At the end, you should get the timings (based on 10 runs, NMAX=1000000, BLOCK_SIZE=128) similar to this: NCHUNKS t, ms - 2.76 - the original (non-staged) version of the code 1 2.72 - result is similar to the non-staged code 2 2.08 - even with only 2 chunks, we already see 33% speedup 4 1.79 5 1.75 10 1.72 - seems to be the best timing, 60% faster than the original code 20 1.87 - as NCHUNKS increases, the results get worse. Why? 100 3.53 - for too many NCHUNKS results can get even worse than in non-staged code To compile: nvcc -arch=sm_20 -O2 staged.cu -o staged The best/average timings: ../best_time.sh ./staged */ #include <hip/hip_runtime.h> #include <sys/time.h> #include <ctype.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <time.h> #include <unistd.h> // Number of times to run the test (for better timings accuracy): #define NTESTS 100 // Number of threads in one block (possible range is 32...1024): #define BLOCK_SIZE 128 // Total number of threads (total number of elements to process in the kernel): #define NMAX 1000000 /* Subtract the `struct timeval' values X and Y, storing the result in RESULT. Return 1 if the difference is negative, otherwise 0. */ // It messes up with y! int timeval_subtract (double *result, struct timeval *x, struct timeval *y) { struct timeval result0; /* Perform the carry for the later subtraction by updating y. */ if (x->tv_usec < y->tv_usec) { int nsec = (y->tv_usec - x->tv_usec) / 1000000 + 1; y->tv_usec -= 1000000 * nsec; y->tv_sec += nsec; } if (x->tv_usec - y->tv_usec > 1000000) { int nsec = (y->tv_usec - x->tv_usec) / 1000000; y->tv_usec += 1000000 * nsec; y->tv_sec -= nsec; } /* Compute the time remaining to wait. tv_usec is certainly positive. */ result0.tv_sec = x->tv_sec - y->tv_sec; result0.tv_usec = x->tv_usec - y->tv_usec; *result = ((double)result0.tv_usec)/1e6 + (double)result0.tv_sec; /* Return 1 if result is negative. */ return x->tv_sec < y->tv_sec; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ // The kernel: __global__ void MyKernel (double *d_A, double *d_B) { double x, y, z; int i = threadIdx.x + blockDim.x * blockIdx.x; if (i >= NMAX) return; // Some meaningless cpu-intensive computation: x = pow(d_A[i], 2.71); y = pow(d_A[i], 0.35); z = 2*x + 5*y; d_B[i] = x + y + z + x*y + x/y + y/z; return; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ int main (int argc,char **argv) { struct timeval tdr0, tdr1, tdr, tdr01; double restime, restime0, restime1; int devid, devcount, error, Max_gridsize; double *h_A, *h_B, *d_A, *d_B; /* find number of device in current "context" */ hipGetDevice(&devid); /* find how many devices are available */ if (hipGetDeviceCount(&devcount) || devcount==0) { printf ("No CUDA devices!\n"); exit (1); } else { hipDeviceProp_t deviceProp; hipGetDeviceProperties (&deviceProp, devid); printf ("Device count, devid: %d %d\n", devcount, devid); printf ("Device: %s\n", deviceProp.name); printf("[deviceProp.major.deviceProp.minor] = [%d.%d]\n\n", deviceProp.major, deviceProp.minor); Max_gridsize = deviceProp.maxGridSize[0]; } // Loop to run the timing test multiple times: int kk; for (kk=0; kk<NTESTS; kk++) { // Using cudaMallocHost (intead of malloc) to accelerate data copying: // Initial data array on host: if (error = hipHostMalloc (&h_A, NMAX*sizeof(double), hipHostMallocDefault)) { printf ("Error %d\n", error); exit (error); } // Results array on host: if (error = hipHostMalloc (&h_B, NMAX*sizeof(double), hipHostMallocDefault)) { printf ("Error %d\n", error); exit (error); } // ALlocating arrays on GPU: if (error = hipMalloc (&d_A, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } if (error = hipMalloc (&d_B, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } // Initializing the input array: for (int i=0; i<NMAX; i++) h_A[i] = (double)rand()/(double)RAND_MAX; // Number of blocks of threads: int Nblocks = (NMAX+BLOCK_SIZE-1) / BLOCK_SIZE; if (Nblocks > Max_gridsize) { printf ("Nblocks > Max_gridsize! %d %d\n", Nblocks, Max_gridsize); exit (1); } if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr0, NULL); //-------------------------------------------------------------------------------- // Copying the data to device (we time it): if (error = hipMemcpy (d_A, h_A, NMAX*sizeof(double), hipMemcpyHostToDevice)) { printf ("Error %d\n", error); exit (error); } // Intermediate timing, to measure timings separately for copying and kernel execution // (Should be removed in the solution code) if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr01, NULL); // The kernel call: MyKernel <<<Nblocks, BLOCK_SIZE>>> (d_A, d_B); //-------------------------------------------------------------------------------- if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr1, NULL); tdr = tdr0; timeval_subtract (&restime, &tdr1, &tdr); // Copying the result back to host (we don't time it): if (error = hipMemcpy (h_B, d_B, NMAX*sizeof(double), hipMemcpyDeviceToHost)) { printf ("Error %d\n", error); exit (error); } if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } // Adding up the results, for accuracy/correctness testing: double result = 0.0; for (int i=0; i<NMAX; i++) { result += h_B[i]; } tdr = tdr0; timeval_subtract (&restime0, &tdr01, &tdr); tdr = tdr01; timeval_subtract (&restime1, &tdr1, &tdr); printf ("Individual timings: %e %e\n", restime0, restime1); printf ("Result: %e\n\n", result); printf ("Time: %e\n", restime); hipHostFree (h_A); hipHostFree (h_B); hipFree (d_A); hipFree (d_B); } // kk loop return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z8MyKernelPdS_ .globl _Z8MyKernelPdS_ .p2align 8 .type _Z8MyKernelPdS_,@function _Z8MyKernelPdS_: s_load_b32 s2, s[0:1], 0x1c s_waitcnt lgkmcnt(0) s_and_b32 s2, s2, 0xffff s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_mad_u64_u32 v[1:2], null, s15, s2, v[0:1] s_mov_b32 s2, exec_lo v_cmpx_gt_i32_e32 0xf4240, v1 s_cbranch_execz .LBB0_2 s_load_b128 s[4:7], s[0:1], 0x0 v_ashrrev_i32_e32 v2, 31, v1 s_mov_b32 s1, 0x3fe55555 s_mov_b32 s0, 0x55555555 s_mov_b32 s3, 0x3fba6564 s_mov_b32 s2, 0x968915a9 v_lshlrev_b64 v[0:1], 3, v[1:2] s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v2, vcc_lo, s4, v0 v_add_co_ci_u32_e32 v3, vcc_lo, s5, v1, vcc_lo s_mov_b32 s5, 0x3fbdee67 s_mov_b32 s4, 0x4222de17 global_load_b64 v[2:3], v[2:3], off s_waitcnt vmcnt(0) v_frexp_mant_f64_e64 v[4:5], |v[2:3]| s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_cmp_gt_f64_e32 vcc_lo, s[0:1], v[4:5] v_cndmask_b32_e64 v6, 0, 1, vcc_lo v_ldexp_f64 v[4:5], v[4:5], v6 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_add_f64 v[6:7], v[4:5], 1.0 v_add_f64 v[12:13], v[4:5], -1.0 v_rcp_f64_e32 v[8:9], v[6:7] v_add_f64 v[14:15], v[6:7], -1.0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_add_f64 v[4:5], v[4:5], -v[14:15] s_waitcnt_depctr 0xfff v_fma_f64 v[10:11], -v[6:7], v[8:9], 1.0 v_fma_f64 v[8:9], v[10:11], v[8:9], v[8:9] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fma_f64 v[10:11], -v[6:7], v[8:9], 1.0 v_fma_f64 v[8:9], v[10:11], v[8:9], v[8:9] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f64 v[10:11], v[12:13], v[8:9] v_mul_f64 v[16:17], v[6:7], v[10:11] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fma_f64 v[6:7], v[10:11], v[6:7], -v[16:17] v_fma_f64 v[4:5], v[10:11], v[4:5], v[6:7] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_f64 v[6:7], v[16:17], v[4:5] v_add_f64 v[14:15], v[12:13], -v[6:7] v_add_f64 v[16:17], v[6:7], -v[16:17] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_f64 v[12:13], v[12:13], -v[14:15] v_add_f64 v[4:5], v[16:17], -v[4:5] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_f64 v[6:7], v[12:13], -v[6:7] v_add_f64 v[4:5], v[4:5], v[6:7] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_f64 v[4:5], v[14:15], v[4:5] v_mul_f64 v[4:5], v[8:9], v[4:5] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_f64 v[6:7], v[10:11], v[4:5] v_add_f64 v[8:9], v[6:7], -v[10:11] v_mul_f64 v[10:11], v[6:7], v[6:7] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_f64 v[4:5], v[4:5], -v[8:9] v_fma_f64 v[8:9], v[6:7], v[6:7], -v[10:11] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_f64 v[12:13], v[4:5], v[4:5] v_fma_f64 v[8:9], v[6:7], v[12:13], v[8:9] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_f64 v[12:13], v[10:11], v[8:9] v_fma_f64 v[14:15], v[12:13], s[4:5], s[2:3] s_mov_b32 s3, 0x3fbe25e4 s_mov_b32 s2, 0x3abe935a v_add_f64 v[10:11], v[12:13], -v[10:11] v_mul_f64 v[20:21], v[6:7], v[12:13] s_mov_b32 s5, 0x3ff71547 s_mov_b32 s4, 0x652b82fe s_delay_alu instid0(VALU_DEP_3) | instskip(SKIP_2) | instid1(VALU_DEP_3) v_fma_f64 v[14:15], v[12:13], v[14:15], s[2:3] s_mov_b32 s3, 0x3fc110ef s_mov_b32 s2, 0x47e6c9c2 v_add_f64 v[8:9], v[8:9], -v[10:11] s_delay_alu instid0(VALU_DEP_2) v_fma_f64 v[14:15], v[12:13], v[14:15], s[2:3] s_mov_b32 s3, 0x3fc3b13b s_mov_b32 s2, 0xcfa74449 s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) v_fma_f64 v[14:15], v[12:13], v[14:15], s[2:3] s_mov_b32 s3, 0x3fc745d1 s_mov_b32 s2, 0x71bf3c30 s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) v_fma_f64 v[14:15], v[12:13], v[14:15], s[2:3] s_mov_b32 s3, 0x3fcc71c7 s_mov_b32 s2, 0x1c7792ce s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) v_fma_f64 v[14:15], v[12:13], v[14:15], s[2:3] s_mov_b32 s3, 0x3fd24924 s_mov_b32 s2, 0x924920da s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) v_fma_f64 v[14:15], v[12:13], v[14:15], s[2:3] s_mov_b32 s3, 0x3fd99999 s_mov_b32 s2, 0x9999999c s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) v_fma_f64 v[14:15], v[12:13], v[14:15], s[2:3] s_mov_b32 s3, 0x3c7abc9e s_mov_b32 s2, 0x3b39803f s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f64 v[16:17], v[12:13], v[14:15] v_fma_f64 v[10:11], v[12:13], v[14:15], -v[16:17] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fma_f64 v[10:11], v[8:9], v[14:15], v[10:11] v_add_f64 v[14:15], v[16:17], v[10:11] s_delay_alu instid0(VALU_DEP_1) v_add_f64 v[18:19], v[14:15], s[0:1] v_add_f64 v[16:17], v[14:15], -v[16:17] s_mov_b32 s1, 0xbfe55555 s_delay_alu instid0(VALU_DEP_2) | instid1(SALU_CYCLE_1) v_add_f64 v[22:23], v[18:19], s[0:1] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_3) | instid1(VALU_DEP_3) v_add_f64 v[10:11], v[10:11], -v[16:17] v_fma_f64 v[16:17], v[12:13], v[6:7], -v[20:21] s_mov_b32 s1, 0x3c8543b0 s_mov_b32 s0, 0xd5df274d v_add_f64 v[14:15], v[14:15], -v[22:23] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_f64 v[10:11], v[10:11], s[0:1] v_fma_f64 v[12:13], v[12:13], v[4:5], v[16:17] s_mov_b32 s1, 0x3fe62e42 s_mov_b32 s0, 0xfefa39ef v_ldexp_f64 v[4:5], v[4:5], 1 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_f64 v[10:11], v[10:11], v[14:15] v_fma_f64 v[8:9], v[8:9], v[6:7], v[12:13] v_ldexp_f64 v[6:7], v[6:7], 1 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_f64 v[12:13], v[18:19], v[10:11] v_add_f64 v[14:15], v[20:21], v[8:9] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_f64 v[16:17], v[18:19], -v[12:13] v_mul_f64 v[18:19], v[14:15], v[12:13] v_add_f64 v[20:21], v[14:15], -v[20:21] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_f64 v[10:11], v[10:11], v[16:17] v_fma_f64 v[16:17], v[14:15], v[12:13], -v[18:19] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_f64 v[8:9], v[8:9], -v[20:21] v_fma_f64 v[10:11], v[14:15], v[10:11], v[16:17] s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_fma_f64 v[8:9], v[8:9], v[12:13], v[10:11] v_frexp_exp_i32_f64_e32 v12, v[2:3] v_add_f64 v[10:11], v[18:19], v[8:9] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_subrev_co_ci_u32_e32 v12, vcc_lo, 0, v12, vcc_lo v_cmp_eq_f64_e32 vcc_lo, 1.0, v[2:3] v_cvt_f64_i32_e32 v[12:13], v12 s_delay_alu instid0(VALU_DEP_4) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_f64 v[14:15], v[6:7], v[10:11] v_add_f64 v[16:17], v[10:11], -v[18:19] v_mul_f64 v[18:19], v[12:13], s[0:1] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_f64 v[6:7], v[14:15], -v[6:7] v_add_f64 v[8:9], v[8:9], -v[16:17] s_delay_alu instid0(VALU_DEP_3) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_fma_f64 v[16:17], v[12:13], s[0:1], -v[18:19] s_mov_b32 s1, 0xbfe62e42 v_add_f64 v[6:7], v[10:11], -v[6:7] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_f64 v[4:5], v[4:5], v[8:9] v_fma_f64 v[8:9], v[12:13], s[2:3], v[16:17] s_mov_b32 s3, 0xbc7abc9e s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_f64 v[4:5], v[4:5], v[6:7] v_add_f64 v[6:7], v[18:19], v[8:9] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_f64 v[10:11], v[14:15], v[4:5] v_add_f64 v[18:19], v[6:7], -v[18:19] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_f64 v[12:13], v[6:7], v[10:11] v_add_f64 v[14:15], v[10:11], -v[14:15] v_add_f64 v[8:9], v[8:9], -v[18:19] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_f64 v[16:17], v[12:13], -v[6:7] v_add_f64 v[4:5], v[4:5], -v[14:15] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_f64 v[20:21], v[12:13], -v[16:17] v_add_f64 v[10:11], v[10:11], -v[16:17] v_add_f64 v[14:15], v[8:9], v[4:5] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_f64 v[6:7], v[6:7], -v[20:21] v_add_f64 v[6:7], v[10:11], v[6:7] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_f64 v[10:11], v[14:15], -v[8:9] v_add_f64 v[6:7], v[14:15], v[6:7] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_f64 v[14:15], v[14:15], -v[10:11] v_add_f64 v[4:5], v[4:5], -v[10:11] v_add_f64 v[16:17], v[12:13], v[6:7] s_delay_alu instid0(VALU_DEP_3) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_add_f64 v[8:9], v[8:9], -v[14:15] v_mov_b32_e32 v14, 0x3ff00000 v_cndmask_b32_e32 v15, 0x3fd66666, v14, vcc_lo s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4) v_add_f64 v[10:11], v[16:17], -v[12:13] v_add_f64 v[4:5], v[4:5], v[8:9] v_cndmask_b32_e32 v9, 0x4005ae14, v14, vcc_lo v_cndmask_b32_e64 v8, 0x7ae147ae, 0, vcc_lo v_cndmask_b32_e64 v14, 0x66666666, 0, vcc_lo v_add_f64 v[6:7], v[6:7], -v[10:11] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_f64 v[4:5], v[4:5], v[6:7] v_add_f64 v[6:7], v[16:17], v[4:5] s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_3) v_add_f64 v[10:11], v[6:7], -v[16:17] v_mul_f64 v[12:13], v[8:9], v[6:7] v_mul_f64 v[16:17], v[14:15], v[6:7] v_add_f64 v[4:5], v[4:5], -v[10:11] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_fma_f64 v[10:11], v[8:9], v[6:7], -v[12:13] v_fma_f64 v[6:7], v[14:15], v[6:7], -v[16:17] v_cmp_class_f64_e64 vcc_lo, v[12:13], 0x204 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_fma_f64 v[10:11], v[8:9], v[4:5], v[10:11] v_fma_f64 v[4:5], v[14:15], v[4:5], v[6:7] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_f64 v[18:19], v[12:13], v[10:11] v_add_f64 v[22:23], v[16:17], v[4:5] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(VALU_DEP_4) v_dual_cndmask_b32 v7, v19, v13 :: v_dual_cndmask_b32 v6, v18, v12 v_cmp_class_f64_e64 vcc_lo, v[16:17], 0x204 v_add_f64 v[12:13], v[18:19], -v[12:13] v_cndmask_b32_e32 v25, v23, v17, vcc_lo s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_3) v_mul_f64 v[20:21], v[6:7], s[4:5] v_add_f64 v[10:11], v[10:11], -v[12:13] v_cndmask_b32_e32 v24, v22, v16, vcc_lo v_cmp_neq_f64_e64 vcc_lo, 0x7ff00000, |v[6:7]| v_add_f64 v[16:17], v[22:23], -v[16:17] v_rndne_f64_e32 v[20:21], v[20:21] v_cndmask_b32_e32 v11, 0, v11, vcc_lo v_mul_f64 v[26:27], v[24:25], s[4:5] s_mov_b32 s5, 0x3e928af3 s_mov_b32 s4, 0xfca7ab0c v_cndmask_b32_e32 v10, 0, v10, vcc_lo v_add_f64 v[4:5], v[4:5], -v[16:17] v_fma_f64 v[28:29], v[20:21], s[0:1], v[6:7] v_cvt_i32_f64_e32 v36, v[20:21] v_rndne_f64_e32 v[26:27], v[26:27] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2) v_fma_f64 v[28:29], v[20:21], s[2:3], v[28:29] v_fma_f64 v[30:31], v[26:27], s[0:1], v[24:25] s_mov_b32 s1, 0x3e5ade15 s_mov_b32 s0, 0x6a5dcb37 s_delay_alu instid0(VALU_DEP_2) | instid1(SALU_CYCLE_1) v_fma_f64 v[32:33], v[28:29], s[0:1], s[4:5] s_delay_alu instid0(VALU_DEP_2) v_fma_f64 v[30:31], v[26:27], s[2:3], v[30:31] s_mov_b32 s3, 0x3ec71dee s_mov_b32 s2, 0x623fde64 s_delay_alu instid0(VALU_DEP_2) | instid1(SALU_CYCLE_1) v_fma_f64 v[32:33], v[28:29], v[32:33], s[2:3] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_4) | instid1(VALU_DEP_4) v_fma_f64 v[34:35], v[30:31], s[0:1], s[4:5] s_mov_b32 s1, 0x3efa0199 s_mov_b32 s0, 0x7c89e6b0 v_cmp_ngt_f64_e64 s4, 0xc090cc00, v[24:25] v_cmp_class_f64_e64 s5, v[2:3], 0x204 v_fma_f64 v[32:33], v[28:29], v[32:33], s[0:1] s_delay_alu instid0(VALU_DEP_4) v_fma_f64 v[34:35], v[30:31], v[34:35], s[2:3] s_mov_b32 s3, 0x3f2a01a0 s_mov_b32 s2, 0x14761f6e s_delay_alu instid0(VALU_DEP_2) | instid1(SALU_CYCLE_1) v_fma_f64 v[32:33], v[28:29], v[32:33], s[2:3] s_delay_alu instid0(VALU_DEP_2) v_fma_f64 v[34:35], v[30:31], v[34:35], s[0:1] s_mov_b32 s1, 0x3f56c16c s_mov_b32 s0, 0x1852b7b0 s_delay_alu instid0(VALU_DEP_2) | instid1(SALU_CYCLE_1) v_fma_f64 v[32:33], v[28:29], v[32:33], s[0:1] s_delay_alu instid0(VALU_DEP_2) v_fma_f64 v[34:35], v[30:31], v[34:35], s[2:3] s_mov_b32 s3, 0x3f811111 s_mov_b32 s2, 0x11122322 s_delay_alu instid0(VALU_DEP_2) | instid1(SALU_CYCLE_1) v_fma_f64 v[32:33], v[28:29], v[32:33], s[2:3] s_delay_alu instid0(VALU_DEP_2) v_fma_f64 v[34:35], v[30:31], v[34:35], s[0:1] s_mov_b32 s1, 0x3fa55555 s_mov_b32 s0, 0x555502a1 s_delay_alu instid0(VALU_DEP_2) | instid1(SALU_CYCLE_1) v_fma_f64 v[32:33], v[28:29], v[32:33], s[0:1] s_delay_alu instid0(VALU_DEP_2) v_fma_f64 v[34:35], v[30:31], v[34:35], s[2:3] s_mov_b32 s3, 0x3fc55555 s_mov_b32 s2, 0x55555511 s_delay_alu instid0(VALU_DEP_2) | instid1(SALU_CYCLE_1) v_fma_f64 v[32:33], v[28:29], v[32:33], s[2:3] s_delay_alu instid0(VALU_DEP_2) v_fma_f64 v[34:35], v[30:31], v[34:35], s[0:1] s_mov_b32 s1, 0x3fe00000 s_mov_b32 s0, 11 s_delay_alu instid0(VALU_DEP_2) | instid1(SALU_CYCLE_1) v_fma_f64 v[32:33], v[28:29], v[32:33], s[0:1] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(VALU_DEP_4) v_fma_f64 v[34:35], v[30:31], v[34:35], s[2:3] v_cmp_nlt_f64_e64 s3, 0x40900000, v[24:25] v_cmp_neq_f64_e64 s2, 0x7ff00000, |v[24:25]| v_fma_f64 v[32:33], v[28:29], v[32:33], 1.0 s_delay_alu instid0(VALU_DEP_4) v_fma_f64 v[34:35], v[30:31], v[34:35], s[0:1] v_cmp_nlt_f64_e64 s0, 0x40900000, v[6:7] v_cmp_ngt_f64_e64 s1, 0xc090cc00, v[6:7] v_trunc_f64_e32 v[6:7], v[8:9] v_cndmask_b32_e64 v5, 0, v5, s2 v_cndmask_b32_e64 v4, 0, v4, s2 s_and_b32 s2, s4, s3 v_fma_f64 v[20:21], v[28:29], v[32:33], 1.0 v_cvt_i32_f64_e32 v32, v[26:27] v_fma_f64 v[28:29], v[30:31], v[34:35], 1.0 s_and_b32 vcc_lo, s1, s0 s_delay_alu instid0(VALU_DEP_3) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_ldexp_f64 v[18:19], v[20:21], v36 v_mul_f64 v[20:21], v[8:9], 0.5 v_fma_f64 v[26:27], v[30:31], v[28:29], 1.0 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_cndmask_b32_e64 v19, 0x7ff00000, v19, s0 v_trunc_f64_e32 v[12:13], v[20:21] v_cndmask_b32_e32 v18, 0, v18, vcc_lo v_cmp_eq_f64_e32 vcc_lo, v[6:7], v[8:9] v_ldexp_f64 v[22:23], v[26:27], v32 v_mul_f64 v[26:27], v[14:15], 0.5 v_cndmask_b32_e64 v19, 0, v19, s1 v_trunc_f64_e32 v[8:9], v[14:15] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_4) | instid1(VALU_DEP_4) v_fma_f64 v[6:7], v[18:19], v[10:11], v[18:19] v_cmp_class_f64_e64 s1, v[18:19], 0x204 v_cmp_neq_f64_e64 s0, v[12:13], v[20:21] v_cndmask_b32_e64 v12, 0x7ff00000, v23, s3 v_trunc_f64_e32 v[10:11], v[26:27] v_cndmask_b32_e64 v16, v6, v18, s1 v_cndmask_b32_e64 v6, v7, v19, s1 v_cmp_eq_f64_e64 s1, 0, v[2:3] v_cndmask_b32_e64 v7, 0, v12, s4 s_delay_alu instid0(VALU_DEP_4) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_cndmask_b32_e32 v18, 0, v16, vcc_lo s_and_b32 s0, vcc_lo, s0 v_cndmask_b32_e64 v13, 0x3ff00000, v3, s0 v_cmp_neq_f64_e64 s3, v[10:11], v[26:27] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_3) | instid1(VALU_DEP_4) v_bfi_b32 v13, 0x7fffffff, v6, v13 v_cndmask_b32_e64 v6, 0, v22, s2 v_cmp_eq_f64_e64 s2, v[8:9], v[14:15] v_cndmask_b32_e64 v9, 0, v3, s0 v_cndmask_b32_e32 v17, 0x7ff80000, v13, vcc_lo s_delay_alu instid0(VALU_DEP_4) v_fma_f64 v[4:5], v[6:7], v[4:5], v[6:7] v_cmp_class_f64_e64 s4, v[6:7], 0x204 v_cmp_gt_f64_e32 vcc_lo, 0, v[2:3] v_cndmask_b32_e64 v8, 0x7ff00000, 0, s1 s_or_b32 s0, s1, s5 v_cmp_u_f64_e64 s1, v[2:3], v[2:3] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_bfi_b32 v9, 0x7fffffff, v8, v9 s_and_b32 s3, s2, s3 v_cndmask_b32_e64 v2, 0x3ff00000, v3, s3 v_cndmask_b32_e64 v3, 0, v3, s3 v_cndmask_b32_e64 v5, v5, v7, s4 v_cndmask_b32_e32 v11, v13, v17, vcc_lo v_cndmask_b32_e64 v4, v4, v6, s4 s_delay_alu instid0(VALU_DEP_4) | instskip(SKIP_3) | instid1(VALU_DEP_2) v_bfi_b32 v3, 0x7fffffff, v8, v3 v_cndmask_b32_e32 v10, v16, v18, vcc_lo v_bfi_b32 v2, 0x7fffffff, v5, v2 v_cndmask_b32_e64 v9, v11, v9, s0 v_cndmask_b32_e64 v5, 0x7ff80000, v2, s2 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_cndmask_b32_e32 v2, v2, v5, vcc_lo v_cndmask_b32_e64 v5, 0, v4, s2 v_cndmask_b32_e64 v6, v2, v3, s0 s_delay_alu instid0(VALU_DEP_2) v_cndmask_b32_e32 v4, v4, v5, vcc_lo s_or_b32 s0, s1, s0 v_cndmask_b32_e64 v3, v9, 0x7ff80000, s1 v_cndmask_b32_e64 v2, v10, 0, s0 v_cndmask_b32_e64 v5, v6, 0x7ff80000, s1 v_cndmask_b32_e64 v4, v4, 0, s0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_4) v_mul_f64 v[6:7], v[4:5], 0x40140000 v_div_scale_f64 v[8:9], null, v[4:5], v[4:5], v[2:3] v_add_f64 v[22:23], v[2:3], v[4:5] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_fma_f64 v[6:7], v[2:3], 2.0, v[6:7] v_rcp_f64_e32 v[10:11], v[8:9] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_3) v_div_scale_f64 v[12:13], null, v[6:7], v[6:7], v[4:5] v_add_f64 v[22:23], v[22:23], v[6:7] s_waitcnt_depctr 0xfff v_fma_f64 v[16:17], -v[8:9], v[10:11], 1.0 v_rcp_f64_e32 v[14:15], v[12:13] s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_2) v_fma_f64 v[10:11], v[10:11], v[16:17], v[10:11] s_waitcnt_depctr 0xfff v_fma_f64 v[16:17], -v[12:13], v[14:15], 1.0 v_fma_f64 v[18:19], -v[8:9], v[10:11], 1.0 v_fma_f64 v[14:15], v[14:15], v[16:17], v[14:15] v_div_scale_f64 v[16:17], vcc_lo, v[2:3], v[4:5], v[2:3] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_fma_f64 v[10:11], v[10:11], v[18:19], v[10:11] v_fma_f64 v[18:19], -v[12:13], v[14:15], 1.0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_f64 v[20:21], v[16:17], v[10:11] v_fma_f64 v[14:15], v[14:15], v[18:19], v[14:15] v_div_scale_f64 v[18:19], s0, v[4:5], v[6:7], v[4:5] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2) v_fma_f64 v[8:9], -v[8:9], v[20:21], v[16:17] v_mul_f64 v[16:17], v[18:19], v[14:15] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_div_fmas_f64 v[8:9], v[8:9], v[10:11], v[20:21] s_mov_b32 vcc_lo, s0 v_fma_f64 v[10:11], -v[12:13], v[16:17], v[18:19] v_fma_f64 v[12:13], v[2:3], v[4:5], v[22:23] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_div_fixup_f64 v[2:3], v[8:9], v[4:5], v[2:3] v_div_fmas_f64 v[8:9], v[10:11], v[14:15], v[16:17] v_add_co_u32 v0, vcc_lo, s6, v0 s_delay_alu instid0(VALU_DEP_3) | instskip(SKIP_1) | instid1(VALU_DEP_4) v_add_f64 v[2:3], v[2:3], v[12:13] v_add_co_ci_u32_e32 v1, vcc_lo, s7, v1, vcc_lo v_div_fixup_f64 v[4:5], v[8:9], v[6:7], v[4:5] s_delay_alu instid0(VALU_DEP_1) v_add_f64 v[2:3], v[4:5], v[2:3] global_store_b64 v[0:1], v[2:3], off .LBB0_2: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z8MyKernelPdS_ .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 272 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 37 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z8MyKernelPdS_, .Lfunc_end0-_Z8MyKernelPdS_ .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .offset: 16 .size: 4 .value_kind: hidden_block_count_x - .offset: 20 .size: 4 .value_kind: hidden_block_count_y - .offset: 24 .size: 4 .value_kind: hidden_block_count_z - .offset: 28 .size: 2 .value_kind: hidden_group_size_x - .offset: 30 .size: 2 .value_kind: hidden_group_size_y - .offset: 32 .size: 2 .value_kind: hidden_group_size_z - .offset: 34 .size: 2 .value_kind: hidden_remainder_x - .offset: 36 .size: 2 .value_kind: hidden_remainder_y - .offset: 38 .size: 2 .value_kind: hidden_remainder_z - .offset: 56 .size: 8 .value_kind: hidden_global_offset_x - .offset: 64 .size: 8 .value_kind: hidden_global_offset_y - .offset: 72 .size: 8 .value_kind: hidden_global_offset_z - .offset: 80 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 272 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z8MyKernelPdS_ .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z8MyKernelPdS_.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 37 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
/* Small CUDA exercise to try to improve efficiency by using two separate streams to set up a staged copying and execution (when instead of one large copy, followed by one large kernel computation, one does it in many small chunks, with copying and computing done in parallel in two streams, after first chunk was copied to the device). For the purpose of this exercise, ignore the copying of the results from the device to host at the end; only do the staged copy and execute for the copying of the initial data to the device + the kernel. We use cudaMallocHost to allocate arrays on host in pinned memory, which is both results in faster copying to/from GPU (compared to malloc), and also a CUDA requirement for copying running concurrently with a kernel. Make sure that the "Result:" value printed by the code is (almost) identical in both original and modified versions of the code. If not, you have a bug! Hints: You will have to use the following CUDA functions: - cudaStreamCreate - cudaMemcpyAsync - cudaStreamDestroy - cudaDeviceSynchronize * You will have to set up a for loop for multiple chunks copying and kernel execution; * Number of chunks should be a variable (or macro parameter); for simplicity, make NMAX dividable by the number of chunks; * In cudaMemcpyAsync the first two arguments should be "&d_A[ind], &h_A[ind]", not "d_A, h_A", ind being the starting index for the current chunk to copy; * You'll have to pass two more arguments to the kernel - ind and number of threads per chunk; * Nblocks will be different - shoul be computed per chunk. * You'll have to modify the kernel slightly; At the end, you should get the timings (based on 10 runs, NMAX=1000000, BLOCK_SIZE=128) similar to this: NCHUNKS t, ms - 2.76 - the original (non-staged) version of the code 1 2.72 - result is similar to the non-staged code 2 2.08 - even with only 2 chunks, we already see 33% speedup 4 1.79 5 1.75 10 1.72 - seems to be the best timing, 60% faster than the original code 20 1.87 - as NCHUNKS increases, the results get worse. Why? 100 3.53 - for too many NCHUNKS results can get even worse than in non-staged code To compile: nvcc -arch=sm_20 -O2 staged.cu -o staged The best/average timings: ../best_time.sh ./staged */ #include <hip/hip_runtime.h> #include <sys/time.h> #include <ctype.h> #include <math.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <time.h> #include <unistd.h> // Number of times to run the test (for better timings accuracy): #define NTESTS 100 // Number of threads in one block (possible range is 32...1024): #define BLOCK_SIZE 128 // Total number of threads (total number of elements to process in the kernel): #define NMAX 1000000 /* Subtract the `struct timeval' values X and Y, storing the result in RESULT. Return 1 if the difference is negative, otherwise 0. */ // It messes up with y! int timeval_subtract (double *result, struct timeval *x, struct timeval *y) { struct timeval result0; /* Perform the carry for the later subtraction by updating y. */ if (x->tv_usec < y->tv_usec) { int nsec = (y->tv_usec - x->tv_usec) / 1000000 + 1; y->tv_usec -= 1000000 * nsec; y->tv_sec += nsec; } if (x->tv_usec - y->tv_usec > 1000000) { int nsec = (y->tv_usec - x->tv_usec) / 1000000; y->tv_usec += 1000000 * nsec; y->tv_sec -= nsec; } /* Compute the time remaining to wait. tv_usec is certainly positive. */ result0.tv_sec = x->tv_sec - y->tv_sec; result0.tv_usec = x->tv_usec - y->tv_usec; *result = ((double)result0.tv_usec)/1e6 + (double)result0.tv_sec; /* Return 1 if result is negative. */ return x->tv_sec < y->tv_sec; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ // The kernel: __global__ void MyKernel (double *d_A, double *d_B) { double x, y, z; int i = threadIdx.x + blockDim.x * blockIdx.x; if (i >= NMAX) return; // Some meaningless cpu-intensive computation: x = pow(d_A[i], 2.71); y = pow(d_A[i], 0.35); z = 2*x + 5*y; d_B[i] = x + y + z + x*y + x/y + y/z; return; } //++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ int main (int argc,char **argv) { struct timeval tdr0, tdr1, tdr, tdr01; double restime, restime0, restime1; int devid, devcount, error, Max_gridsize; double *h_A, *h_B, *d_A, *d_B; /* find number of device in current "context" */ hipGetDevice(&devid); /* find how many devices are available */ if (hipGetDeviceCount(&devcount) || devcount==0) { printf ("No CUDA devices!\n"); exit (1); } else { hipDeviceProp_t deviceProp; hipGetDeviceProperties (&deviceProp, devid); printf ("Device count, devid: %d %d\n", devcount, devid); printf ("Device: %s\n", deviceProp.name); printf("[deviceProp.major.deviceProp.minor] = [%d.%d]\n\n", deviceProp.major, deviceProp.minor); Max_gridsize = deviceProp.maxGridSize[0]; } // Loop to run the timing test multiple times: int kk; for (kk=0; kk<NTESTS; kk++) { // Using cudaMallocHost (intead of malloc) to accelerate data copying: // Initial data array on host: if (error = hipHostMalloc (&h_A, NMAX*sizeof(double), hipHostMallocDefault)) { printf ("Error %d\n", error); exit (error); } // Results array on host: if (error = hipHostMalloc (&h_B, NMAX*sizeof(double), hipHostMallocDefault)) { printf ("Error %d\n", error); exit (error); } // ALlocating arrays on GPU: if (error = hipMalloc (&d_A, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } if (error = hipMalloc (&d_B, NMAX*sizeof(double))) { printf ("Error %d\n", error); exit (error); } // Initializing the input array: for (int i=0; i<NMAX; i++) h_A[i] = (double)rand()/(double)RAND_MAX; // Number of blocks of threads: int Nblocks = (NMAX+BLOCK_SIZE-1) / BLOCK_SIZE; if (Nblocks > Max_gridsize) { printf ("Nblocks > Max_gridsize! %d %d\n", Nblocks, Max_gridsize); exit (1); } if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr0, NULL); //-------------------------------------------------------------------------------- // Copying the data to device (we time it): if (error = hipMemcpy (d_A, h_A, NMAX*sizeof(double), hipMemcpyHostToDevice)) { printf ("Error %d\n", error); exit (error); } // Intermediate timing, to measure timings separately for copying and kernel execution // (Should be removed in the solution code) if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr01, NULL); // The kernel call: MyKernel <<<Nblocks, BLOCK_SIZE>>> (d_A, d_B); //-------------------------------------------------------------------------------- if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } gettimeofday (&tdr1, NULL); tdr = tdr0; timeval_subtract (&restime, &tdr1, &tdr); // Copying the result back to host (we don't time it): if (error = hipMemcpy (h_B, d_B, NMAX*sizeof(double), hipMemcpyDeviceToHost)) { printf ("Error %d\n", error); exit (error); } if (error = hipDeviceSynchronize()) { printf ("Error %d\n", error); exit (error); } // Adding up the results, for accuracy/correctness testing: double result = 0.0; for (int i=0; i<NMAX; i++) { result += h_B[i]; } tdr = tdr0; timeval_subtract (&restime0, &tdr01, &tdr); tdr = tdr01; timeval_subtract (&restime1, &tdr1, &tdr); printf ("Individual timings: %e %e\n", restime0, restime1); printf ("Result: %e\n\n", result); printf ("Time: %e\n", restime); hipHostFree (h_A); hipHostFree (h_B); hipFree (d_A); hipFree (d_B); } // kk loop return 0; }
.text .file "staged.hip" .section .rodata.cst8,"aM",@progbits,8 .p2align 3, 0x0 # -- Begin function _Z16timeval_subtractPdP7timevalS1_ .LCPI0_0: .quad 0x412e848000000000 # double 1.0E+6 .text .globl _Z16timeval_subtractPdP7timevalS1_ .p2align 4, 0x90 .type _Z16timeval_subtractPdP7timevalS1_,@function _Z16timeval_subtractPdP7timevalS1_: # @_Z16timeval_subtractPdP7timevalS1_ .cfi_startproc # %bb.0: movq %rdx, %rcx movq 8(%rdx), %r9 movq %r9, %rax movabsq $4835703278458516699, %r8 # imm = 0x431BDE82D7B634DB subq 8(%rsi), %rax jle .LBB0_2 # %bb.1: imulq %r8 movq %rdx, %rax shrq $18, %rax shrq $63, %rdx addl %edx, %eax incl %eax imull $1000000, %eax, %edx # imm = 0xF4240 movslq %edx, %rdx subq %rdx, %r9 movq %r9, 8(%rcx) cltq addq %rax, (%rcx) .LBB0_2: movq 8(%rsi), %rdx movq 8(%rcx), %r9 movq %rdx, %rax subq %r9, %rax cmpq $1000001, %rax # imm = 0xF4241 jl .LBB0_4 # %bb.3: movq %r9, %rax subq %rdx, %rax imulq %r8 movq %rdx, %rax shrq $63, %rax sarq $18, %rdx addq %rax, %rdx movabsq $4294967296000000, %rax # imm = 0xF424000000000 imulq %rdx, %rax sarq $32, %rax addq %r9, %rax movq %rax, 8(%rcx) movslq %edx, %rax subq %rax, (%rcx) .LBB0_4: movq 8(%rsi), %rax subq 8(%rcx), %rax cvtsi2sd %rax, %xmm0 movq (%rsi), %rdx divsd .LCPI0_0(%rip), %xmm0 xorl %eax, %eax subq (%rcx), %rdx cvtsi2sd %rdx, %xmm1 addsd %xmm0, %xmm1 movsd %xmm1, (%rdi) setl %al retq .Lfunc_end0: .size _Z16timeval_subtractPdP7timevalS1_, .Lfunc_end0-_Z16timeval_subtractPdP7timevalS1_ .cfi_endproc # -- End function .globl _Z23__device_stub__MyKernelPdS_ # -- Begin function _Z23__device_stub__MyKernelPdS_ .p2align 4, 0x90 .type _Z23__device_stub__MyKernelPdS_,@function _Z23__device_stub__MyKernelPdS_: # @_Z23__device_stub__MyKernelPdS_ .cfi_startproc # %bb.0: subq $88, %rsp .cfi_def_cfa_offset 96 movq %rdi, 56(%rsp) movq %rsi, 48(%rsp) leaq 56(%rsp), %rax movq %rax, 64(%rsp) leaq 48(%rsp), %rax movq %rax, 72(%rsp) leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 64(%rsp), %r9 movl $_Z8MyKernelPdS_, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $104, %rsp .cfi_adjust_cfa_offset -104 retq .Lfunc_end1: .size _Z23__device_stub__MyKernelPdS_, .Lfunc_end1-_Z23__device_stub__MyKernelPdS_ .cfi_endproc # -- End function .section .rodata.cst8,"aM",@progbits,8 .p2align 3, 0x0 # -- Begin function main .LCPI2_0: .quad 0x41dfffffffc00000 # double 2147483647 .LCPI2_1: .quad 0x412e848000000000 # double 1.0E+6 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $1672, %rsp # imm = 0x688 .cfi_def_cfa_offset 1728 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 leaq 20(%rsp), %rdi callq hipGetDevice leaq 16(%rsp), %rdi callq hipGetDeviceCount testl %eax, %eax jne .LBB2_36 # %bb.1: cmpl $0, 16(%rsp) je .LBB2_36 # %bb.2: movabsq $4294967424, %rbx # imm = 0x100000080 movl 20(%rsp), %esi leaq 192(%rsp), %r14 movq %r14, %rdi callq hipGetDevicePropertiesR0600 movl 16(%rsp), %esi movl 20(%rsp), %edx xorl %r15d, %r15d movl $.L.str.1, %edi xorl %eax, %eax callq printf movl $.L.str.2, %edi movq %r14, %rsi xorl %eax, %eax callq printf movl 552(%rsp), %esi movl 556(%rsp), %edx movl $.L.str.3, %edi xorl %eax, %eax callq printf movl 528(%rsp), %eax movl %eax, 12(%rsp) # 4-byte Spill leaq 7685(%rbx), %rax movq %rax, 72(%rsp) # 8-byte Spill jmp .LBB2_3 .p2align 4, 0x90 .LBB2_33: # in Loop: Header=BB2_3 Depth=1 movsd %xmm1, 64(%rsp) # 8-byte Spill movq %rcx, %rax subq %r14, %rax movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $63, %rax sarq $18, %rdx addq %rax, %rdx movslq %edx, %rax movabsq $4294967296000000, %r8 # imm = 0xF424000000000 imulq %r8, %rdx sarq $32, %rdx addq %rdx, %rcx subq %rax, %rdi .LBB2_34: # %_Z16timeval_subtractPdP7timevalS1_.exit66 # in Loop: Header=BB2_3 Depth=1 subq %r13, %rsi xorps %xmm1, %xmm1 cvtsi2sd %rsi, %xmm1 addsd %xmm1, %xmm0 subq %rdi, %rbp subq %rcx, %r14 cvtsi2sd %r14, %xmm2 divsd .LCPI2_1(%rip), %xmm2 xorps %xmm1, %xmm1 cvtsi2sd %rbp, %xmm1 addsd %xmm2, %xmm1 movl $.L.str.6, %edi movb $2, %al callq printf movl $.L.str.7, %edi movsd 64(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero movb $1, %al callq printf movl $.L.str.8, %edi movsd 24(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero movb $1, %al callq printf movq 56(%rsp), %rdi callq hipHostFree movq 48(%rsp), %rdi callq hipHostFree movq 40(%rsp), %rdi callq hipFree movq 32(%rsp), %rdi callq hipFree incl %r15d cmpl $100, %r15d je .LBB2_35 .LBB2_3: # =>This Loop Header: Depth=1 # Child Loop BB2_8 Depth 2 # Child Loop BB2_23 Depth 2 movl $8000000, %esi # imm = 0x7A1200 leaq 56(%rsp), %rdi xorl %edx, %edx callq hipHostMalloc testl %eax, %eax jne .LBB2_37 # %bb.4: # in Loop: Header=BB2_3 Depth=1 movl $8000000, %esi # imm = 0x7A1200 leaq 48(%rsp), %rdi xorl %edx, %edx callq hipHostMalloc testl %eax, %eax jne .LBB2_38 # %bb.5: # in Loop: Header=BB2_3 Depth=1 movl $8000000, %esi # imm = 0x7A1200 leaq 40(%rsp), %rdi callq hipMalloc testl %eax, %eax jne .LBB2_38 # %bb.6: # in Loop: Header=BB2_3 Depth=1 movl $8000000, %esi # imm = 0x7A1200 leaq 32(%rsp), %rdi callq hipMalloc testl %eax, %eax jne .LBB2_38 # %bb.7: # %.preheader95.preheader # in Loop: Header=BB2_3 Depth=1 xorl %ebx, %ebx .p2align 4, 0x90 .LBB2_8: # %.preheader95 # Parent Loop BB2_3 Depth=1 # => This Inner Loop Header: Depth=2 callq rand xorps %xmm0, %xmm0 cvtsi2sd %eax, %xmm0 divsd .LCPI2_0(%rip), %xmm0 movq 56(%rsp), %rax movsd %xmm0, (%rax,%rbx,8) incq %rbx cmpq $1000000, %rbx # imm = 0xF4240 jne .LBB2_8 # %bb.9: # in Loop: Header=BB2_3 Depth=1 cmpl $7813, 12(%rsp) # 4-byte Folded Reload # imm = 0x1E85 jl .LBB2_39 # %bb.10: # in Loop: Header=BB2_3 Depth=1 callq hipDeviceSynchronize testl %eax, %eax jne .LBB2_38 # %bb.11: # in Loop: Header=BB2_3 Depth=1 leaq 176(%rsp), %rdi xorl %esi, %esi callq gettimeofday movq 40(%rsp), %rdi movq 56(%rsp), %rsi movl $8000000, %edx # imm = 0x7A1200 movl $1, %ecx callq hipMemcpy testl %eax, %eax jne .LBB2_38 # %bb.12: # in Loop: Header=BB2_3 Depth=1 callq hipDeviceSynchronize testl %eax, %eax jne .LBB2_38 # %bb.13: # in Loop: Header=BB2_3 Depth=1 leaq 144(%rsp), %rdi xorl %esi, %esi callq gettimeofday movq 72(%rsp), %rdi # 8-byte Reload movl $1, %esi movabsq $4294967424, %rdx # imm = 0x100000080 movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB2_15 # %bb.14: # in Loop: Header=BB2_3 Depth=1 movq 40(%rsp), %rax movq 32(%rsp), %rcx movq %rax, 136(%rsp) movq %rcx, 128(%rsp) leaq 136(%rsp), %rax movq %rax, 192(%rsp) leaq 128(%rsp), %rax movq %rax, 200(%rsp) leaq 112(%rsp), %rdi leaq 96(%rsp), %rsi leaq 88(%rsp), %rdx leaq 80(%rsp), %rcx callq __hipPopCallConfiguration movq 112(%rsp), %rsi movl 120(%rsp), %edx movq 96(%rsp), %rcx movl 104(%rsp), %r8d movl $_Z8MyKernelPdS_, %edi leaq 192(%rsp), %r9 pushq 80(%rsp) .cfi_adjust_cfa_offset 8 pushq 96(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB2_15: # in Loop: Header=BB2_3 Depth=1 callq hipDeviceSynchronize testl %eax, %eax jne .LBB2_38 # %bb.16: # in Loop: Header=BB2_3 Depth=1 leaq 160(%rsp), %rdi xorl %esi, %esi callq gettimeofday movq 176(%rsp), %r13 movq 184(%rsp), %r12 movq 168(%rsp), %r14 movq %r12, %rax movq %r13, %rbx movq %r12, %rcx subq %r14, %rax jle .LBB2_18 # %bb.17: # in Loop: Header=BB2_3 Depth=1 movabsq $4835703278458516699, %rcx # imm = 0x431BDE82D7B634DB imulq %rcx movq %rdx, %rax shrq $18, %rax shrq $63, %rdx addl %edx, %eax incl %eax imull $1000000, %eax, %ecx # imm = 0xF4240 movslq %ecx, %rdx movq %r12, %rcx subq %rdx, %rcx movslq %eax, %rbx addq %r13, %rbx .LBB2_18: # in Loop: Header=BB2_3 Depth=1 movq %r14, %rax subq %rcx, %rax cmpq $1000001, %rax # imm = 0xF4241 jl .LBB2_20 # %bb.19: # in Loop: Header=BB2_3 Depth=1 movq %rcx, %rax subq %r14, %rax movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $63, %rax sarq $18, %rdx addq %rax, %rdx movslq %edx, %rax movabsq $4294967296000000, %rsi # imm = 0xF424000000000 imulq %rsi, %rdx sarq $32, %rdx addq %rdx, %rcx subq %rax, %rbx .LBB2_20: # %_Z16timeval_subtractPdP7timevalS1_.exit # in Loop: Header=BB2_3 Depth=1 movq %r14, %rax subq %rcx, %rax xorps %xmm0, %xmm0 cvtsi2sd %rax, %xmm0 movq 160(%rsp), %rbp divsd .LCPI2_1(%rip), %xmm0 movsd %xmm0, 24(%rsp) # 8-byte Spill movq 48(%rsp), %rdi movq 32(%rsp), %rsi movl $8000000, %edx # imm = 0x7A1200 movl $2, %ecx callq hipMemcpy testl %eax, %eax jne .LBB2_38 # %bb.21: # in Loop: Header=BB2_3 Depth=1 callq hipDeviceSynchronize testl %eax, %eax jne .LBB2_38 # %bb.22: # %.preheader # in Loop: Header=BB2_3 Depth=1 movq %rbp, %rax subq %rbx, %rax xorps %xmm0, %xmm0 cvtsi2sd %rax, %xmm0 movsd 24(%rsp), %xmm1 # 8-byte Reload # xmm1 = mem[0],zero addsd %xmm0, %xmm1 movapd %xmm1, %xmm0 xorpd %xmm1, %xmm1 movq 48(%rsp), %rax xorl %ecx, %ecx .p2align 4, 0x90 .LBB2_23: # Parent Loop BB2_3 Depth=1 # => This Inner Loop Header: Depth=2 addsd (%rax,%rcx,8), %xmm1 incq %rcx cmpq $1000000, %rcx # imm = 0xF4240 jne .LBB2_23 # %bb.24: # in Loop: Header=BB2_3 Depth=1 movq 152(%rsp), %rcx movq %r12, %rax subq %rcx, %rax jle .LBB2_26 # %bb.25: # in Loop: Header=BB2_3 Depth=1 movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $18, %rax shrq $63, %rdx addl %edx, %eax incl %eax imull $1000000, %eax, %edx # imm = 0xF4240 movslq %edx, %rdx subq %rdx, %r12 cltq addq %rax, %r13 .LBB2_26: # in Loop: Header=BB2_3 Depth=1 movq %rcx, %rax subq %r12, %rax cmpq $1000001, %rax # imm = 0xF4241 movsd %xmm0, 24(%rsp) # 8-byte Spill jl .LBB2_28 # %bb.27: # in Loop: Header=BB2_3 Depth=1 movq %r12, %rax subq %rcx, %rax movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $63, %rax sarq $18, %rdx addq %rax, %rdx movslq %edx, %rax movabsq $4294967296000000, %rsi # imm = 0xF424000000000 imulq %rsi, %rdx sarq $32, %rdx addq %rdx, %r12 subq %rax, %r13 .LBB2_28: # %_Z16timeval_subtractPdP7timevalS1_.exit63 # in Loop: Header=BB2_3 Depth=1 movq %rcx, %rax subq %r12, %rax xorps %xmm0, %xmm0 cvtsi2sd %rax, %xmm0 movq 144(%rsp), %rsi divsd .LCPI2_1(%rip), %xmm0 movq %rcx, %rax subq %r14, %rax jle .LBB2_29 # %bb.30: # in Loop: Header=BB2_3 Depth=1 movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $18, %rax shrq $63, %rdx addl %edx, %eax incl %eax imull $1000000, %eax, %edx # imm = 0xF4240 movslq %edx, %rdx subq %rdx, %rcx movslq %eax, %rdi addq %rsi, %rdi jmp .LBB2_31 .p2align 4, 0x90 .LBB2_29: # in Loop: Header=BB2_3 Depth=1 movq %rsi, %rdi .LBB2_31: # in Loop: Header=BB2_3 Depth=1 movq %r14, %rax subq %rcx, %rax cmpq $1000001, %rax # imm = 0xF4241 jge .LBB2_33 # %bb.32: # in Loop: Header=BB2_3 Depth=1 movsd %xmm1, 64(%rsp) # 8-byte Spill jmp .LBB2_34 .LBB2_35: xorl %eax, %eax addq $1672, %rsp # imm = 0x688 .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .LBB2_38: .cfi_def_cfa_offset 1728 movl $.L.str.4, %edi movl %eax, %esi movl %eax, %ebx xorl %eax, %eax callq printf movl %ebx, %edi callq exit .LBB2_39: movl $.L.str.5, %edi movl $7813, %esi # imm = 0x1E85 movl 12(%rsp), %edx # 4-byte Reload xorl %eax, %eax callq printf movl $1, %edi callq exit .LBB2_37: movl %eax, %r14d movl $.L.str.4, %edi movl %eax, %esi xorl %eax, %eax callq printf movl %r14d, %edi callq exit .LBB2_36: movl $.Lstr, %edi callq puts@PLT movl $1, %edi callq exit .Lfunc_end2: .size main, .Lfunc_end2-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB3_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB3_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z8MyKernelPdS_, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end3: .size __hip_module_ctor, .Lfunc_end3-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB4_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB4_2: retq .Lfunc_end4: .size __hip_module_dtor, .Lfunc_end4-__hip_module_dtor .cfi_endproc # -- End function .type _Z8MyKernelPdS_,@object # @_Z8MyKernelPdS_ .section .rodata,"a",@progbits .globl _Z8MyKernelPdS_ .p2align 3, 0x0 _Z8MyKernelPdS_: .quad _Z23__device_stub__MyKernelPdS_ .size _Z8MyKernelPdS_, 8 .type .L.str.1,@object # @.str.1 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.1: .asciz "Device count, devid: %d %d\n" .size .L.str.1, 28 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "Device: %s\n" .size .L.str.2, 12 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "[deviceProp.major.deviceProp.minor] = [%d.%d]\n\n" .size .L.str.3, 48 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "Error %d\n" .size .L.str.4, 10 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "Nblocks > Max_gridsize! %d %d\n" .size .L.str.5, 33 .type .L.str.6,@object # @.str.6 .L.str.6: .asciz "Individual timings: %e %e\n" .size .L.str.6, 27 .type .L.str.7,@object # @.str.7 .L.str.7: .asciz "Result: %e\n\n" .size .L.str.7, 13 .type .L.str.8,@object # @.str.8 .L.str.8: .asciz "Time: %e\n" .size .L.str.8, 10 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z8MyKernelPdS_" .size .L__unnamed_1, 16 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "No CUDA devices!" .size .Lstr, 17 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z23__device_stub__MyKernelPdS_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z8MyKernelPdS_ .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_000f5ee8_00000000-6_staged.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2074: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2074: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z16timeval_subtractPdP7timevalS1_ .type _Z16timeval_subtractPdP7timevalS1_, @function _Z16timeval_subtractPdP7timevalS1_: .LFB2070: .cfi_startproc endbr64 movq %rdx, %rcx movq 8(%rsi), %rax movq 8(%rdx), %r8 cmpq %r8, %rax jge .L4 movq %r8, %r9 subq %rax, %r9 movabsq $4835703278458516699, %rdx movq %r9, %rax imulq %rdx sarq $18, %rdx sarq $63, %r9 subq %r9, %rdx addl $1, %edx imull $1000000, %edx, %eax cltq subq %rax, %r8 movq %r8, 8(%rcx) movslq %edx, %rdx addq %rdx, (%rcx) .L4: movq 8(%rsi), %rax movq 8(%rcx), %r8 movq %rax, %rdx subq %r8, %rdx cmpq $1000000, %rdx jle .L5 movq %r8, %r9 subq %rax, %r9 movabsq $4835703278458516699, %rdx movq %r9, %rax imulq %rdx sarq $18, %rdx sarq $63, %r9 subq %r9, %rdx imull $1000000, %edx, %eax cltq addq %r8, %rax movq %rax, 8(%rcx) movslq %edx, %rdx subq %rdx, (%rcx) .L5: movq 8(%rsi), %rax subq 8(%rcx), %rax pxor %xmm0, %xmm0 cvtsi2sdq %rax, %xmm0 divsd .LC0(%rip), %xmm0 movq (%rsi), %rax subq (%rcx), %rax pxor %xmm1, %xmm1 cvtsi2sdq %rax, %xmm1 addsd %xmm1, %xmm0 movsd %xmm0, (%rdi) movq (%rcx), %rax cmpq %rax, (%rsi) setl %al movzbl %al, %eax ret .cfi_endproc .LFE2070: .size _Z16timeval_subtractPdP7timevalS1_, .-_Z16timeval_subtractPdP7timevalS1_ .globl _Z29__device_stub__Z8MyKernelPdS_PdS_ .type _Z29__device_stub__Z8MyKernelPdS_PdS_, @function _Z29__device_stub__Z8MyKernelPdS_PdS_: .LFB2096: .cfi_startproc endbr64 subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 8(%rsp) movq %rsi, (%rsp) movq %fs:40, %rax movq %rax, 104(%rsp) xorl %eax, %eax leaq 8(%rsp), %rax movq %rax, 80(%rsp) movq %rsp, %rax movq %rax, 88(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) leaq 24(%rsp), %rcx leaq 16(%rsp), %rdx leaq 44(%rsp), %rsi leaq 32(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L10 .L6: movq 104(%rsp), %rax subq %fs:40, %rax jne .L11 addq $120, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L10: .cfi_restore_state pushq 24(%rsp) .cfi_def_cfa_offset 136 pushq 24(%rsp) .cfi_def_cfa_offset 144 leaq 96(%rsp), %r9 movq 60(%rsp), %rcx movl 68(%rsp), %r8d movq 48(%rsp), %rsi movl 56(%rsp), %edx leaq _Z8MyKernelPdS_(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 128 jmp .L6 .L11: call __stack_chk_fail@PLT .cfi_endproc .LFE2096: .size _Z29__device_stub__Z8MyKernelPdS_PdS_, .-_Z29__device_stub__Z8MyKernelPdS_PdS_ .globl _Z8MyKernelPdS_ .type _Z8MyKernelPdS_, @function _Z8MyKernelPdS_: .LFB2097: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z8MyKernelPdS_PdS_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2097: .size _Z8MyKernelPdS_, .-_Z8MyKernelPdS_ .section .rodata.str1.1,"aMS",@progbits,1 .LC2: .string "Device count, devid: %d %d\n" .LC3: .string "Device: %s\n" .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC4: .string "[deviceProp.major.deviceProp.minor] = [%d.%d]\n\n" .section .rodata.str1.1 .LC5: .string "No CUDA devices!\n" .LC6: .string "Error %d\n" .section .rodata.str1.8 .align 8 .LC8: .string "Nblocks > Max_gridsize! %d %d\n" .section .rodata.str1.1 .LC9: .string "Individual timings: %e %e\n" .LC10: .string "Result: %e\n\n" .LC11: .string "Time: %e\n" .text .globl main .type main, @function main: .LFB2071: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $1224, %rsp .cfi_def_cfa_offset 1280 movq %fs:40, %rax movq %rax, 1208(%rsp) xorl %eax, %eax leaq 24(%rsp), %rdi call cudaGetDevice@PLT leaq 28(%rsp), %rdi call cudaGetDeviceCount@PLT testl %eax, %eax jne .L15 cmpl $0, 28(%rsp) je .L15 leaq 176(%rsp), %rbx movl 24(%rsp), %esi movq %rbx, %rdi call cudaGetDeviceProperties_v2@PLT movl 24(%rsp), %ecx movl 28(%rsp), %edx leaq .LC2(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq %rbx, %rdx leaq .LC3(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl 540(%rsp), %ecx movl 536(%rsp), %edx leaq .LC4(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl 512(%rsp), %r12d movl $100, %ebx leaq .LC9(%rip), %r14 leaq .LC10(%rip), %r13 leaq .LC11(%rip), %r15 jmp .L16 .L15: leaq .LC5(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl $1, %edi call exit@PLT .L40: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L41: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L42: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L43: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L44: movl %r12d, %ecx movl $7813, %edx leaq .LC8(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl $1, %edi call exit@PLT .L45: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L46: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L47: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L25: call cudaDeviceSynchronize@PLT movl %eax, %ebp testl %eax, %eax jne .L36 leaq 128(%rsp), %rbp movl $0, %esi movq %rbp, %rdi call gettimeofday@PLT movdqa 112(%rsp), %xmm3 movaps %xmm3, 144(%rsp) leaq 144(%rsp), %rdx leaq 32(%rsp), %rdi movq %rbp, %rsi call _Z16timeval_subtractPdP7timevalS1_ movl $2, %ecx movl $8000000, %edx movq 80(%rsp), %rsi movq 64(%rsp), %rdi call cudaMemcpy@PLT movl %eax, %ebp testl %eax, %eax jne .L37 call cudaDeviceSynchronize@PLT movl %eax, %ebp testl %eax, %eax jne .L38 movq 64(%rsp), %rax leaq 8000000(%rax), %rdx movq $0x000000000, 8(%rsp) .L29: movsd 8(%rsp), %xmm2 addsd (%rax), %xmm2 movsd %xmm2, 8(%rsp) addq $8, %rax cmpq %rdx, %rax jne .L29 movdqa 112(%rsp), %xmm4 movaps %xmm4, 144(%rsp) leaq 144(%rsp), %rbp leaq 160(%rsp), %rsi leaq 40(%rsp), %rdi movq %rbp, %rdx call _Z16timeval_subtractPdP7timevalS1_ movdqa 160(%rsp), %xmm5 movaps %xmm5, 144(%rsp) leaq 128(%rsp), %rsi leaq 48(%rsp), %rdi movq %rbp, %rdx call _Z16timeval_subtractPdP7timevalS1_ movsd 48(%rsp), %xmm1 movsd 40(%rsp), %xmm0 movq %r14, %rsi movl $2, %edi movl $2, %eax call __printf_chk@PLT movsd 8(%rsp), %xmm0 movq %r13, %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movsd 32(%rsp), %xmm0 movq %r15, %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movq 56(%rsp), %rdi call cudaFreeHost@PLT movq 64(%rsp), %rdi call cudaFreeHost@PLT movq 72(%rsp), %rdi call cudaFree@PLT movq 80(%rsp), %rdi call cudaFree@PLT subl $1, %ebx je .L39 .L16: leaq 56(%rsp), %rdi movl $0, %edx movl $8000000, %esi call cudaHostAlloc@PLT movl %eax, %ebp testl %eax, %eax jne .L40 leaq 64(%rsp), %rdi movl $0, %edx movl $8000000, %esi call cudaHostAlloc@PLT movl %eax, %ebp testl %eax, %eax jne .L41 leaq 72(%rsp), %rdi movl $8000000, %esi call cudaMalloc@PLT movl %eax, %ebp testl %eax, %eax jne .L42 leaq 80(%rsp), %rdi movl $8000000, %esi call cudaMalloc@PLT movl %eax, %ebp testl %eax, %eax jne .L43 movl $0, %ebp .L20: call rand@PLT pxor %xmm0, %xmm0 cvtsi2sdl %eax, %xmm0 divsd .LC7(%rip), %xmm0 movq 56(%rsp), %rax movsd %xmm0, (%rax,%rbp) addq $8, %rbp cmpq $8000000, %rbp jne .L20 cmpl $7812, %r12d jle .L44 call cudaDeviceSynchronize@PLT movl %eax, %ebp testl %eax, %eax jne .L45 leaq 112(%rsp), %rdi movl $0, %esi call gettimeofday@PLT movl $1, %ecx movl $8000000, %edx movq 56(%rsp), %rsi movq 72(%rsp), %rdi call cudaMemcpy@PLT movl %eax, %ebp testl %eax, %eax jne .L46 call cudaDeviceSynchronize@PLT movl %eax, %ebp testl %eax, %eax jne .L47 leaq 160(%rsp), %rdi movl $0, %esi call gettimeofday@PLT movl $128, 100(%rsp) movl $1, 104(%rsp) movl $1, 108(%rsp) movl $7813, 88(%rsp) movl $1, 92(%rsp) movl $1, 96(%rsp) movl $0, %r9d movl $0, %r8d movq 100(%rsp), %rdx movl $1, %ecx movq 88(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax jne .L25 movq 80(%rsp), %rsi movq 72(%rsp), %rdi call _Z29__device_stub__Z8MyKernelPdS_PdS_ jmp .L25 .L36: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L37: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L38: movl %eax, %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %ebp, %edi call exit@PLT .L39: movq 1208(%rsp), %rax subq %fs:40, %rax jne .L48 movl $0, %eax addq $1224, %rsp .cfi_remember_state .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .L48: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE2071: .size main, .-main .section .rodata.str1.1 .LC12: .string "_Z8MyKernelPdS_" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2099: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC12(%rip), %rdx movq %rdx, %rcx leaq _Z8MyKernelPdS_(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2099: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst8,"aM",@progbits,8 .align 8 .LC0: .long 0 .long 1093567616 .align 8 .LC7: .long -4194304 .long 1105199103 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "staged.hip" .section .rodata.cst8,"aM",@progbits,8 .p2align 3, 0x0 # -- Begin function _Z16timeval_subtractPdP7timevalS1_ .LCPI0_0: .quad 0x412e848000000000 # double 1.0E+6 .text .globl _Z16timeval_subtractPdP7timevalS1_ .p2align 4, 0x90 .type _Z16timeval_subtractPdP7timevalS1_,@function _Z16timeval_subtractPdP7timevalS1_: # @_Z16timeval_subtractPdP7timevalS1_ .cfi_startproc # %bb.0: movq %rdx, %rcx movq 8(%rdx), %r9 movq %r9, %rax movabsq $4835703278458516699, %r8 # imm = 0x431BDE82D7B634DB subq 8(%rsi), %rax jle .LBB0_2 # %bb.1: imulq %r8 movq %rdx, %rax shrq $18, %rax shrq $63, %rdx addl %edx, %eax incl %eax imull $1000000, %eax, %edx # imm = 0xF4240 movslq %edx, %rdx subq %rdx, %r9 movq %r9, 8(%rcx) cltq addq %rax, (%rcx) .LBB0_2: movq 8(%rsi), %rdx movq 8(%rcx), %r9 movq %rdx, %rax subq %r9, %rax cmpq $1000001, %rax # imm = 0xF4241 jl .LBB0_4 # %bb.3: movq %r9, %rax subq %rdx, %rax imulq %r8 movq %rdx, %rax shrq $63, %rax sarq $18, %rdx addq %rax, %rdx movabsq $4294967296000000, %rax # imm = 0xF424000000000 imulq %rdx, %rax sarq $32, %rax addq %r9, %rax movq %rax, 8(%rcx) movslq %edx, %rax subq %rax, (%rcx) .LBB0_4: movq 8(%rsi), %rax subq 8(%rcx), %rax cvtsi2sd %rax, %xmm0 movq (%rsi), %rdx divsd .LCPI0_0(%rip), %xmm0 xorl %eax, %eax subq (%rcx), %rdx cvtsi2sd %rdx, %xmm1 addsd %xmm0, %xmm1 movsd %xmm1, (%rdi) setl %al retq .Lfunc_end0: .size _Z16timeval_subtractPdP7timevalS1_, .Lfunc_end0-_Z16timeval_subtractPdP7timevalS1_ .cfi_endproc # -- End function .globl _Z23__device_stub__MyKernelPdS_ # -- Begin function _Z23__device_stub__MyKernelPdS_ .p2align 4, 0x90 .type _Z23__device_stub__MyKernelPdS_,@function _Z23__device_stub__MyKernelPdS_: # @_Z23__device_stub__MyKernelPdS_ .cfi_startproc # %bb.0: subq $88, %rsp .cfi_def_cfa_offset 96 movq %rdi, 56(%rsp) movq %rsi, 48(%rsp) leaq 56(%rsp), %rax movq %rax, 64(%rsp) leaq 48(%rsp), %rax movq %rax, 72(%rsp) leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 64(%rsp), %r9 movl $_Z8MyKernelPdS_, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $104, %rsp .cfi_adjust_cfa_offset -104 retq .Lfunc_end1: .size _Z23__device_stub__MyKernelPdS_, .Lfunc_end1-_Z23__device_stub__MyKernelPdS_ .cfi_endproc # -- End function .section .rodata.cst8,"aM",@progbits,8 .p2align 3, 0x0 # -- Begin function main .LCPI2_0: .quad 0x41dfffffffc00000 # double 2147483647 .LCPI2_1: .quad 0x412e848000000000 # double 1.0E+6 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $1672, %rsp # imm = 0x688 .cfi_def_cfa_offset 1728 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 leaq 20(%rsp), %rdi callq hipGetDevice leaq 16(%rsp), %rdi callq hipGetDeviceCount testl %eax, %eax jne .LBB2_36 # %bb.1: cmpl $0, 16(%rsp) je .LBB2_36 # %bb.2: movabsq $4294967424, %rbx # imm = 0x100000080 movl 20(%rsp), %esi leaq 192(%rsp), %r14 movq %r14, %rdi callq hipGetDevicePropertiesR0600 movl 16(%rsp), %esi movl 20(%rsp), %edx xorl %r15d, %r15d movl $.L.str.1, %edi xorl %eax, %eax callq printf movl $.L.str.2, %edi movq %r14, %rsi xorl %eax, %eax callq printf movl 552(%rsp), %esi movl 556(%rsp), %edx movl $.L.str.3, %edi xorl %eax, %eax callq printf movl 528(%rsp), %eax movl %eax, 12(%rsp) # 4-byte Spill leaq 7685(%rbx), %rax movq %rax, 72(%rsp) # 8-byte Spill jmp .LBB2_3 .p2align 4, 0x90 .LBB2_33: # in Loop: Header=BB2_3 Depth=1 movsd %xmm1, 64(%rsp) # 8-byte Spill movq %rcx, %rax subq %r14, %rax movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $63, %rax sarq $18, %rdx addq %rax, %rdx movslq %edx, %rax movabsq $4294967296000000, %r8 # imm = 0xF424000000000 imulq %r8, %rdx sarq $32, %rdx addq %rdx, %rcx subq %rax, %rdi .LBB2_34: # %_Z16timeval_subtractPdP7timevalS1_.exit66 # in Loop: Header=BB2_3 Depth=1 subq %r13, %rsi xorps %xmm1, %xmm1 cvtsi2sd %rsi, %xmm1 addsd %xmm1, %xmm0 subq %rdi, %rbp subq %rcx, %r14 cvtsi2sd %r14, %xmm2 divsd .LCPI2_1(%rip), %xmm2 xorps %xmm1, %xmm1 cvtsi2sd %rbp, %xmm1 addsd %xmm2, %xmm1 movl $.L.str.6, %edi movb $2, %al callq printf movl $.L.str.7, %edi movsd 64(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero movb $1, %al callq printf movl $.L.str.8, %edi movsd 24(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero movb $1, %al callq printf movq 56(%rsp), %rdi callq hipHostFree movq 48(%rsp), %rdi callq hipHostFree movq 40(%rsp), %rdi callq hipFree movq 32(%rsp), %rdi callq hipFree incl %r15d cmpl $100, %r15d je .LBB2_35 .LBB2_3: # =>This Loop Header: Depth=1 # Child Loop BB2_8 Depth 2 # Child Loop BB2_23 Depth 2 movl $8000000, %esi # imm = 0x7A1200 leaq 56(%rsp), %rdi xorl %edx, %edx callq hipHostMalloc testl %eax, %eax jne .LBB2_37 # %bb.4: # in Loop: Header=BB2_3 Depth=1 movl $8000000, %esi # imm = 0x7A1200 leaq 48(%rsp), %rdi xorl %edx, %edx callq hipHostMalloc testl %eax, %eax jne .LBB2_38 # %bb.5: # in Loop: Header=BB2_3 Depth=1 movl $8000000, %esi # imm = 0x7A1200 leaq 40(%rsp), %rdi callq hipMalloc testl %eax, %eax jne .LBB2_38 # %bb.6: # in Loop: Header=BB2_3 Depth=1 movl $8000000, %esi # imm = 0x7A1200 leaq 32(%rsp), %rdi callq hipMalloc testl %eax, %eax jne .LBB2_38 # %bb.7: # %.preheader95.preheader # in Loop: Header=BB2_3 Depth=1 xorl %ebx, %ebx .p2align 4, 0x90 .LBB2_8: # %.preheader95 # Parent Loop BB2_3 Depth=1 # => This Inner Loop Header: Depth=2 callq rand xorps %xmm0, %xmm0 cvtsi2sd %eax, %xmm0 divsd .LCPI2_0(%rip), %xmm0 movq 56(%rsp), %rax movsd %xmm0, (%rax,%rbx,8) incq %rbx cmpq $1000000, %rbx # imm = 0xF4240 jne .LBB2_8 # %bb.9: # in Loop: Header=BB2_3 Depth=1 cmpl $7813, 12(%rsp) # 4-byte Folded Reload # imm = 0x1E85 jl .LBB2_39 # %bb.10: # in Loop: Header=BB2_3 Depth=1 callq hipDeviceSynchronize testl %eax, %eax jne .LBB2_38 # %bb.11: # in Loop: Header=BB2_3 Depth=1 leaq 176(%rsp), %rdi xorl %esi, %esi callq gettimeofday movq 40(%rsp), %rdi movq 56(%rsp), %rsi movl $8000000, %edx # imm = 0x7A1200 movl $1, %ecx callq hipMemcpy testl %eax, %eax jne .LBB2_38 # %bb.12: # in Loop: Header=BB2_3 Depth=1 callq hipDeviceSynchronize testl %eax, %eax jne .LBB2_38 # %bb.13: # in Loop: Header=BB2_3 Depth=1 leaq 144(%rsp), %rdi xorl %esi, %esi callq gettimeofday movq 72(%rsp), %rdi # 8-byte Reload movl $1, %esi movabsq $4294967424, %rdx # imm = 0x100000080 movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB2_15 # %bb.14: # in Loop: Header=BB2_3 Depth=1 movq 40(%rsp), %rax movq 32(%rsp), %rcx movq %rax, 136(%rsp) movq %rcx, 128(%rsp) leaq 136(%rsp), %rax movq %rax, 192(%rsp) leaq 128(%rsp), %rax movq %rax, 200(%rsp) leaq 112(%rsp), %rdi leaq 96(%rsp), %rsi leaq 88(%rsp), %rdx leaq 80(%rsp), %rcx callq __hipPopCallConfiguration movq 112(%rsp), %rsi movl 120(%rsp), %edx movq 96(%rsp), %rcx movl 104(%rsp), %r8d movl $_Z8MyKernelPdS_, %edi leaq 192(%rsp), %r9 pushq 80(%rsp) .cfi_adjust_cfa_offset 8 pushq 96(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB2_15: # in Loop: Header=BB2_3 Depth=1 callq hipDeviceSynchronize testl %eax, %eax jne .LBB2_38 # %bb.16: # in Loop: Header=BB2_3 Depth=1 leaq 160(%rsp), %rdi xorl %esi, %esi callq gettimeofday movq 176(%rsp), %r13 movq 184(%rsp), %r12 movq 168(%rsp), %r14 movq %r12, %rax movq %r13, %rbx movq %r12, %rcx subq %r14, %rax jle .LBB2_18 # %bb.17: # in Loop: Header=BB2_3 Depth=1 movabsq $4835703278458516699, %rcx # imm = 0x431BDE82D7B634DB imulq %rcx movq %rdx, %rax shrq $18, %rax shrq $63, %rdx addl %edx, %eax incl %eax imull $1000000, %eax, %ecx # imm = 0xF4240 movslq %ecx, %rdx movq %r12, %rcx subq %rdx, %rcx movslq %eax, %rbx addq %r13, %rbx .LBB2_18: # in Loop: Header=BB2_3 Depth=1 movq %r14, %rax subq %rcx, %rax cmpq $1000001, %rax # imm = 0xF4241 jl .LBB2_20 # %bb.19: # in Loop: Header=BB2_3 Depth=1 movq %rcx, %rax subq %r14, %rax movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $63, %rax sarq $18, %rdx addq %rax, %rdx movslq %edx, %rax movabsq $4294967296000000, %rsi # imm = 0xF424000000000 imulq %rsi, %rdx sarq $32, %rdx addq %rdx, %rcx subq %rax, %rbx .LBB2_20: # %_Z16timeval_subtractPdP7timevalS1_.exit # in Loop: Header=BB2_3 Depth=1 movq %r14, %rax subq %rcx, %rax xorps %xmm0, %xmm0 cvtsi2sd %rax, %xmm0 movq 160(%rsp), %rbp divsd .LCPI2_1(%rip), %xmm0 movsd %xmm0, 24(%rsp) # 8-byte Spill movq 48(%rsp), %rdi movq 32(%rsp), %rsi movl $8000000, %edx # imm = 0x7A1200 movl $2, %ecx callq hipMemcpy testl %eax, %eax jne .LBB2_38 # %bb.21: # in Loop: Header=BB2_3 Depth=1 callq hipDeviceSynchronize testl %eax, %eax jne .LBB2_38 # %bb.22: # %.preheader # in Loop: Header=BB2_3 Depth=1 movq %rbp, %rax subq %rbx, %rax xorps %xmm0, %xmm0 cvtsi2sd %rax, %xmm0 movsd 24(%rsp), %xmm1 # 8-byte Reload # xmm1 = mem[0],zero addsd %xmm0, %xmm1 movapd %xmm1, %xmm0 xorpd %xmm1, %xmm1 movq 48(%rsp), %rax xorl %ecx, %ecx .p2align 4, 0x90 .LBB2_23: # Parent Loop BB2_3 Depth=1 # => This Inner Loop Header: Depth=2 addsd (%rax,%rcx,8), %xmm1 incq %rcx cmpq $1000000, %rcx # imm = 0xF4240 jne .LBB2_23 # %bb.24: # in Loop: Header=BB2_3 Depth=1 movq 152(%rsp), %rcx movq %r12, %rax subq %rcx, %rax jle .LBB2_26 # %bb.25: # in Loop: Header=BB2_3 Depth=1 movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $18, %rax shrq $63, %rdx addl %edx, %eax incl %eax imull $1000000, %eax, %edx # imm = 0xF4240 movslq %edx, %rdx subq %rdx, %r12 cltq addq %rax, %r13 .LBB2_26: # in Loop: Header=BB2_3 Depth=1 movq %rcx, %rax subq %r12, %rax cmpq $1000001, %rax # imm = 0xF4241 movsd %xmm0, 24(%rsp) # 8-byte Spill jl .LBB2_28 # %bb.27: # in Loop: Header=BB2_3 Depth=1 movq %r12, %rax subq %rcx, %rax movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $63, %rax sarq $18, %rdx addq %rax, %rdx movslq %edx, %rax movabsq $4294967296000000, %rsi # imm = 0xF424000000000 imulq %rsi, %rdx sarq $32, %rdx addq %rdx, %r12 subq %rax, %r13 .LBB2_28: # %_Z16timeval_subtractPdP7timevalS1_.exit63 # in Loop: Header=BB2_3 Depth=1 movq %rcx, %rax subq %r12, %rax xorps %xmm0, %xmm0 cvtsi2sd %rax, %xmm0 movq 144(%rsp), %rsi divsd .LCPI2_1(%rip), %xmm0 movq %rcx, %rax subq %r14, %rax jle .LBB2_29 # %bb.30: # in Loop: Header=BB2_3 Depth=1 movabsq $4835703278458516699, %rdx # imm = 0x431BDE82D7B634DB imulq %rdx movq %rdx, %rax shrq $18, %rax shrq $63, %rdx addl %edx, %eax incl %eax imull $1000000, %eax, %edx # imm = 0xF4240 movslq %edx, %rdx subq %rdx, %rcx movslq %eax, %rdi addq %rsi, %rdi jmp .LBB2_31 .p2align 4, 0x90 .LBB2_29: # in Loop: Header=BB2_3 Depth=1 movq %rsi, %rdi .LBB2_31: # in Loop: Header=BB2_3 Depth=1 movq %r14, %rax subq %rcx, %rax cmpq $1000001, %rax # imm = 0xF4241 jge .LBB2_33 # %bb.32: # in Loop: Header=BB2_3 Depth=1 movsd %xmm1, 64(%rsp) # 8-byte Spill jmp .LBB2_34 .LBB2_35: xorl %eax, %eax addq $1672, %rsp # imm = 0x688 .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .LBB2_38: .cfi_def_cfa_offset 1728 movl $.L.str.4, %edi movl %eax, %esi movl %eax, %ebx xorl %eax, %eax callq printf movl %ebx, %edi callq exit .LBB2_39: movl $.L.str.5, %edi movl $7813, %esi # imm = 0x1E85 movl 12(%rsp), %edx # 4-byte Reload xorl %eax, %eax callq printf movl $1, %edi callq exit .LBB2_37: movl %eax, %r14d movl $.L.str.4, %edi movl %eax, %esi xorl %eax, %eax callq printf movl %r14d, %edi callq exit .LBB2_36: movl $.Lstr, %edi callq puts@PLT movl $1, %edi callq exit .Lfunc_end2: .size main, .Lfunc_end2-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB3_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB3_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z8MyKernelPdS_, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end3: .size __hip_module_ctor, .Lfunc_end3-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB4_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB4_2: retq .Lfunc_end4: .size __hip_module_dtor, .Lfunc_end4-__hip_module_dtor .cfi_endproc # -- End function .type _Z8MyKernelPdS_,@object # @_Z8MyKernelPdS_ .section .rodata,"a",@progbits .globl _Z8MyKernelPdS_ .p2align 3, 0x0 _Z8MyKernelPdS_: .quad _Z23__device_stub__MyKernelPdS_ .size _Z8MyKernelPdS_, 8 .type .L.str.1,@object # @.str.1 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.1: .asciz "Device count, devid: %d %d\n" .size .L.str.1, 28 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "Device: %s\n" .size .L.str.2, 12 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "[deviceProp.major.deviceProp.minor] = [%d.%d]\n\n" .size .L.str.3, 48 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "Error %d\n" .size .L.str.4, 10 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "Nblocks > Max_gridsize! %d %d\n" .size .L.str.5, 33 .type .L.str.6,@object # @.str.6 .L.str.6: .asciz "Individual timings: %e %e\n" .size .L.str.6, 27 .type .L.str.7,@object # @.str.7 .L.str.7: .asciz "Result: %e\n\n" .size .L.str.7, 13 .type .L.str.8,@object # @.str.8 .L.str.8: .asciz "Time: %e\n" .size .L.str.8, 10 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z8MyKernelPdS_" .size .L__unnamed_1, 16 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "No CUDA devices!" .size .Lstr, 17 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z23__device_stub__MyKernelPdS_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z8MyKernelPdS_ .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
#include <iostream> #define N 10000 float *a_cpu, *b_cpu, *c_cpu; float *a_gpu, *b_gpu, *c_gpu; __global__ void add(float *a_gpu, float *b_gpu, float *c_gpu, int n){ for(int i = 0; i < n; i++){ c_gpu[i] = a_gpu[i] + b_gpu[i]; } } int main() { a_cpu = (float*)malloc(sizeof(float)*N); b_cpu = (float*)malloc(sizeof(float)*N); c_cpu = (float*)malloc(sizeof(float)*N); cudaMalloc((float**)&a_gpu, sizeof(float) * N); cudaMalloc((float**)&b_gpu, sizeof(float) * N); cudaMalloc((float**)&c_gpu, sizeof(float) * N); for (int i = 1; i <= N; i++){ a_cpu[i] = -i; b_cpu[i] = i*2; c_cpu[i] = a_cpu[i] + b_cpu[i]; } cudaMemcpy(a_gpu, a_cpu, sizeof(float) * N, cudaMemcpyHostToDevice); cudaMemcpy(b_gpu, b_cpu, sizeof(float) * N, cudaMemcpyHostToDevice); cudaMemcpy(c_gpu, c_cpu, sizeof(float) * N, cudaMemcpyHostToDevice); add<<<1,10>>>(a_gpu, b_gpu, c_gpu, N); cudaMemcpy(c_cpu, c_gpu, sizeof(float) * N, cudaMemcpyDeviceToHost); for (int i = 1; i <= N; i++){ printf("%.0f\n", c_cpu[i]); } cudaFree(a_gpu); cudaFree(b_gpu); cudaFree(c_gpu); free(a_cpu); free(b_cpu); free(c_cpu); getchar(); return 0; }
code for sm_80 Function : _Z3addPfS_S_i .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ IMAD.MOV.U32 R1, RZ, RZ, c[0x0][0x28] ; /* 0x00000a00ff017624 */ /* 0x000fe400078e00ff */ /*0010*/ IMAD.MOV.U32 R0, RZ, RZ, c[0x0][0x178] ; /* 0x00005e00ff007624 */ /* 0x000fca00078e00ff */ /*0020*/ ISETP.GE.AND P0, PT, R0, 0x1, PT ; /* 0x000000010000780c */ /* 0x000fda0003f06270 */ /*0030*/ @!P0 EXIT ; /* 0x000000000000894d */ /* 0x000fea0003800000 */ /*0040*/ IADD3 R2, R0.reuse, -0x1, RZ ; /* 0xffffffff00027810 */ /* 0x040fe20007ffe0ff */ /*0050*/ UMOV UR4, URZ ; /* 0x0000003f00047c82 */ /* 0x000fe20008000000 */ /*0060*/ LOP3.LUT R0, R0, 0x3, RZ, 0xc0, !PT ; /* 0x0000000300007812 */ /* 0x000fe200078ec0ff */ /*0070*/ ULDC.64 UR12, c[0x0][0x118] ; /* 0x00004600000c7ab9 */ /* 0x000fe20000000a00 */ /*0080*/ ISETP.GE.U32.AND P0, PT, R2, 0x3, PT ; /* 0x000000030200780c */ /* 0x000fda0003f06070 */ /*0090*/ @!P0 BRA 0xb80 ; /* 0x00000ae000008947 */ /* 0x000fea0003800000 */ /*00a0*/ IADD3 R8, -R0, c[0x0][0x178], RZ ; /* 0x00005e0000087a10 */ /* 0x000fe20007ffe1ff */ /*00b0*/ UMOV UR4, URZ ; /* 0x0000003f00047c82 */ /* 0x000fe20008000000 */ /*00c0*/ MOV R4, c[0x0][0x170] ; /* 0x00005c0000047a02 */ /* 0x000fe20000000f00 */ /*00d0*/ IMAD.MOV.U32 R5, RZ, RZ, c[0x0][0x174] ; /* 0x00005d00ff057624 */ /* 0x000fe200078e00ff */ /*00e0*/ ISETP.GT.AND P0, PT, R8, RZ, PT ; /* 0x000000ff0800720c */ /* 0x000fe20003f04270 */ /*00f0*/ IMAD.MOV.U32 R3, RZ, RZ, c[0x0][0x16c] ; /* 0x00005b00ff037624 */ /* 0x000fe200078e00ff */ /*0100*/ MOV R2, c[0x0][0x168] ; /* 0x00005a0000027a02 */ /* 0x000fe20000000f00 */ /*0110*/ IMAD.MOV.U32 R7, RZ, RZ, c[0x0][0x164] ; /* 0x00005900ff077624 */ /* 0x000fe200078e00ff */ /*0120*/ MOV R6, c[0x0][0x160] ; /* 0x0000580000067a02 */ /* 0x000fd20000000f00 */ /*0130*/ @!P0 BRA 0x990 ; /* 0x0000085000008947 */ /* 0x000fea0003800000 */ /*0140*/ ISETP.GT.AND P1, PT, R8, 0xc, PT ; /* 0x0000000c0800780c */ /* 0x000fe40003f24270 */ /*0150*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x80, 0x0 ; /* 0x000000000000781c */ /* 0x000fd60003f0f070 */ /*0160*/ @!P1 BRA 0x670 ; /* 0x0000050000009947 */ /* 0x000fea0003800000 */ /*0170*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe40003f0e170 */ /*0180*/ LDG.E R9, [R2.64] ; /* 0x0000000c02097981 */ /* 0x000ea8000c1e1900 */ /*0190*/ LDG.E R10, [R6.64] ; /* 0x0000000c060a7981 */ /* 0x000ea4000c1e1900 */ /*01a0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x004fca0000000000 */ /*01b0*/ STG.E [R4.64], R9 ; /* 0x0000000904007986 */ /* 0x0001e8000c10190c */ /*01c0*/ LDG.E R10, [R2.64+0x4] ; /* 0x0000040c020a7981 */ /* 0x000ea8000c1e1900 */ /*01d0*/ LDG.E R11, [R6.64+0x4] ; /* 0x0000040c060b7981 */ /* 0x000ea4000c1e1900 */ /*01e0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x004fca0000000000 */ /*01f0*/ STG.E [R4.64+0x4], R11 ; /* 0x0000040b04007986 */ /* 0x0003e8000c10190c */ /*0200*/ LDG.E R10, [R2.64+0x8] ; /* 0x0000080c020a7981 */ /* 0x000ea8000c1e1900 */ /*0210*/ LDG.E R13, [R6.64+0x8] ; /* 0x0000080c060d7981 */ /* 0x000ea4000c1e1900 */ /*0220*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x004fca0000000000 */ /*0230*/ STG.E [R4.64+0x8], R13 ; /* 0x0000080d04007986 */ /* 0x0005e8000c10190c */ /*0240*/ LDG.E R10, [R2.64+0xc] ; /* 0x00000c0c020a7981 */ /* 0x000ee8000c1e1900 */ /*0250*/ LDG.E R15, [R6.64+0xc] ; /* 0x00000c0c060f7981 */ /* 0x000ee4000c1e1900 */ /*0260*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x008fca0000000000 */ /*0270*/ STG.E [R4.64+0xc], R15 ; /* 0x00000c0f04007986 */ /* 0x0007e8000c10190c */ /*0280*/ LDG.E R9, [R2.64+0x10] ; /* 0x0000100c02097981 */ /* 0x001f28000c1e1900 */ /*0290*/ LDG.E R10, [R6.64+0x10] ; /* 0x0000100c060a7981 */ /* 0x000f24000c1e1900 */ /*02a0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x010fca0000000000 */ /*02b0*/ STG.E [R4.64+0x10], R9 ; /* 0x0000100904007986 */ /* 0x0001e8000c10190c */ /*02c0*/ LDG.E R10, [R2.64+0x14] ; /* 0x0000140c020a7981 */ /* 0x000f28000c1e1900 */ /*02d0*/ LDG.E R11, [R6.64+0x14] ; /* 0x0000140c060b7981 */ /* 0x002f24000c1e1900 */ /*02e0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x010fca0000000000 */ /*02f0*/ STG.E [R4.64+0x14], R11 ; /* 0x0000140b04007986 */ /* 0x0003e8000c10190c */ /*0300*/ LDG.E R10, [R2.64+0x18] ; /* 0x0000180c020a7981 */ /* 0x000f28000c1e1900 */ /*0310*/ LDG.E R13, [R6.64+0x18] ; /* 0x0000180c060d7981 */ /* 0x004f24000c1e1900 */ /*0320*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x010fca0000000000 */ /*0330*/ STG.E [R4.64+0x18], R13 ; /* 0x0000180d04007986 */ /* 0x0005e8000c10190c */ /*0340*/ LDG.E R10, [R2.64+0x1c] ; /* 0x00001c0c020a7981 */ /* 0x000f28000c1e1900 */ /*0350*/ LDG.E R15, [R6.64+0x1c] ; /* 0x00001c0c060f7981 */ /* 0x008f24000c1e1900 */ /*0360*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x010fca0000000000 */ /*0370*/ STG.E [R4.64+0x1c], R15 ; /* 0x00001c0f04007986 */ /* 0x0007e8000c10190c */ /*0380*/ LDG.E R9, [R2.64+0x20] ; /* 0x0000200c02097981 */ /* 0x001f28000c1e1900 */ /*0390*/ LDG.E R10, [R6.64+0x20] ; /* 0x0000200c060a7981 */ /* 0x000f24000c1e1900 */ /*03a0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x010fca0000000000 */ /*03b0*/ STG.E [R4.64+0x20], R9 ; /* 0x0000200904007986 */ /* 0x0001e8000c10190c */ /*03c0*/ LDG.E R10, [R2.64+0x24] ; /* 0x0000240c020a7981 */ /* 0x000f28000c1e1900 */ /*03d0*/ LDG.E R11, [R6.64+0x24] ; /* 0x0000240c060b7981 */ /* 0x002f24000c1e1900 */ /*03e0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x010fca0000000000 */ /*03f0*/ STG.E [R4.64+0x24], R11 ; /* 0x0000240b04007986 */ /* 0x0003e8000c10190c */ /*0400*/ LDG.E R10, [R2.64+0x28] ; /* 0x0000280c020a7981 */ /* 0x000f28000c1e1900 */ /*0410*/ LDG.E R13, [R6.64+0x28] ; /* 0x0000280c060d7981 */ /* 0x004f24000c1e1900 */ /*0420*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x010fca0000000000 */ /*0430*/ STG.E [R4.64+0x28], R13 ; /* 0x0000280d04007986 */ /* 0x0005e8000c10190c */ /*0440*/ LDG.E R10, [R2.64+0x2c] ; /* 0x00002c0c020a7981 */ /* 0x000f28000c1e1900 */ /*0450*/ LDG.E R15, [R6.64+0x2c] ; /* 0x00002c0c060f7981 */ /* 0x008f24000c1e1900 */ /*0460*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x010fca0000000000 */ /*0470*/ STG.E [R4.64+0x2c], R15 ; /* 0x00002c0f04007986 */ /* 0x0007e8000c10190c */ /*0480*/ LDG.E R9, [R2.64+0x30] ; /* 0x0000300c02097981 */ /* 0x001f28000c1e1900 */ /*0490*/ LDG.E R10, [R6.64+0x30] ; /* 0x0000300c060a7981 */ /* 0x000f24000c1e1900 */ /*04a0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x010fca0000000000 */ /*04b0*/ STG.E [R4.64+0x30], R9 ; /* 0x0000300904007986 */ /* 0x000fe8000c10190c */ /*04c0*/ LDG.E R10, [R2.64+0x34] ; /* 0x0000340c020a7981 */ /* 0x000f28000c1e1900 */ /*04d0*/ LDG.E R11, [R6.64+0x34] ; /* 0x0000340c060b7981 */ /* 0x002f24000c1e1900 */ /*04e0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x010fca0000000000 */ /*04f0*/ STG.E [R4.64+0x34], R11 ; /* 0x0000340b04007986 */ /* 0x0001e8000c10190c */ /*0500*/ LDG.E R10, [R2.64+0x38] ; /* 0x0000380c020a7981 */ /* 0x000f28000c1e1900 */ /*0510*/ LDG.E R13, [R6.64+0x38] ; /* 0x0000380c060d7981 */ /* 0x004f22000c1e1900 */ /*0520*/ IADD3 R12, P1, R2, 0x40, RZ ; /* 0x00000040020c7810 */ /* 0x000fe40007f3e0ff */ /*0530*/ IADD3 R8, R8, -0x10, RZ ; /* 0xfffffff008087810 */ /* 0x000fe20007ffe0ff */ /*0540*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x010fca0000000000 */ /*0550*/ STG.E [R4.64+0x38], R13 ; /* 0x0000380d04007986 */ /* 0x000fe8000c10190c */ /*0560*/ LDG.E R10, [R2.64+0x3c] ; /* 0x00003c0c020a7981 */ /* 0x0002a8000c1e1900 */ /*0570*/ LDG.E R15, [R6.64+0x3c] ; /* 0x00003c0c060f7981 */ /* 0x0086a2000c1e1900 */ /*0580*/ IADD3.X R11, RZ, R3, RZ, P1, !PT ; /* 0x00000003ff0b7210 */ /* 0x001fe20000ffe4ff */ /*0590*/ UIADD3 UR4, UR4, 0x10, URZ ; /* 0x0000001004047890 */ /* 0x000fe2000fffe03f */ /*05a0*/ ISETP.GT.AND P1, PT, R8, 0xc, PT ; /* 0x0000000c0800780c */ /* 0x000fc40003f24270 */ /*05b0*/ IADD3 R9, P3, R4, 0x40, RZ ; /* 0x0000004004097810 */ /* 0x000fe20007f7e0ff */ /*05c0*/ IMAD.MOV.U32 R2, RZ, RZ, R12 ; /* 0x000000ffff027224 */ /* 0x002fe200078e000c */ /*05d0*/ IADD3 R14, P2, R6, 0x40, RZ ; /* 0x00000040060e7810 */ /* 0x000fe40007f5e0ff */ /*05e0*/ MOV R3, R11 ; /* 0x0000000b00037202 */ /* 0x000fc60000000f00 */ /*05f0*/ IMAD.X R7, RZ, RZ, R7, P2 ; /* 0x000000ffff077224 */ /* 0x008fe400010e0607 */ /*0600*/ IMAD.MOV.U32 R6, RZ, RZ, R14 ; /* 0x000000ffff067224 */ /* 0x000fe400078e000e */ /*0610*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x004fe20000000000 */ /*0620*/ IADD3.X R10, RZ, R5, RZ, P3, !PT ; /* 0x00000005ff0a7210 */ /* 0x000fc80001ffe4ff */ /*0630*/ STG.E [R4.64+0x3c], R15 ; /* 0x00003c0f04007986 */ /* 0x0001e4000c10190c */ /*0640*/ MOV R4, R9 ; /* 0x0000000900047202 */ /* 0x001fe20000000f00 */ /*0650*/ IMAD.MOV.U32 R5, RZ, RZ, R10 ; /* 0x000000ffff057224 */ /* 0x000fe200078e000a */ /*0660*/ @P1 BRA 0x180 ; /* 0xfffffb1000001947 */ /* 0x000fea000383ffff */ /*0670*/ ISETP.GT.AND P1, PT, R8, 0x4, PT ; /* 0x000000040800780c */ /* 0x000fda0003f24270 */ /*0680*/ @!P1 BRA 0x970 ; /* 0x000002e000009947 */ /* 0x000fea0003800000 */ /*0690*/ LDG.E R9, [R2.64] ; /* 0x0000000c02097981 */ /* 0x000ea8000c1e1900 */ /*06a0*/ LDG.E R10, [R6.64] ; /* 0x0000000c060a7981 */ /* 0x000ea4000c1e1900 */ /*06b0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x004fca0000000000 */ /*06c0*/ STG.E [R4.64], R9 ; /* 0x0000000904007986 */ /* 0x0001e8000c10190c */ /*06d0*/ LDG.E R10, [R2.64+0x4] ; /* 0x0000040c020a7981 */ /* 0x000ea8000c1e1900 */ /*06e0*/ LDG.E R11, [R6.64+0x4] ; /* 0x0000040c060b7981 */ /* 0x000ea4000c1e1900 */ /*06f0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x004fca0000000000 */ /*0700*/ STG.E [R4.64+0x4], R11 ; /* 0x0000040b04007986 */ /* 0x0003e8000c10190c */ /*0710*/ LDG.E R10, [R2.64+0x8] ; /* 0x0000080c020a7981 */ /* 0x000ea8000c1e1900 */ /*0720*/ LDG.E R13, [R6.64+0x8] ; /* 0x0000080c060d7981 */ /* 0x000ea4000c1e1900 */ /*0730*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x004fca0000000000 */ /*0740*/ STG.E [R4.64+0x8], R13 ; /* 0x0000080d04007986 */ /* 0x0005e8000c10190c */ /*0750*/ LDG.E R10, [R2.64+0xc] ; /* 0x00000c0c020a7981 */ /* 0x000ee8000c1e1900 */ /*0760*/ LDG.E R15, [R6.64+0xc] ; /* 0x00000c0c060f7981 */ /* 0x000ee4000c1e1900 */ /*0770*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x008fca0000000000 */ /*0780*/ STG.E [R4.64+0xc], R15 ; /* 0x00000c0f04007986 */ /* 0x0007e8000c10190c */ /*0790*/ LDG.E R9, [R2.64+0x10] ; /* 0x0000100c02097981 */ /* 0x001f28000c1e1900 */ /*07a0*/ LDG.E R10, [R6.64+0x10] ; /* 0x0000100c060a7981 */ /* 0x000f24000c1e1900 */ /*07b0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x010fca0000000000 */ /*07c0*/ STG.E [R4.64+0x10], R9 ; /* 0x0000100904007986 */ /* 0x000fe8000c10190c */ /*07d0*/ LDG.E R10, [R2.64+0x14] ; /* 0x0000140c020a7981 */ /* 0x000f28000c1e1900 */ /*07e0*/ LDG.E R11, [R6.64+0x14] ; /* 0x0000140c060b7981 */ /* 0x002f24000c1e1900 */ /*07f0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x010fca0000000000 */ /*0800*/ STG.E [R4.64+0x14], R11 ; /* 0x0000140b04007986 */ /* 0x0001e8000c10190c */ /*0810*/ LDG.E R10, [R2.64+0x18] ; /* 0x0000180c020a7981 */ /* 0x000f28000c1e1900 */ /*0820*/ LDG.E R13, [R6.64+0x18] ; /* 0x0000180c060d7981 */ /* 0x004f24000c1e1900 */ /*0830*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x010fca0000000000 */ /*0840*/ STG.E [R4.64+0x18], R13 ; /* 0x0000180d04007986 */ /* 0x000fe8000c10190c */ /*0850*/ LDG.E R10, [R2.64+0x1c] ; /* 0x00001c0c020a7981 */ /* 0x0002a8000c1e1900 */ /*0860*/ LDG.E R15, [R6.64+0x1c] ; /* 0x00001c0c060f7981 */ /* 0x0086a2000c1e1900 */ /*0870*/ IADD3 R11, P2, R2, 0x20, RZ ; /* 0x00000020020b7810 */ /* 0x001fe20007f5e0ff */ /*0880*/ UIADD3 UR4, UR4, 0x8, URZ ; /* 0x0000000804047890 */ /* 0x000fe2000fffe03f */ /*0890*/ IADD3 R9, P3, R4, 0x20, RZ ; /* 0x0000002004097810 */ /* 0x000fc40007f7e0ff */ /*08a0*/ IADD3 R14, P1, R6, 0x20, RZ ; /* 0x00000020060e7810 */ /* 0x000fe20007f3e0ff */ /*08b0*/ IMAD.X R12, RZ, RZ, R3, P2 ; /* 0x000000ffff0c7224 */ /* 0x000fe200010e0603 */ /*08c0*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe20003f0e170 */ /*08d0*/ IMAD.MOV.U32 R2, RZ, RZ, R11 ; /* 0x000000ffff027224 */ /* 0x002fe200078e000b */ /*08e0*/ IADD3 R8, R8, -0x8, RZ ; /* 0xfffffff808087810 */ /* 0x000fe20007ffe0ff */ /*08f0*/ IMAD.MOV.U32 R6, RZ, RZ, R14 ; /* 0x000000ffff067224 */ /* 0x008fe200078e000e */ /*0900*/ IADD3.X R7, RZ, R7, RZ, P1, !PT ; /* 0x00000007ff077210 */ /* 0x000fe40000ffe4ff */ /*0910*/ MOV R3, R12 ; /* 0x0000000c00037202 */ /* 0x000fe20000000f00 */ /*0920*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x004fe20000000000 */ /*0930*/ IADD3.X R10, RZ, R5, RZ, P3, !PT ; /* 0x00000005ff0a7210 */ /* 0x000fc80001ffe4ff */ /*0940*/ STG.E [R4.64+0x1c], R15 ; /* 0x00001c0f04007986 */ /* 0x0001e4000c10190c */ /*0950*/ MOV R4, R9 ; /* 0x0000000900047202 */ /* 0x001fe20000000f00 */ /*0960*/ IMAD.MOV.U32 R5, RZ, RZ, R10 ; /* 0x000000ffff057224 */ /* 0x000fe400078e000a */ /*0970*/ ISETP.NE.OR P0, PT, R8, RZ, P0 ; /* 0x000000ff0800720c */ /* 0x000fda0000705670 */ /*0980*/ @!P0 BRA 0xb80 ; /* 0x000001f000008947 */ /* 0x000fea0003800000 */ /*0990*/ LDG.E R9, [R2.64] ; /* 0x0000000c02097981 */ /* 0x000ea8000c1e1900 */ /*09a0*/ LDG.E R10, [R6.64] ; /* 0x0000000c060a7981 */ /* 0x000ea4000c1e1900 */ /*09b0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x004fca0000000000 */ /*09c0*/ STG.E [R4.64], R9 ; /* 0x0000000904007986 */ /* 0x000fe8000c10190c */ /*09d0*/ LDG.E R10, [R2.64+0x4] ; /* 0x0000040c020a7981 */ /* 0x000ea8000c1e1900 */ /*09e0*/ LDG.E R11, [R6.64+0x4] ; /* 0x0000040c060b7981 */ /* 0x000ea4000c1e1900 */ /*09f0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x004fca0000000000 */ /*0a00*/ STG.E [R4.64+0x4], R11 ; /* 0x0000040b04007986 */ /* 0x0001e8000c10190c */ /*0a10*/ LDG.E R10, [R2.64+0x8] ; /* 0x0000080c020a7981 */ /* 0x000ea8000c1e1900 */ /*0a20*/ LDG.E R13, [R6.64+0x8] ; /* 0x0000080c060d7981 */ /* 0x000ea2000c1e1900 */ /*0a30*/ IADD3 R12, P0, R2, 0x10, RZ ; /* 0x00000010020c7810 */ /* 0x000fe40007f1e0ff */ /*0a40*/ IADD3 R8, R8, -0x4, RZ ; /* 0xfffffffc08087810 */ /* 0x000fe20007ffe0ff */ /*0a50*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x004fca0000000000 */ /*0a60*/ STG.E [R4.64+0x8], R13 ; /* 0x0000080d04007986 */ /* 0x000fe8000c10190c */ /*0a70*/ LDG.E R10, [R2.64+0xc] ; /* 0x00000c0c020a7981 */ /* 0x0002a8000c1e1900 */ /*0a80*/ LDG.E R15, [R6.64+0xc] ; /* 0x00000c0c060f7981 */ /* 0x0006a2000c1e1900 */ /*0a90*/ IADD3.X R11, RZ, R3, RZ, P0, !PT ; /* 0x00000003ff0b7210 */ /* 0x001fe200007fe4ff */ /*0aa0*/ UIADD3 UR4, UR4, 0x4, URZ ; /* 0x0000000404047890 */ /* 0x000fe2000fffe03f */ /*0ab0*/ ISETP.NE.AND P0, PT, R8, RZ, PT ; /* 0x000000ff0800720c */ /* 0x000fc40003f05270 */ /*0ac0*/ IADD3 R9, P2, R4, 0x10, RZ ; /* 0x0000001004097810 */ /* 0x000fe20007f5e0ff */ /*0ad0*/ IMAD.MOV.U32 R2, RZ, RZ, R12 ; /* 0x000000ffff027224 */ /* 0x002fe200078e000c */ /*0ae0*/ IADD3 R14, P1, R6, 0x10, RZ ; /* 0x00000010060e7810 */ /* 0x000fe40007f3e0ff */ /*0af0*/ MOV R3, R11 ; /* 0x0000000b00037202 */ /* 0x000fc60000000f00 */ /*0b00*/ IMAD.X R7, RZ, RZ, R7, P1 ; /* 0x000000ffff077224 */ /* 0x008fe400008e0607 */ /*0b10*/ IMAD.MOV.U32 R6, RZ, RZ, R14 ; /* 0x000000ffff067224 */ /* 0x000fe400078e000e */ /*0b20*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x004fe20000000000 */ /*0b30*/ IADD3.X R10, RZ, R5, RZ, P2, !PT ; /* 0x00000005ff0a7210 */ /* 0x000fc800017fe4ff */ /*0b40*/ STG.E [R4.64+0xc], R15 ; /* 0x00000c0f04007986 */ /* 0x0001e4000c10190c */ /*0b50*/ MOV R4, R9 ; /* 0x0000000900047202 */ /* 0x001fe20000000f00 */ /*0b60*/ IMAD.MOV.U32 R5, RZ, RZ, R10 ; /* 0x000000ffff057224 */ /* 0x000fe200078e000a */ /*0b70*/ @P0 BRA 0x990 ; /* 0xfffffe1000000947 */ /* 0x000fea000383ffff */ /*0b80*/ ISETP.NE.AND P0, PT, R0, RZ, PT ; /* 0x000000ff0000720c */ /* 0x000fda0003f05270 */ /*0b90*/ @!P0 EXIT ; /* 0x000000000000894d */ /* 0x000fea0003800000 */ /*0ba0*/ UMOV UR5, 0x4 ; /* 0x0000000400057882 */ /* 0x000fe40000000000 */ /*0bb0*/ ULDC.64 UR6, c[0x0][0x170] ; /* 0x00005c0000067ab9 */ /* 0x000fe40000000a00 */ /*0bc0*/ ULDC.64 UR8, c[0x0][0x168] ; /* 0x00005a0000087ab9 */ /* 0x000fe40000000a00 */ /*0bd0*/ ULDC.64 UR10, c[0x0][0x160] ; /* 0x00005800000a7ab9 */ /* 0x000fe40000000a00 */ /*0be0*/ UIMAD.WIDE UR6, UR4, UR5, UR6 ; /* 0x00000005040672a5 */ /* 0x000fe4000f8e0206 */ /*0bf0*/ UIMAD.WIDE UR8, UR4, UR5, UR8 ; /* 0x00000005040872a5 */ /* 0x000fc4000f8e0208 */ /*0c00*/ UIMAD.WIDE UR4, UR4, UR5, UR10 ; /* 0x00000005040472a5 */ /* 0x000fca000f8e020a */ /*0c10*/ MOV R2, UR8 ; /* 0x0000000800027c02 */ /* 0x000fe20008000f00 */ /*0c20*/ IMAD.U32 R5, RZ, RZ, UR5 ; /* 0x00000005ff057e24 */ /* 0x000fe2000f8e00ff */ /*0c30*/ MOV R4, UR4 ; /* 0x0000000400047c02 */ /* 0x000fe20008000f00 */ /*0c40*/ IMAD.U32 R3, RZ, RZ, UR9 ; /* 0x00000009ff037e24 */ /* 0x000fc8000f8e00ff */ /*0c50*/ LDG.E R5, [R4.64] ; /* 0x0000000c04057981 */ /* 0x000ea8000c1e1900 */ /*0c60*/ LDG.E R2, [R2.64] ; /* 0x0000000c02027981 */ /* 0x000ea2000c1e1900 */ /*0c70*/ IADD3 R0, R0, -0x1, RZ ; /* 0xffffffff00007810 */ /* 0x000fc80007ffe0ff */ /*0c80*/ ISETP.NE.AND P0, PT, R0, RZ, PT ; /* 0x000000ff0000720c */ /* 0x000fe20003f05270 */ /*0c90*/ UIADD3 UR8, UP1, UR8, 0x4, URZ ; /* 0x0000000408087890 */ /* 0x000fe2000ff3e03f */ /*0ca0*/ MOV R6, UR6 ; /* 0x0000000600067c02 */ /* 0x001fe20008000f00 */ /*0cb0*/ UIADD3 UR6, UP0, UR6, 0x4, URZ ; /* 0x0000000406067890 */ /* 0x000fe2000ff1e03f */ /*0cc0*/ MOV R7, UR7 ; /* 0x0000000700077c02 */ /* 0x000fe20008000f00 */ /*0cd0*/ UIADD3 UR4, UP2, UR4, 0x4, URZ ; /* 0x0000000404047890 */ /* 0x000fe4000ff5e03f */ /*0ce0*/ UIADD3.X UR7, URZ, UR7, URZ, UP0, !UPT ; /* 0x000000073f077290 */ /* 0x000fe400087fe43f */ /*0cf0*/ UIADD3.X UR9, URZ, UR9, URZ, UP1, !UPT ; /* 0x000000093f097290 */ /* 0x000fe40008ffe43f */ /*0d00*/ UIADD3.X UR5, URZ, UR5, URZ, UP2, !UPT ; /* 0x000000053f057290 */ /* 0x000fe200097fe43f */ /*0d10*/ FADD R9, R2, R5 ; /* 0x0000000502097221 */ /* 0x004fca0000000000 */ /*0d20*/ STG.E [R6.64], R9 ; /* 0x0000000906007986 */ /* 0x0001e2000c10190c */ /*0d30*/ @P0 BRA 0xc10 ; /* 0xfffffed000000947 */ /* 0x000fea000383ffff */ /*0d40*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0d50*/ BRA 0xd50; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0d60*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0d70*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0d80*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0d90*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0da0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0db0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0dc0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0dd0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0de0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0df0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include <iostream> #define N 10000 float *a_cpu, *b_cpu, *c_cpu; float *a_gpu, *b_gpu, *c_gpu; __global__ void add(float *a_gpu, float *b_gpu, float *c_gpu, int n){ for(int i = 0; i < n; i++){ c_gpu[i] = a_gpu[i] + b_gpu[i]; } } int main() { a_cpu = (float*)malloc(sizeof(float)*N); b_cpu = (float*)malloc(sizeof(float)*N); c_cpu = (float*)malloc(sizeof(float)*N); cudaMalloc((float**)&a_gpu, sizeof(float) * N); cudaMalloc((float**)&b_gpu, sizeof(float) * N); cudaMalloc((float**)&c_gpu, sizeof(float) * N); for (int i = 1; i <= N; i++){ a_cpu[i] = -i; b_cpu[i] = i*2; c_cpu[i] = a_cpu[i] + b_cpu[i]; } cudaMemcpy(a_gpu, a_cpu, sizeof(float) * N, cudaMemcpyHostToDevice); cudaMemcpy(b_gpu, b_cpu, sizeof(float) * N, cudaMemcpyHostToDevice); cudaMemcpy(c_gpu, c_cpu, sizeof(float) * N, cudaMemcpyHostToDevice); add<<<1,10>>>(a_gpu, b_gpu, c_gpu, N); cudaMemcpy(c_cpu, c_gpu, sizeof(float) * N, cudaMemcpyDeviceToHost); for (int i = 1; i <= N; i++){ printf("%.0f\n", c_cpu[i]); } cudaFree(a_gpu); cudaFree(b_gpu); cudaFree(c_gpu); free(a_cpu); free(b_cpu); free(c_cpu); getchar(); return 0; }
.file "tmpxft_0018e52c_00000000-6_othermain.cudafe1.cpp" .text #APP .globl _ZSt21ios_base_library_initv #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB3672: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3672: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z27__device_stub__Z3addPfS_S_iPfS_S_i .type _Z27__device_stub__Z3addPfS_S_iPfS_S_i, @function _Z27__device_stub__Z3addPfS_S_iPfS_S_i: .LFB3694: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movl %ecx, 4(%rsp) movq %fs:40, %rax movq %rax, 136(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) leaq 4(%rsp), %rax movq %rax, 120(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 136(%rsp), %rax subq %fs:40, %rax jne .L8 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 168 pushq 40(%rsp) .cfi_def_cfa_offset 176 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z3addPfS_S_i(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE3694: .size _Z27__device_stub__Z3addPfS_S_iPfS_S_i, .-_Z27__device_stub__Z3addPfS_S_iPfS_S_i .globl _Z3addPfS_S_i .type _Z3addPfS_S_i, @function _Z3addPfS_S_i: .LFB3695: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z27__device_stub__Z3addPfS_S_iPfS_S_i addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3695: .size _Z3addPfS_S_i, .-_Z3addPfS_S_i .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "%.0f\n" .text .globl main .type main, @function main: .LFB3669: .cfi_startproc endbr64 pushq %rbp .cfi_def_cfa_offset 16 .cfi_offset 6, -16 pushq %rbx .cfi_def_cfa_offset 24 .cfi_offset 3, -24 subq $40, %rsp .cfi_def_cfa_offset 64 movl $40000, %edi call malloc@PLT movq %rax, a_cpu(%rip) movl $40000, %edi call malloc@PLT movq %rax, b_cpu(%rip) movl $40000, %edi call malloc@PLT movq %rax, c_cpu(%rip) movl $40000, %esi leaq a_gpu(%rip), %rdi call cudaMalloc@PLT movl $40000, %esi leaq b_gpu(%rip), %rdi call cudaMalloc@PLT movl $40000, %esi leaq c_gpu(%rip), %rdi call cudaMalloc@PLT movl $4, %eax movl $1, %edx .L12: movl %edx, %ecx negl %ecx pxor %xmm0, %xmm0 cvtsi2ssl %ecx, %xmm0 movq a_cpu(%rip), %rcx movss %xmm0, (%rcx,%rax) leal (%rdx,%rdx), %ecx pxor %xmm0, %xmm0 cvtsi2ssl %ecx, %xmm0 movq b_cpu(%rip), %rcx movss %xmm0, (%rcx,%rax) movq a_cpu(%rip), %rcx movss (%rcx,%rax), %xmm0 movq b_cpu(%rip), %rcx addss (%rcx,%rax), %xmm0 movq c_cpu(%rip), %rcx movss %xmm0, (%rcx,%rax) addl $1, %edx addq $4, %rax cmpl $10001, %edx jne .L12 movl $1, %ecx movl $40000, %edx movq a_cpu(%rip), %rsi movq a_gpu(%rip), %rdi call cudaMemcpy@PLT movl $1, %ecx movl $40000, %edx movq b_cpu(%rip), %rsi movq b_gpu(%rip), %rdi call cudaMemcpy@PLT movl $1, %ecx movl $40000, %edx movq c_cpu(%rip), %rsi movq c_gpu(%rip), %rdi call cudaMemcpy@PLT movl $10, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $1, 8(%rsp) movl $1, 12(%rsp) movl $1, 16(%rsp) movl $0, %r9d movl $0, %r8d movq 20(%rsp), %rdx movl $1, %ecx movq 8(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L18 .L13: movl $2, %ecx movl $40000, %edx movq c_gpu(%rip), %rsi movq c_cpu(%rip), %rdi call cudaMemcpy@PLT movl $4, %ebx leaq .LC0(%rip), %rbp .L14: movq c_cpu(%rip), %rax pxor %xmm0, %xmm0 cvtss2sd (%rax,%rbx), %xmm0 movq %rbp, %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT addq $4, %rbx cmpq $40004, %rbx jne .L14 movq a_gpu(%rip), %rdi call cudaFree@PLT movq b_gpu(%rip), %rdi call cudaFree@PLT movq c_gpu(%rip), %rdi call cudaFree@PLT movq a_cpu(%rip), %rdi call free@PLT movq b_cpu(%rip), %rdi call free@PLT movq c_cpu(%rip), %rdi call free@PLT movq stdin(%rip), %rdi call getc@PLT movl $0, %eax addq $40, %rsp .cfi_remember_state .cfi_def_cfa_offset 24 popq %rbx .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 ret .L18: .cfi_restore_state movl $10000, %ecx movq c_gpu(%rip), %rdx movq b_gpu(%rip), %rsi movq a_gpu(%rip), %rdi call _Z27__device_stub__Z3addPfS_S_iPfS_S_i jmp .L13 .cfi_endproc .LFE3669: .size main, .-main .section .rodata.str1.1 .LC1: .string "_Z3addPfS_S_i" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB3697: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC1(%rip), %rdx movq %rdx, %rcx leaq _Z3addPfS_S_i(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3697: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .globl c_gpu .bss .align 8 .type c_gpu, @object .size c_gpu, 8 c_gpu: .zero 8 .globl b_gpu .align 8 .type b_gpu, @object .size b_gpu, 8 b_gpu: .zero 8 .globl a_gpu .align 8 .type a_gpu, @object .size a_gpu, 8 a_gpu: .zero 8 .globl c_cpu .align 8 .type c_cpu, @object .size c_cpu, 8 c_cpu: .zero 8 .globl b_cpu .align 8 .type b_cpu, @object .size b_cpu, 8 b_cpu: .zero 8 .globl a_cpu .align 8 .type a_cpu, @object .size a_cpu, 8 a_cpu: .zero 8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include <iostream> #define N 10000 float *a_cpu, *b_cpu, *c_cpu; float *a_gpu, *b_gpu, *c_gpu; __global__ void add(float *a_gpu, float *b_gpu, float *c_gpu, int n){ for(int i = 0; i < n; i++){ c_gpu[i] = a_gpu[i] + b_gpu[i]; } } int main() { a_cpu = (float*)malloc(sizeof(float)*N); b_cpu = (float*)malloc(sizeof(float)*N); c_cpu = (float*)malloc(sizeof(float)*N); cudaMalloc((float**)&a_gpu, sizeof(float) * N); cudaMalloc((float**)&b_gpu, sizeof(float) * N); cudaMalloc((float**)&c_gpu, sizeof(float) * N); for (int i = 1; i <= N; i++){ a_cpu[i] = -i; b_cpu[i] = i*2; c_cpu[i] = a_cpu[i] + b_cpu[i]; } cudaMemcpy(a_gpu, a_cpu, sizeof(float) * N, cudaMemcpyHostToDevice); cudaMemcpy(b_gpu, b_cpu, sizeof(float) * N, cudaMemcpyHostToDevice); cudaMemcpy(c_gpu, c_cpu, sizeof(float) * N, cudaMemcpyHostToDevice); add<<<1,10>>>(a_gpu, b_gpu, c_gpu, N); cudaMemcpy(c_cpu, c_gpu, sizeof(float) * N, cudaMemcpyDeviceToHost); for (int i = 1; i <= N; i++){ printf("%.0f\n", c_cpu[i]); } cudaFree(a_gpu); cudaFree(b_gpu); cudaFree(c_gpu); free(a_cpu); free(b_cpu); free(c_cpu); getchar(); return 0; }
#include <hip/hip_runtime.h> #include <iostream> #define N 10000 float *a_cpu, *b_cpu, *c_cpu; float *a_gpu, *b_gpu, *c_gpu; __global__ void add(float *a_gpu, float *b_gpu, float *c_gpu, int n){ for(int i = 0; i < n; i++){ c_gpu[i] = a_gpu[i] + b_gpu[i]; } } int main() { a_cpu = (float*)malloc(sizeof(float)*N); b_cpu = (float*)malloc(sizeof(float)*N); c_cpu = (float*)malloc(sizeof(float)*N); hipMalloc((float**)&a_gpu, sizeof(float) * N); hipMalloc((float**)&b_gpu, sizeof(float) * N); hipMalloc((float**)&c_gpu, sizeof(float) * N); for (int i = 1; i <= N; i++){ a_cpu[i] = -i; b_cpu[i] = i*2; c_cpu[i] = a_cpu[i] + b_cpu[i]; } hipMemcpy(a_gpu, a_cpu, sizeof(float) * N, hipMemcpyHostToDevice); hipMemcpy(b_gpu, b_cpu, sizeof(float) * N, hipMemcpyHostToDevice); hipMemcpy(c_gpu, c_cpu, sizeof(float) * N, hipMemcpyHostToDevice); add<<<1,10>>>(a_gpu, b_gpu, c_gpu, N); hipMemcpy(c_cpu, c_gpu, sizeof(float) * N, hipMemcpyDeviceToHost); for (int i = 1; i <= N; i++){ printf("%.0f\n", c_cpu[i]); } hipFree(a_gpu); hipFree(b_gpu); hipFree(c_gpu); free(a_cpu); free(b_cpu); free(c_cpu); getchar(); return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> #include <iostream> #define N 10000 float *a_cpu, *b_cpu, *c_cpu; float *a_gpu, *b_gpu, *c_gpu; __global__ void add(float *a_gpu, float *b_gpu, float *c_gpu, int n){ for(int i = 0; i < n; i++){ c_gpu[i] = a_gpu[i] + b_gpu[i]; } } int main() { a_cpu = (float*)malloc(sizeof(float)*N); b_cpu = (float*)malloc(sizeof(float)*N); c_cpu = (float*)malloc(sizeof(float)*N); hipMalloc((float**)&a_gpu, sizeof(float) * N); hipMalloc((float**)&b_gpu, sizeof(float) * N); hipMalloc((float**)&c_gpu, sizeof(float) * N); for (int i = 1; i <= N; i++){ a_cpu[i] = -i; b_cpu[i] = i*2; c_cpu[i] = a_cpu[i] + b_cpu[i]; } hipMemcpy(a_gpu, a_cpu, sizeof(float) * N, hipMemcpyHostToDevice); hipMemcpy(b_gpu, b_cpu, sizeof(float) * N, hipMemcpyHostToDevice); hipMemcpy(c_gpu, c_cpu, sizeof(float) * N, hipMemcpyHostToDevice); add<<<1,10>>>(a_gpu, b_gpu, c_gpu, N); hipMemcpy(c_cpu, c_gpu, sizeof(float) * N, hipMemcpyDeviceToHost); for (int i = 1; i <= N; i++){ printf("%.0f\n", c_cpu[i]); } hipFree(a_gpu); hipFree(b_gpu); hipFree(c_gpu); free(a_cpu); free(b_cpu); free(c_cpu); getchar(); return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z3addPfS_S_i .globl _Z3addPfS_S_i .p2align 8 .type _Z3addPfS_S_i,@function _Z3addPfS_S_i: s_load_b32 s2, s[0:1], 0x18 s_waitcnt lgkmcnt(0) s_cmp_lt_i32 s2, 1 s_cbranch_scc1 .LBB0_3 s_clause 0x1 s_load_b128 s[4:7], s[0:1], 0x0 s_load_b64 s[0:1], s[0:1], 0x10 v_mov_b32_e32 v0, 0 .LBB0_2: s_waitcnt lgkmcnt(0) s_clause 0x1 global_load_b32 v1, v0, s[4:5] global_load_b32 v2, v0, s[6:7] s_add_i32 s2, s2, -1 s_add_u32 s4, s4, 4 s_addc_u32 s5, s5, 0 s_add_u32 s6, s6, 4 s_addc_u32 s7, s7, 0 s_waitcnt vmcnt(0) v_add_f32_e32 v1, v1, v2 global_store_b32 v0, v1, s[0:1] s_add_u32 s0, s0, 4 s_addc_u32 s1, s1, 0 s_cmp_eq_u32 s2, 0 s_cbranch_scc0 .LBB0_2 .LBB0_3: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z3addPfS_S_i .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 28 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 3 .amdhsa_next_free_sgpr 8 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z3addPfS_S_i, .Lfunc_end0-_Z3addPfS_S_i .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: by_value .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 28 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z3addPfS_S_i .private_segment_fixed_size: 0 .sgpr_count: 8 .sgpr_spill_count: 0 .symbol: _Z3addPfS_S_i.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 3 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <hip/hip_runtime.h> #include <iostream> #define N 10000 float *a_cpu, *b_cpu, *c_cpu; float *a_gpu, *b_gpu, *c_gpu; __global__ void add(float *a_gpu, float *b_gpu, float *c_gpu, int n){ for(int i = 0; i < n; i++){ c_gpu[i] = a_gpu[i] + b_gpu[i]; } } int main() { a_cpu = (float*)malloc(sizeof(float)*N); b_cpu = (float*)malloc(sizeof(float)*N); c_cpu = (float*)malloc(sizeof(float)*N); hipMalloc((float**)&a_gpu, sizeof(float) * N); hipMalloc((float**)&b_gpu, sizeof(float) * N); hipMalloc((float**)&c_gpu, sizeof(float) * N); for (int i = 1; i <= N; i++){ a_cpu[i] = -i; b_cpu[i] = i*2; c_cpu[i] = a_cpu[i] + b_cpu[i]; } hipMemcpy(a_gpu, a_cpu, sizeof(float) * N, hipMemcpyHostToDevice); hipMemcpy(b_gpu, b_cpu, sizeof(float) * N, hipMemcpyHostToDevice); hipMemcpy(c_gpu, c_cpu, sizeof(float) * N, hipMemcpyHostToDevice); add<<<1,10>>>(a_gpu, b_gpu, c_gpu, N); hipMemcpy(c_cpu, c_gpu, sizeof(float) * N, hipMemcpyDeviceToHost); for (int i = 1; i <= N; i++){ printf("%.0f\n", c_cpu[i]); } hipFree(a_gpu); hipFree(b_gpu); hipFree(c_gpu); free(a_cpu); free(b_cpu); free(c_cpu); getchar(); return 0; }
.text .file "othermain.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl _Z18__device_stub__addPfS_S_i # -- Begin function _Z18__device_stub__addPfS_S_i .p2align 4, 0x90 .type _Z18__device_stub__addPfS_S_i,@function _Z18__device_stub__addPfS_S_i: # @_Z18__device_stub__addPfS_S_i .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movq %rdx, 56(%rsp) movl %ecx, 4(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 4(%rsp), %rax movq %rax, 104(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z3addPfS_S_i, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $136, %rsp .cfi_adjust_cfa_offset -136 retq .Lfunc_end0: .size _Z18__device_stub__addPfS_S_i, .Lfunc_end0-_Z18__device_stub__addPfS_S_i .cfi_endproc # -- End function .globl main # -- Begin function main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 subq $112, %rsp .cfi_def_cfa_offset 128 .cfi_offset %rbx, -16 movl $40000, %edi # imm = 0x9C40 callq malloc movq %rax, a_cpu(%rip) movl $40000, %edi # imm = 0x9C40 callq malloc movq %rax, b_cpu(%rip) movl $40000, %edi # imm = 0x9C40 callq malloc movq %rax, c_cpu(%rip) movl $a_gpu, %edi movl $40000, %esi # imm = 0x9C40 callq hipMalloc movl $b_gpu, %edi movl $40000, %esi # imm = 0x9C40 callq hipMalloc movl $c_gpu, %edi movl $40000, %esi # imm = 0x9C40 callq hipMalloc movq a_cpu(%rip), %rax movq b_cpu(%rip), %rcx movl $-1, %edx xorl %edi, %edi movq c_cpu(%rip), %rsi .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 xorps %xmm0, %xmm0 cvtsi2ss %edx, %xmm0 leaq 2(%rdi), %r8 xorps %xmm1, %xmm1 cvtsi2ss %r8d, %xmm1 movss %xmm0, 4(%rax,%rdi,2) movss %xmm1, 4(%rcx,%rdi,2) addss 4(%rax,%rdi,2), %xmm1 movss %xmm1, 4(%rsi,%rdi,2) decl %edx movq %r8, %rdi cmpq $20000, %r8 # imm = 0x4E20 jne .LBB1_1 # %bb.2: movq a_gpu(%rip), %rdi movq a_cpu(%rip), %rsi movl $40000, %edx # imm = 0x9C40 movl $1, %ecx callq hipMemcpy movq b_gpu(%rip), %rdi movq b_cpu(%rip), %rsi movl $40000, %edx # imm = 0x9C40 movl $1, %ecx callq hipMemcpy movq c_gpu(%rip), %rdi movq c_cpu(%rip), %rsi movl $40000, %edx # imm = 0x9C40 movl $1, %ecx callq hipMemcpy movabsq $4294967297, %rdi # imm = 0x100000001 leaq 9(%rdi), %rdx movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_4 # %bb.3: movq a_gpu(%rip), %rax movq b_gpu(%rip), %rcx movq c_gpu(%rip), %rdx movq %rax, 72(%rsp) movq %rcx, 64(%rsp) movq %rdx, 56(%rsp) movl $10000, 4(%rsp) # imm = 0x2710 leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 4(%rsp), %rax movq %rax, 104(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z3addPfS_S_i, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_4: movq c_cpu(%rip), %rdi movq c_gpu(%rip), %rsi movl $40000, %edx # imm = 0x9C40 movl $2, %ecx callq hipMemcpy movl $1, %ebx .p2align 4, 0x90 .LBB1_5: # =>This Inner Loop Header: Depth=1 movq c_cpu(%rip), %rax movss (%rax,%rbx,4), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str, %edi movb $1, %al callq printf incq %rbx cmpq $10001, %rbx # imm = 0x2711 jne .LBB1_5 # %bb.6: movq a_gpu(%rip), %rdi callq hipFree movq b_gpu(%rip), %rdi callq hipFree movq c_gpu(%rip), %rdi callq hipFree movq a_cpu(%rip), %rdi callq free movq b_cpu(%rip), %rdi callq free movq c_cpu(%rip), %rdi callq free movq stdin(%rip), %rdi callq getc xorl %eax, %eax addq $112, %rsp .cfi_def_cfa_offset 16 popq %rbx .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z3addPfS_S_i, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type a_cpu,@object # @a_cpu .bss .globl a_cpu .p2align 3, 0x0 a_cpu: .quad 0 .size a_cpu, 8 .type b_cpu,@object # @b_cpu .globl b_cpu .p2align 3, 0x0 b_cpu: .quad 0 .size b_cpu, 8 .type c_cpu,@object # @c_cpu .globl c_cpu .p2align 3, 0x0 c_cpu: .quad 0 .size c_cpu, 8 .type a_gpu,@object # @a_gpu .globl a_gpu .p2align 3, 0x0 a_gpu: .quad 0 .size a_gpu, 8 .type b_gpu,@object # @b_gpu .globl b_gpu .p2align 3, 0x0 b_gpu: .quad 0 .size b_gpu, 8 .type c_gpu,@object # @c_gpu .globl c_gpu .p2align 3, 0x0 c_gpu: .quad 0 .size c_gpu, 8 .type _Z3addPfS_S_i,@object # @_Z3addPfS_S_i .section .rodata,"a",@progbits .globl _Z3addPfS_S_i .p2align 3, 0x0 _Z3addPfS_S_i: .quad _Z18__device_stub__addPfS_S_i .size _Z3addPfS_S_i, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "%.0f\n" .size .L.str, 6 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z3addPfS_S_i" .size .L__unnamed_1, 14 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z18__device_stub__addPfS_S_i .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym a_gpu .addrsig_sym b_gpu .addrsig_sym c_gpu .addrsig_sym _Z3addPfS_S_i .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z3addPfS_S_i .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ IMAD.MOV.U32 R1, RZ, RZ, c[0x0][0x28] ; /* 0x00000a00ff017624 */ /* 0x000fe400078e00ff */ /*0010*/ IMAD.MOV.U32 R0, RZ, RZ, c[0x0][0x178] ; /* 0x00005e00ff007624 */ /* 0x000fca00078e00ff */ /*0020*/ ISETP.GE.AND P0, PT, R0, 0x1, PT ; /* 0x000000010000780c */ /* 0x000fda0003f06270 */ /*0030*/ @!P0 EXIT ; /* 0x000000000000894d */ /* 0x000fea0003800000 */ /*0040*/ IADD3 R2, R0.reuse, -0x1, RZ ; /* 0xffffffff00027810 */ /* 0x040fe20007ffe0ff */ /*0050*/ UMOV UR4, URZ ; /* 0x0000003f00047c82 */ /* 0x000fe20008000000 */ /*0060*/ LOP3.LUT R0, R0, 0x3, RZ, 0xc0, !PT ; /* 0x0000000300007812 */ /* 0x000fe200078ec0ff */ /*0070*/ ULDC.64 UR12, c[0x0][0x118] ; /* 0x00004600000c7ab9 */ /* 0x000fe20000000a00 */ /*0080*/ ISETP.GE.U32.AND P0, PT, R2, 0x3, PT ; /* 0x000000030200780c */ /* 0x000fda0003f06070 */ /*0090*/ @!P0 BRA 0xb80 ; /* 0x00000ae000008947 */ /* 0x000fea0003800000 */ /*00a0*/ IADD3 R8, -R0, c[0x0][0x178], RZ ; /* 0x00005e0000087a10 */ /* 0x000fe20007ffe1ff */ /*00b0*/ UMOV UR4, URZ ; /* 0x0000003f00047c82 */ /* 0x000fe20008000000 */ /*00c0*/ MOV R4, c[0x0][0x170] ; /* 0x00005c0000047a02 */ /* 0x000fe20000000f00 */ /*00d0*/ IMAD.MOV.U32 R5, RZ, RZ, c[0x0][0x174] ; /* 0x00005d00ff057624 */ /* 0x000fe200078e00ff */ /*00e0*/ ISETP.GT.AND P0, PT, R8, RZ, PT ; /* 0x000000ff0800720c */ /* 0x000fe20003f04270 */ /*00f0*/ IMAD.MOV.U32 R3, RZ, RZ, c[0x0][0x16c] ; /* 0x00005b00ff037624 */ /* 0x000fe200078e00ff */ /*0100*/ MOV R2, c[0x0][0x168] ; /* 0x00005a0000027a02 */ /* 0x000fe20000000f00 */ /*0110*/ IMAD.MOV.U32 R7, RZ, RZ, c[0x0][0x164] ; /* 0x00005900ff077624 */ /* 0x000fe200078e00ff */ /*0120*/ MOV R6, c[0x0][0x160] ; /* 0x0000580000067a02 */ /* 0x000fd20000000f00 */ /*0130*/ @!P0 BRA 0x990 ; /* 0x0000085000008947 */ /* 0x000fea0003800000 */ /*0140*/ ISETP.GT.AND P1, PT, R8, 0xc, PT ; /* 0x0000000c0800780c */ /* 0x000fe40003f24270 */ /*0150*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x80, 0x0 ; /* 0x000000000000781c */ /* 0x000fd60003f0f070 */ /*0160*/ @!P1 BRA 0x670 ; /* 0x0000050000009947 */ /* 0x000fea0003800000 */ /*0170*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe40003f0e170 */ /*0180*/ LDG.E R9, [R2.64] ; /* 0x0000000c02097981 */ /* 0x000ea8000c1e1900 */ /*0190*/ LDG.E R10, [R6.64] ; /* 0x0000000c060a7981 */ /* 0x000ea4000c1e1900 */ /*01a0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x004fca0000000000 */ /*01b0*/ STG.E [R4.64], R9 ; /* 0x0000000904007986 */ /* 0x0001e8000c10190c */ /*01c0*/ LDG.E R10, [R2.64+0x4] ; /* 0x0000040c020a7981 */ /* 0x000ea8000c1e1900 */ /*01d0*/ LDG.E R11, [R6.64+0x4] ; /* 0x0000040c060b7981 */ /* 0x000ea4000c1e1900 */ /*01e0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x004fca0000000000 */ /*01f0*/ STG.E [R4.64+0x4], R11 ; /* 0x0000040b04007986 */ /* 0x0003e8000c10190c */ /*0200*/ LDG.E R10, [R2.64+0x8] ; /* 0x0000080c020a7981 */ /* 0x000ea8000c1e1900 */ /*0210*/ LDG.E R13, [R6.64+0x8] ; /* 0x0000080c060d7981 */ /* 0x000ea4000c1e1900 */ /*0220*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x004fca0000000000 */ /*0230*/ STG.E [R4.64+0x8], R13 ; /* 0x0000080d04007986 */ /* 0x0005e8000c10190c */ /*0240*/ LDG.E R10, [R2.64+0xc] ; /* 0x00000c0c020a7981 */ /* 0x000ee8000c1e1900 */ /*0250*/ LDG.E R15, [R6.64+0xc] ; /* 0x00000c0c060f7981 */ /* 0x000ee4000c1e1900 */ /*0260*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x008fca0000000000 */ /*0270*/ STG.E [R4.64+0xc], R15 ; /* 0x00000c0f04007986 */ /* 0x0007e8000c10190c */ /*0280*/ LDG.E R9, [R2.64+0x10] ; /* 0x0000100c02097981 */ /* 0x001f28000c1e1900 */ /*0290*/ LDG.E R10, [R6.64+0x10] ; /* 0x0000100c060a7981 */ /* 0x000f24000c1e1900 */ /*02a0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x010fca0000000000 */ /*02b0*/ STG.E [R4.64+0x10], R9 ; /* 0x0000100904007986 */ /* 0x0001e8000c10190c */ /*02c0*/ LDG.E R10, [R2.64+0x14] ; /* 0x0000140c020a7981 */ /* 0x000f28000c1e1900 */ /*02d0*/ LDG.E R11, [R6.64+0x14] ; /* 0x0000140c060b7981 */ /* 0x002f24000c1e1900 */ /*02e0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x010fca0000000000 */ /*02f0*/ STG.E [R4.64+0x14], R11 ; /* 0x0000140b04007986 */ /* 0x0003e8000c10190c */ /*0300*/ LDG.E R10, [R2.64+0x18] ; /* 0x0000180c020a7981 */ /* 0x000f28000c1e1900 */ /*0310*/ LDG.E R13, [R6.64+0x18] ; /* 0x0000180c060d7981 */ /* 0x004f24000c1e1900 */ /*0320*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x010fca0000000000 */ /*0330*/ STG.E [R4.64+0x18], R13 ; /* 0x0000180d04007986 */ /* 0x0005e8000c10190c */ /*0340*/ LDG.E R10, [R2.64+0x1c] ; /* 0x00001c0c020a7981 */ /* 0x000f28000c1e1900 */ /*0350*/ LDG.E R15, [R6.64+0x1c] ; /* 0x00001c0c060f7981 */ /* 0x008f24000c1e1900 */ /*0360*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x010fca0000000000 */ /*0370*/ STG.E [R4.64+0x1c], R15 ; /* 0x00001c0f04007986 */ /* 0x0007e8000c10190c */ /*0380*/ LDG.E R9, [R2.64+0x20] ; /* 0x0000200c02097981 */ /* 0x001f28000c1e1900 */ /*0390*/ LDG.E R10, [R6.64+0x20] ; /* 0x0000200c060a7981 */ /* 0x000f24000c1e1900 */ /*03a0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x010fca0000000000 */ /*03b0*/ STG.E [R4.64+0x20], R9 ; /* 0x0000200904007986 */ /* 0x0001e8000c10190c */ /*03c0*/ LDG.E R10, [R2.64+0x24] ; /* 0x0000240c020a7981 */ /* 0x000f28000c1e1900 */ /*03d0*/ LDG.E R11, [R6.64+0x24] ; /* 0x0000240c060b7981 */ /* 0x002f24000c1e1900 */ /*03e0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x010fca0000000000 */ /*03f0*/ STG.E [R4.64+0x24], R11 ; /* 0x0000240b04007986 */ /* 0x0003e8000c10190c */ /*0400*/ LDG.E R10, [R2.64+0x28] ; /* 0x0000280c020a7981 */ /* 0x000f28000c1e1900 */ /*0410*/ LDG.E R13, [R6.64+0x28] ; /* 0x0000280c060d7981 */ /* 0x004f24000c1e1900 */ /*0420*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x010fca0000000000 */ /*0430*/ STG.E [R4.64+0x28], R13 ; /* 0x0000280d04007986 */ /* 0x0005e8000c10190c */ /*0440*/ LDG.E R10, [R2.64+0x2c] ; /* 0x00002c0c020a7981 */ /* 0x000f28000c1e1900 */ /*0450*/ LDG.E R15, [R6.64+0x2c] ; /* 0x00002c0c060f7981 */ /* 0x008f24000c1e1900 */ /*0460*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x010fca0000000000 */ /*0470*/ STG.E [R4.64+0x2c], R15 ; /* 0x00002c0f04007986 */ /* 0x0007e8000c10190c */ /*0480*/ LDG.E R9, [R2.64+0x30] ; /* 0x0000300c02097981 */ /* 0x001f28000c1e1900 */ /*0490*/ LDG.E R10, [R6.64+0x30] ; /* 0x0000300c060a7981 */ /* 0x000f24000c1e1900 */ /*04a0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x010fca0000000000 */ /*04b0*/ STG.E [R4.64+0x30], R9 ; /* 0x0000300904007986 */ /* 0x000fe8000c10190c */ /*04c0*/ LDG.E R10, [R2.64+0x34] ; /* 0x0000340c020a7981 */ /* 0x000f28000c1e1900 */ /*04d0*/ LDG.E R11, [R6.64+0x34] ; /* 0x0000340c060b7981 */ /* 0x002f24000c1e1900 */ /*04e0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x010fca0000000000 */ /*04f0*/ STG.E [R4.64+0x34], R11 ; /* 0x0000340b04007986 */ /* 0x0001e8000c10190c */ /*0500*/ LDG.E R10, [R2.64+0x38] ; /* 0x0000380c020a7981 */ /* 0x000f28000c1e1900 */ /*0510*/ LDG.E R13, [R6.64+0x38] ; /* 0x0000380c060d7981 */ /* 0x004f22000c1e1900 */ /*0520*/ IADD3 R12, P1, R2, 0x40, RZ ; /* 0x00000040020c7810 */ /* 0x000fe40007f3e0ff */ /*0530*/ IADD3 R8, R8, -0x10, RZ ; /* 0xfffffff008087810 */ /* 0x000fe20007ffe0ff */ /*0540*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x010fca0000000000 */ /*0550*/ STG.E [R4.64+0x38], R13 ; /* 0x0000380d04007986 */ /* 0x000fe8000c10190c */ /*0560*/ LDG.E R10, [R2.64+0x3c] ; /* 0x00003c0c020a7981 */ /* 0x0002a8000c1e1900 */ /*0570*/ LDG.E R15, [R6.64+0x3c] ; /* 0x00003c0c060f7981 */ /* 0x0086a2000c1e1900 */ /*0580*/ IADD3.X R11, RZ, R3, RZ, P1, !PT ; /* 0x00000003ff0b7210 */ /* 0x001fe20000ffe4ff */ /*0590*/ UIADD3 UR4, UR4, 0x10, URZ ; /* 0x0000001004047890 */ /* 0x000fe2000fffe03f */ /*05a0*/ ISETP.GT.AND P1, PT, R8, 0xc, PT ; /* 0x0000000c0800780c */ /* 0x000fc40003f24270 */ /*05b0*/ IADD3 R9, P3, R4, 0x40, RZ ; /* 0x0000004004097810 */ /* 0x000fe20007f7e0ff */ /*05c0*/ IMAD.MOV.U32 R2, RZ, RZ, R12 ; /* 0x000000ffff027224 */ /* 0x002fe200078e000c */ /*05d0*/ IADD3 R14, P2, R6, 0x40, RZ ; /* 0x00000040060e7810 */ /* 0x000fe40007f5e0ff */ /*05e0*/ MOV R3, R11 ; /* 0x0000000b00037202 */ /* 0x000fc60000000f00 */ /*05f0*/ IMAD.X R7, RZ, RZ, R7, P2 ; /* 0x000000ffff077224 */ /* 0x008fe400010e0607 */ /*0600*/ IMAD.MOV.U32 R6, RZ, RZ, R14 ; /* 0x000000ffff067224 */ /* 0x000fe400078e000e */ /*0610*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x004fe20000000000 */ /*0620*/ IADD3.X R10, RZ, R5, RZ, P3, !PT ; /* 0x00000005ff0a7210 */ /* 0x000fc80001ffe4ff */ /*0630*/ STG.E [R4.64+0x3c], R15 ; /* 0x00003c0f04007986 */ /* 0x0001e4000c10190c */ /*0640*/ MOV R4, R9 ; /* 0x0000000900047202 */ /* 0x001fe20000000f00 */ /*0650*/ IMAD.MOV.U32 R5, RZ, RZ, R10 ; /* 0x000000ffff057224 */ /* 0x000fe200078e000a */ /*0660*/ @P1 BRA 0x180 ; /* 0xfffffb1000001947 */ /* 0x000fea000383ffff */ /*0670*/ ISETP.GT.AND P1, PT, R8, 0x4, PT ; /* 0x000000040800780c */ /* 0x000fda0003f24270 */ /*0680*/ @!P1 BRA 0x970 ; /* 0x000002e000009947 */ /* 0x000fea0003800000 */ /*0690*/ LDG.E R9, [R2.64] ; /* 0x0000000c02097981 */ /* 0x000ea8000c1e1900 */ /*06a0*/ LDG.E R10, [R6.64] ; /* 0x0000000c060a7981 */ /* 0x000ea4000c1e1900 */ /*06b0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x004fca0000000000 */ /*06c0*/ STG.E [R4.64], R9 ; /* 0x0000000904007986 */ /* 0x0001e8000c10190c */ /*06d0*/ LDG.E R10, [R2.64+0x4] ; /* 0x0000040c020a7981 */ /* 0x000ea8000c1e1900 */ /*06e0*/ LDG.E R11, [R6.64+0x4] ; /* 0x0000040c060b7981 */ /* 0x000ea4000c1e1900 */ /*06f0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x004fca0000000000 */ /*0700*/ STG.E [R4.64+0x4], R11 ; /* 0x0000040b04007986 */ /* 0x0003e8000c10190c */ /*0710*/ LDG.E R10, [R2.64+0x8] ; /* 0x0000080c020a7981 */ /* 0x000ea8000c1e1900 */ /*0720*/ LDG.E R13, [R6.64+0x8] ; /* 0x0000080c060d7981 */ /* 0x000ea4000c1e1900 */ /*0730*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x004fca0000000000 */ /*0740*/ STG.E [R4.64+0x8], R13 ; /* 0x0000080d04007986 */ /* 0x0005e8000c10190c */ /*0750*/ LDG.E R10, [R2.64+0xc] ; /* 0x00000c0c020a7981 */ /* 0x000ee8000c1e1900 */ /*0760*/ LDG.E R15, [R6.64+0xc] ; /* 0x00000c0c060f7981 */ /* 0x000ee4000c1e1900 */ /*0770*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x008fca0000000000 */ /*0780*/ STG.E [R4.64+0xc], R15 ; /* 0x00000c0f04007986 */ /* 0x0007e8000c10190c */ /*0790*/ LDG.E R9, [R2.64+0x10] ; /* 0x0000100c02097981 */ /* 0x001f28000c1e1900 */ /*07a0*/ LDG.E R10, [R6.64+0x10] ; /* 0x0000100c060a7981 */ /* 0x000f24000c1e1900 */ /*07b0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x010fca0000000000 */ /*07c0*/ STG.E [R4.64+0x10], R9 ; /* 0x0000100904007986 */ /* 0x000fe8000c10190c */ /*07d0*/ LDG.E R10, [R2.64+0x14] ; /* 0x0000140c020a7981 */ /* 0x000f28000c1e1900 */ /*07e0*/ LDG.E R11, [R6.64+0x14] ; /* 0x0000140c060b7981 */ /* 0x002f24000c1e1900 */ /*07f0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x010fca0000000000 */ /*0800*/ STG.E [R4.64+0x14], R11 ; /* 0x0000140b04007986 */ /* 0x0001e8000c10190c */ /*0810*/ LDG.E R10, [R2.64+0x18] ; /* 0x0000180c020a7981 */ /* 0x000f28000c1e1900 */ /*0820*/ LDG.E R13, [R6.64+0x18] ; /* 0x0000180c060d7981 */ /* 0x004f24000c1e1900 */ /*0830*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x010fca0000000000 */ /*0840*/ STG.E [R4.64+0x18], R13 ; /* 0x0000180d04007986 */ /* 0x000fe8000c10190c */ /*0850*/ LDG.E R10, [R2.64+0x1c] ; /* 0x00001c0c020a7981 */ /* 0x0002a8000c1e1900 */ /*0860*/ LDG.E R15, [R6.64+0x1c] ; /* 0x00001c0c060f7981 */ /* 0x0086a2000c1e1900 */ /*0870*/ IADD3 R11, P2, R2, 0x20, RZ ; /* 0x00000020020b7810 */ /* 0x001fe20007f5e0ff */ /*0880*/ UIADD3 UR4, UR4, 0x8, URZ ; /* 0x0000000804047890 */ /* 0x000fe2000fffe03f */ /*0890*/ IADD3 R9, P3, R4, 0x20, RZ ; /* 0x0000002004097810 */ /* 0x000fc40007f7e0ff */ /*08a0*/ IADD3 R14, P1, R6, 0x20, RZ ; /* 0x00000020060e7810 */ /* 0x000fe20007f3e0ff */ /*08b0*/ IMAD.X R12, RZ, RZ, R3, P2 ; /* 0x000000ffff0c7224 */ /* 0x000fe200010e0603 */ /*08c0*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe20003f0e170 */ /*08d0*/ IMAD.MOV.U32 R2, RZ, RZ, R11 ; /* 0x000000ffff027224 */ /* 0x002fe200078e000b */ /*08e0*/ IADD3 R8, R8, -0x8, RZ ; /* 0xfffffff808087810 */ /* 0x000fe20007ffe0ff */ /*08f0*/ IMAD.MOV.U32 R6, RZ, RZ, R14 ; /* 0x000000ffff067224 */ /* 0x008fe200078e000e */ /*0900*/ IADD3.X R7, RZ, R7, RZ, P1, !PT ; /* 0x00000007ff077210 */ /* 0x000fe40000ffe4ff */ /*0910*/ MOV R3, R12 ; /* 0x0000000c00037202 */ /* 0x000fe20000000f00 */ /*0920*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x004fe20000000000 */ /*0930*/ IADD3.X R10, RZ, R5, RZ, P3, !PT ; /* 0x00000005ff0a7210 */ /* 0x000fc80001ffe4ff */ /*0940*/ STG.E [R4.64+0x1c], R15 ; /* 0x00001c0f04007986 */ /* 0x0001e4000c10190c */ /*0950*/ MOV R4, R9 ; /* 0x0000000900047202 */ /* 0x001fe20000000f00 */ /*0960*/ IMAD.MOV.U32 R5, RZ, RZ, R10 ; /* 0x000000ffff057224 */ /* 0x000fe400078e000a */ /*0970*/ ISETP.NE.OR P0, PT, R8, RZ, P0 ; /* 0x000000ff0800720c */ /* 0x000fda0000705670 */ /*0980*/ @!P0 BRA 0xb80 ; /* 0x000001f000008947 */ /* 0x000fea0003800000 */ /*0990*/ LDG.E R9, [R2.64] ; /* 0x0000000c02097981 */ /* 0x000ea8000c1e1900 */ /*09a0*/ LDG.E R10, [R6.64] ; /* 0x0000000c060a7981 */ /* 0x000ea4000c1e1900 */ /*09b0*/ FADD R9, R9, R10 ; /* 0x0000000a09097221 */ /* 0x004fca0000000000 */ /*09c0*/ STG.E [R4.64], R9 ; /* 0x0000000904007986 */ /* 0x000fe8000c10190c */ /*09d0*/ LDG.E R10, [R2.64+0x4] ; /* 0x0000040c020a7981 */ /* 0x000ea8000c1e1900 */ /*09e0*/ LDG.E R11, [R6.64+0x4] ; /* 0x0000040c060b7981 */ /* 0x000ea4000c1e1900 */ /*09f0*/ FADD R11, R10, R11 ; /* 0x0000000b0a0b7221 */ /* 0x004fca0000000000 */ /*0a00*/ STG.E [R4.64+0x4], R11 ; /* 0x0000040b04007986 */ /* 0x0001e8000c10190c */ /*0a10*/ LDG.E R10, [R2.64+0x8] ; /* 0x0000080c020a7981 */ /* 0x000ea8000c1e1900 */ /*0a20*/ LDG.E R13, [R6.64+0x8] ; /* 0x0000080c060d7981 */ /* 0x000ea2000c1e1900 */ /*0a30*/ IADD3 R12, P0, R2, 0x10, RZ ; /* 0x00000010020c7810 */ /* 0x000fe40007f1e0ff */ /*0a40*/ IADD3 R8, R8, -0x4, RZ ; /* 0xfffffffc08087810 */ /* 0x000fe20007ffe0ff */ /*0a50*/ FADD R13, R10, R13 ; /* 0x0000000d0a0d7221 */ /* 0x004fca0000000000 */ /*0a60*/ STG.E [R4.64+0x8], R13 ; /* 0x0000080d04007986 */ /* 0x000fe8000c10190c */ /*0a70*/ LDG.E R10, [R2.64+0xc] ; /* 0x00000c0c020a7981 */ /* 0x0002a8000c1e1900 */ /*0a80*/ LDG.E R15, [R6.64+0xc] ; /* 0x00000c0c060f7981 */ /* 0x0006a2000c1e1900 */ /*0a90*/ IADD3.X R11, RZ, R3, RZ, P0, !PT ; /* 0x00000003ff0b7210 */ /* 0x001fe200007fe4ff */ /*0aa0*/ UIADD3 UR4, UR4, 0x4, URZ ; /* 0x0000000404047890 */ /* 0x000fe2000fffe03f */ /*0ab0*/ ISETP.NE.AND P0, PT, R8, RZ, PT ; /* 0x000000ff0800720c */ /* 0x000fc40003f05270 */ /*0ac0*/ IADD3 R9, P2, R4, 0x10, RZ ; /* 0x0000001004097810 */ /* 0x000fe20007f5e0ff */ /*0ad0*/ IMAD.MOV.U32 R2, RZ, RZ, R12 ; /* 0x000000ffff027224 */ /* 0x002fe200078e000c */ /*0ae0*/ IADD3 R14, P1, R6, 0x10, RZ ; /* 0x00000010060e7810 */ /* 0x000fe40007f3e0ff */ /*0af0*/ MOV R3, R11 ; /* 0x0000000b00037202 */ /* 0x000fc60000000f00 */ /*0b00*/ IMAD.X R7, RZ, RZ, R7, P1 ; /* 0x000000ffff077224 */ /* 0x008fe400008e0607 */ /*0b10*/ IMAD.MOV.U32 R6, RZ, RZ, R14 ; /* 0x000000ffff067224 */ /* 0x000fe400078e000e */ /*0b20*/ FADD R15, R10, R15 ; /* 0x0000000f0a0f7221 */ /* 0x004fe20000000000 */ /*0b30*/ IADD3.X R10, RZ, R5, RZ, P2, !PT ; /* 0x00000005ff0a7210 */ /* 0x000fc800017fe4ff */ /*0b40*/ STG.E [R4.64+0xc], R15 ; /* 0x00000c0f04007986 */ /* 0x0001e4000c10190c */ /*0b50*/ MOV R4, R9 ; /* 0x0000000900047202 */ /* 0x001fe20000000f00 */ /*0b60*/ IMAD.MOV.U32 R5, RZ, RZ, R10 ; /* 0x000000ffff057224 */ /* 0x000fe200078e000a */ /*0b70*/ @P0 BRA 0x990 ; /* 0xfffffe1000000947 */ /* 0x000fea000383ffff */ /*0b80*/ ISETP.NE.AND P0, PT, R0, RZ, PT ; /* 0x000000ff0000720c */ /* 0x000fda0003f05270 */ /*0b90*/ @!P0 EXIT ; /* 0x000000000000894d */ /* 0x000fea0003800000 */ /*0ba0*/ UMOV UR5, 0x4 ; /* 0x0000000400057882 */ /* 0x000fe40000000000 */ /*0bb0*/ ULDC.64 UR6, c[0x0][0x170] ; /* 0x00005c0000067ab9 */ /* 0x000fe40000000a00 */ /*0bc0*/ ULDC.64 UR8, c[0x0][0x168] ; /* 0x00005a0000087ab9 */ /* 0x000fe40000000a00 */ /*0bd0*/ ULDC.64 UR10, c[0x0][0x160] ; /* 0x00005800000a7ab9 */ /* 0x000fe40000000a00 */ /*0be0*/ UIMAD.WIDE UR6, UR4, UR5, UR6 ; /* 0x00000005040672a5 */ /* 0x000fe4000f8e0206 */ /*0bf0*/ UIMAD.WIDE UR8, UR4, UR5, UR8 ; /* 0x00000005040872a5 */ /* 0x000fc4000f8e0208 */ /*0c00*/ UIMAD.WIDE UR4, UR4, UR5, UR10 ; /* 0x00000005040472a5 */ /* 0x000fca000f8e020a */ /*0c10*/ MOV R2, UR8 ; /* 0x0000000800027c02 */ /* 0x000fe20008000f00 */ /*0c20*/ IMAD.U32 R5, RZ, RZ, UR5 ; /* 0x00000005ff057e24 */ /* 0x000fe2000f8e00ff */ /*0c30*/ MOV R4, UR4 ; /* 0x0000000400047c02 */ /* 0x000fe20008000f00 */ /*0c40*/ IMAD.U32 R3, RZ, RZ, UR9 ; /* 0x00000009ff037e24 */ /* 0x000fc8000f8e00ff */ /*0c50*/ LDG.E R5, [R4.64] ; /* 0x0000000c04057981 */ /* 0x000ea8000c1e1900 */ /*0c60*/ LDG.E R2, [R2.64] ; /* 0x0000000c02027981 */ /* 0x000ea2000c1e1900 */ /*0c70*/ IADD3 R0, R0, -0x1, RZ ; /* 0xffffffff00007810 */ /* 0x000fc80007ffe0ff */ /*0c80*/ ISETP.NE.AND P0, PT, R0, RZ, PT ; /* 0x000000ff0000720c */ /* 0x000fe20003f05270 */ /*0c90*/ UIADD3 UR8, UP1, UR8, 0x4, URZ ; /* 0x0000000408087890 */ /* 0x000fe2000ff3e03f */ /*0ca0*/ MOV R6, UR6 ; /* 0x0000000600067c02 */ /* 0x001fe20008000f00 */ /*0cb0*/ UIADD3 UR6, UP0, UR6, 0x4, URZ ; /* 0x0000000406067890 */ /* 0x000fe2000ff1e03f */ /*0cc0*/ MOV R7, UR7 ; /* 0x0000000700077c02 */ /* 0x000fe20008000f00 */ /*0cd0*/ UIADD3 UR4, UP2, UR4, 0x4, URZ ; /* 0x0000000404047890 */ /* 0x000fe4000ff5e03f */ /*0ce0*/ UIADD3.X UR7, URZ, UR7, URZ, UP0, !UPT ; /* 0x000000073f077290 */ /* 0x000fe400087fe43f */ /*0cf0*/ UIADD3.X UR9, URZ, UR9, URZ, UP1, !UPT ; /* 0x000000093f097290 */ /* 0x000fe40008ffe43f */ /*0d00*/ UIADD3.X UR5, URZ, UR5, URZ, UP2, !UPT ; /* 0x000000053f057290 */ /* 0x000fe200097fe43f */ /*0d10*/ FADD R9, R2, R5 ; /* 0x0000000502097221 */ /* 0x004fca0000000000 */ /*0d20*/ STG.E [R6.64], R9 ; /* 0x0000000906007986 */ /* 0x0001e2000c10190c */ /*0d30*/ @P0 BRA 0xc10 ; /* 0xfffffed000000947 */ /* 0x000fea000383ffff */ /*0d40*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0d50*/ BRA 0xd50; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0d60*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0d70*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0d80*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0d90*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0da0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0db0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0dc0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0dd0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0de0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0df0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z3addPfS_S_i .globl _Z3addPfS_S_i .p2align 8 .type _Z3addPfS_S_i,@function _Z3addPfS_S_i: s_load_b32 s2, s[0:1], 0x18 s_waitcnt lgkmcnt(0) s_cmp_lt_i32 s2, 1 s_cbranch_scc1 .LBB0_3 s_clause 0x1 s_load_b128 s[4:7], s[0:1], 0x0 s_load_b64 s[0:1], s[0:1], 0x10 v_mov_b32_e32 v0, 0 .LBB0_2: s_waitcnt lgkmcnt(0) s_clause 0x1 global_load_b32 v1, v0, s[4:5] global_load_b32 v2, v0, s[6:7] s_add_i32 s2, s2, -1 s_add_u32 s4, s4, 4 s_addc_u32 s5, s5, 0 s_add_u32 s6, s6, 4 s_addc_u32 s7, s7, 0 s_waitcnt vmcnt(0) v_add_f32_e32 v1, v1, v2 global_store_b32 v0, v1, s[0:1] s_add_u32 s0, s0, 4 s_addc_u32 s1, s1, 0 s_cmp_eq_u32 s2, 0 s_cbranch_scc0 .LBB0_2 .LBB0_3: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z3addPfS_S_i .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 28 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 3 .amdhsa_next_free_sgpr 8 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z3addPfS_S_i, .Lfunc_end0-_Z3addPfS_S_i .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: by_value .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 28 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z3addPfS_S_i .private_segment_fixed_size: 0 .sgpr_count: 8 .sgpr_spill_count: 0 .symbol: _Z3addPfS_S_i.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 3 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_0018e52c_00000000-6_othermain.cudafe1.cpp" .text #APP .globl _ZSt21ios_base_library_initv #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB3672: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3672: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z27__device_stub__Z3addPfS_S_iPfS_S_i .type _Z27__device_stub__Z3addPfS_S_iPfS_S_i, @function _Z27__device_stub__Z3addPfS_S_iPfS_S_i: .LFB3694: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movl %ecx, 4(%rsp) movq %fs:40, %rax movq %rax, 136(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) leaq 4(%rsp), %rax movq %rax, 120(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 136(%rsp), %rax subq %fs:40, %rax jne .L8 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 168 pushq 40(%rsp) .cfi_def_cfa_offset 176 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z3addPfS_S_i(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE3694: .size _Z27__device_stub__Z3addPfS_S_iPfS_S_i, .-_Z27__device_stub__Z3addPfS_S_iPfS_S_i .globl _Z3addPfS_S_i .type _Z3addPfS_S_i, @function _Z3addPfS_S_i: .LFB3695: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z27__device_stub__Z3addPfS_S_iPfS_S_i addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3695: .size _Z3addPfS_S_i, .-_Z3addPfS_S_i .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "%.0f\n" .text .globl main .type main, @function main: .LFB3669: .cfi_startproc endbr64 pushq %rbp .cfi_def_cfa_offset 16 .cfi_offset 6, -16 pushq %rbx .cfi_def_cfa_offset 24 .cfi_offset 3, -24 subq $40, %rsp .cfi_def_cfa_offset 64 movl $40000, %edi call malloc@PLT movq %rax, a_cpu(%rip) movl $40000, %edi call malloc@PLT movq %rax, b_cpu(%rip) movl $40000, %edi call malloc@PLT movq %rax, c_cpu(%rip) movl $40000, %esi leaq a_gpu(%rip), %rdi call cudaMalloc@PLT movl $40000, %esi leaq b_gpu(%rip), %rdi call cudaMalloc@PLT movl $40000, %esi leaq c_gpu(%rip), %rdi call cudaMalloc@PLT movl $4, %eax movl $1, %edx .L12: movl %edx, %ecx negl %ecx pxor %xmm0, %xmm0 cvtsi2ssl %ecx, %xmm0 movq a_cpu(%rip), %rcx movss %xmm0, (%rcx,%rax) leal (%rdx,%rdx), %ecx pxor %xmm0, %xmm0 cvtsi2ssl %ecx, %xmm0 movq b_cpu(%rip), %rcx movss %xmm0, (%rcx,%rax) movq a_cpu(%rip), %rcx movss (%rcx,%rax), %xmm0 movq b_cpu(%rip), %rcx addss (%rcx,%rax), %xmm0 movq c_cpu(%rip), %rcx movss %xmm0, (%rcx,%rax) addl $1, %edx addq $4, %rax cmpl $10001, %edx jne .L12 movl $1, %ecx movl $40000, %edx movq a_cpu(%rip), %rsi movq a_gpu(%rip), %rdi call cudaMemcpy@PLT movl $1, %ecx movl $40000, %edx movq b_cpu(%rip), %rsi movq b_gpu(%rip), %rdi call cudaMemcpy@PLT movl $1, %ecx movl $40000, %edx movq c_cpu(%rip), %rsi movq c_gpu(%rip), %rdi call cudaMemcpy@PLT movl $10, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $1, 8(%rsp) movl $1, 12(%rsp) movl $1, 16(%rsp) movl $0, %r9d movl $0, %r8d movq 20(%rsp), %rdx movl $1, %ecx movq 8(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L18 .L13: movl $2, %ecx movl $40000, %edx movq c_gpu(%rip), %rsi movq c_cpu(%rip), %rdi call cudaMemcpy@PLT movl $4, %ebx leaq .LC0(%rip), %rbp .L14: movq c_cpu(%rip), %rax pxor %xmm0, %xmm0 cvtss2sd (%rax,%rbx), %xmm0 movq %rbp, %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT addq $4, %rbx cmpq $40004, %rbx jne .L14 movq a_gpu(%rip), %rdi call cudaFree@PLT movq b_gpu(%rip), %rdi call cudaFree@PLT movq c_gpu(%rip), %rdi call cudaFree@PLT movq a_cpu(%rip), %rdi call free@PLT movq b_cpu(%rip), %rdi call free@PLT movq c_cpu(%rip), %rdi call free@PLT movq stdin(%rip), %rdi call getc@PLT movl $0, %eax addq $40, %rsp .cfi_remember_state .cfi_def_cfa_offset 24 popq %rbx .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 ret .L18: .cfi_restore_state movl $10000, %ecx movq c_gpu(%rip), %rdx movq b_gpu(%rip), %rsi movq a_gpu(%rip), %rdi call _Z27__device_stub__Z3addPfS_S_iPfS_S_i jmp .L13 .cfi_endproc .LFE3669: .size main, .-main .section .rodata.str1.1 .LC1: .string "_Z3addPfS_S_i" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB3697: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC1(%rip), %rdx movq %rdx, %rcx leaq _Z3addPfS_S_i(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3697: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .globl c_gpu .bss .align 8 .type c_gpu, @object .size c_gpu, 8 c_gpu: .zero 8 .globl b_gpu .align 8 .type b_gpu, @object .size b_gpu, 8 b_gpu: .zero 8 .globl a_gpu .align 8 .type a_gpu, @object .size a_gpu, 8 a_gpu: .zero 8 .globl c_cpu .align 8 .type c_cpu, @object .size c_cpu, 8 c_cpu: .zero 8 .globl b_cpu .align 8 .type b_cpu, @object .size b_cpu, 8 b_cpu: .zero 8 .globl a_cpu .align 8 .type a_cpu, @object .size a_cpu, 8 a_cpu: .zero 8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "othermain.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl _Z18__device_stub__addPfS_S_i # -- Begin function _Z18__device_stub__addPfS_S_i .p2align 4, 0x90 .type _Z18__device_stub__addPfS_S_i,@function _Z18__device_stub__addPfS_S_i: # @_Z18__device_stub__addPfS_S_i .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movq %rdx, 56(%rsp) movl %ecx, 4(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 4(%rsp), %rax movq %rax, 104(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z3addPfS_S_i, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $136, %rsp .cfi_adjust_cfa_offset -136 retq .Lfunc_end0: .size _Z18__device_stub__addPfS_S_i, .Lfunc_end0-_Z18__device_stub__addPfS_S_i .cfi_endproc # -- End function .globl main # -- Begin function main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 subq $112, %rsp .cfi_def_cfa_offset 128 .cfi_offset %rbx, -16 movl $40000, %edi # imm = 0x9C40 callq malloc movq %rax, a_cpu(%rip) movl $40000, %edi # imm = 0x9C40 callq malloc movq %rax, b_cpu(%rip) movl $40000, %edi # imm = 0x9C40 callq malloc movq %rax, c_cpu(%rip) movl $a_gpu, %edi movl $40000, %esi # imm = 0x9C40 callq hipMalloc movl $b_gpu, %edi movl $40000, %esi # imm = 0x9C40 callq hipMalloc movl $c_gpu, %edi movl $40000, %esi # imm = 0x9C40 callq hipMalloc movq a_cpu(%rip), %rax movq b_cpu(%rip), %rcx movl $-1, %edx xorl %edi, %edi movq c_cpu(%rip), %rsi .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 xorps %xmm0, %xmm0 cvtsi2ss %edx, %xmm0 leaq 2(%rdi), %r8 xorps %xmm1, %xmm1 cvtsi2ss %r8d, %xmm1 movss %xmm0, 4(%rax,%rdi,2) movss %xmm1, 4(%rcx,%rdi,2) addss 4(%rax,%rdi,2), %xmm1 movss %xmm1, 4(%rsi,%rdi,2) decl %edx movq %r8, %rdi cmpq $20000, %r8 # imm = 0x4E20 jne .LBB1_1 # %bb.2: movq a_gpu(%rip), %rdi movq a_cpu(%rip), %rsi movl $40000, %edx # imm = 0x9C40 movl $1, %ecx callq hipMemcpy movq b_gpu(%rip), %rdi movq b_cpu(%rip), %rsi movl $40000, %edx # imm = 0x9C40 movl $1, %ecx callq hipMemcpy movq c_gpu(%rip), %rdi movq c_cpu(%rip), %rsi movl $40000, %edx # imm = 0x9C40 movl $1, %ecx callq hipMemcpy movabsq $4294967297, %rdi # imm = 0x100000001 leaq 9(%rdi), %rdx movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_4 # %bb.3: movq a_gpu(%rip), %rax movq b_gpu(%rip), %rcx movq c_gpu(%rip), %rdx movq %rax, 72(%rsp) movq %rcx, 64(%rsp) movq %rdx, 56(%rsp) movl $10000, 4(%rsp) # imm = 0x2710 leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 4(%rsp), %rax movq %rax, 104(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z3addPfS_S_i, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_4: movq c_cpu(%rip), %rdi movq c_gpu(%rip), %rsi movl $40000, %edx # imm = 0x9C40 movl $2, %ecx callq hipMemcpy movl $1, %ebx .p2align 4, 0x90 .LBB1_5: # =>This Inner Loop Header: Depth=1 movq c_cpu(%rip), %rax movss (%rax,%rbx,4), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str, %edi movb $1, %al callq printf incq %rbx cmpq $10001, %rbx # imm = 0x2711 jne .LBB1_5 # %bb.6: movq a_gpu(%rip), %rdi callq hipFree movq b_gpu(%rip), %rdi callq hipFree movq c_gpu(%rip), %rdi callq hipFree movq a_cpu(%rip), %rdi callq free movq b_cpu(%rip), %rdi callq free movq c_cpu(%rip), %rdi callq free movq stdin(%rip), %rdi callq getc xorl %eax, %eax addq $112, %rsp .cfi_def_cfa_offset 16 popq %rbx .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z3addPfS_S_i, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type a_cpu,@object # @a_cpu .bss .globl a_cpu .p2align 3, 0x0 a_cpu: .quad 0 .size a_cpu, 8 .type b_cpu,@object # @b_cpu .globl b_cpu .p2align 3, 0x0 b_cpu: .quad 0 .size b_cpu, 8 .type c_cpu,@object # @c_cpu .globl c_cpu .p2align 3, 0x0 c_cpu: .quad 0 .size c_cpu, 8 .type a_gpu,@object # @a_gpu .globl a_gpu .p2align 3, 0x0 a_gpu: .quad 0 .size a_gpu, 8 .type b_gpu,@object # @b_gpu .globl b_gpu .p2align 3, 0x0 b_gpu: .quad 0 .size b_gpu, 8 .type c_gpu,@object # @c_gpu .globl c_gpu .p2align 3, 0x0 c_gpu: .quad 0 .size c_gpu, 8 .type _Z3addPfS_S_i,@object # @_Z3addPfS_S_i .section .rodata,"a",@progbits .globl _Z3addPfS_S_i .p2align 3, 0x0 _Z3addPfS_S_i: .quad _Z18__device_stub__addPfS_S_i .size _Z3addPfS_S_i, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "%.0f\n" .size .L.str, 6 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z3addPfS_S_i" .size .L__unnamed_1, 14 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z18__device_stub__addPfS_S_i .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym a_gpu .addrsig_sym b_gpu .addrsig_sym c_gpu .addrsig_sym _Z3addPfS_S_i .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
/********************************************************************************************** Source Code : sharedMemoryReadingSameWord.cu Objective : Example code to demonstrate the different access patterns of float3 array in the global memory the corresponding advantages in terms of the bandwidth that is achievable Descritpion : Example code to demonstrate that while reading the same word by all the threads, there will not be any serialization since all threads are accessing from the same bank the 32-bit word gets broadcasted to all the threads -- hence bandwidth can be comparable to the value got when there were no bank conflicts output: The different bandwidths of the shared memory that are achieved while accessing the same word and the access pattern without any bank conflicts Modified : Aug 2011 Author : RarchK ****************************************************************************************/ #include <stdio.h> #include <cuda.h> #include <float.h> #define BLOCK_SIZE 128 // 128 threads per block #define TRANSFERED_DATA_SIZE 2000000 //2 MB #define NO_OF_PATTERNS 2 #define NTIMES 10 #define MIN(x,y) ((x)<(y)?(x):(y)) #define MAX(x,y) ((x)>(y)?(x):(y)) void printResults(void); //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving accessing the elements with a stride of one 32-bit word // the access pattern causes no bank conflicts // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemAccessWithStride1(void) { __shared__ __attribute__((unused)) float array[BLOCK_SIZE]; int idx = threadIdx.x; float scalar = 5.0f; array[idx] = scalar; } //////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving reading the same word by all the threads // the access pattern is conflict-free // since the word gets broadcasted to all the threads // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemReadSameWord(void) { __shared__ float array[BLOCK_SIZE]; int idx = 3; float __attribute__((unused)) scalar ; scalar = array[idx]; } /////////////////////////////////////////////////////////////////////////////////////////////////// static double avgtime[NO_OF_PATTERNS] = {0}, maxtime[NO_OF_PATTERNS] = {0}, mintime[NO_OF_PATTERNS]; static float bandWidths[NO_OF_PATTERNS] = {0}; //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // the main routene // for timing the two kernals // finding the corresponding band widths // //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// int main(int argc, char* argv[]) { float elapsedTimes[NO_OF_PATTERNS][NTIMES]; cudaEvent_t start,stop; double bytes = sizeof(float) * TRANSFERED_DATA_SIZE; // event creation, which will be used for timing the code cudaEventCreate(&start); cudaEventCreate(&stop); //finding the 1D grid size int gridSize = TRANSFERED_DATA_SIZE/BLOCK_SIZE; if( TRANSFERED_DATA_SIZE % BLOCK_SIZE != 0 ) gridSize += 1; // running each pattern NTIMES for(int k=0; k < NTIMES; k++) { // timing the kernels corresponding to different access patterns // PATTERN 1 cudaEventRecord(start,0); sharedMemAccessWithStride1 <<< gridSize,BLOCK_SIZE >>> (); cudaEventRecord(stop,0); cudaEventSynchronize(stop); cudaEventElapsedTime(&elapsedTimes[0][k],start,stop); // PATTERN 2 cudaEventRecord(start,0); sharedMemReadSameWord <<< gridSize,BLOCK_SIZE >>> (); cudaEventRecord(stop,0); cudaEventSynchronize(stop); cudaEventElapsedTime(&elapsedTimes[1][k],start,stop); } // end of the for loop involving NTIMES // intializing the mintime array for(int i=0; i < NO_OF_PATTERNS;i++) mintime[i] = FLT_MAX; for (int k=1; k < NTIMES; k++) // skiping the first iteration { for (int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i] + elapsedTimes[i][k]; mintime[i] = MIN(mintime[i],elapsedTimes[i][k]); maxtime[i] = MAX(maxtime[i], elapsedTimes[i][k]); } } // calculation of the different band widths that are achieved by different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i]/(double)(NTIMES-1); // finding the average time bandWidths[i] = bytes/mintime[i]; } printResults(); printf("\n\n**** successful termination of the program ****\n\n"); cudaEventDestroy(start); cudaEventDestroy(stop); return 0; } //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // prints the results containig the minimum, maximum, average times taken by the two kernels // the associated maximum bandwidth achieved by the different kernels // /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// void printResults() { printf("Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts\n"); printf("The transfered data size (Bytes): %ld\n",TRANSFERED_DATA_SIZE*sizeof(float)); printf("\n-------------------------------------------------------------------------------------------------------------------------------\n"); printf("Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)\n"); printf("-------------------------------------------------------------------------------------------------------------------------------\n"); // printing the results for different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { switch(i) { case 0: printf("Stride of one 32-bit word "); break; case 1: printf("Reading the same 32-bit word"); break; } printf("\t %.6f \t\t %f \t\t %f \t\t %f\n",bandWidths[i]/1000000,avgtime[i],mintime[i],maxtime[i]); } printf("\n ------------------------------------------------------------------------------------------------------------------------------\n"); } ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
code for sm_80 Function : _Z21sharedMemReadSameWordv .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0020*/ BRA 0x20; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0030*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0040*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0050*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0060*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0070*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0080*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0090*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ .......... Function : _Z26sharedMemAccessWithStride1v .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0020*/ BRA 0x20; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0030*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0040*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0050*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0060*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0070*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0080*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0090*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
/********************************************************************************************** Source Code : sharedMemoryReadingSameWord.cu Objective : Example code to demonstrate the different access patterns of float3 array in the global memory the corresponding advantages in terms of the bandwidth that is achievable Descritpion : Example code to demonstrate that while reading the same word by all the threads, there will not be any serialization since all threads are accessing from the same bank the 32-bit word gets broadcasted to all the threads -- hence bandwidth can be comparable to the value got when there were no bank conflicts output: The different bandwidths of the shared memory that are achieved while accessing the same word and the access pattern without any bank conflicts Modified : Aug 2011 Author : RarchK ****************************************************************************************/ #include <stdio.h> #include <cuda.h> #include <float.h> #define BLOCK_SIZE 128 // 128 threads per block #define TRANSFERED_DATA_SIZE 2000000 //2 MB #define NO_OF_PATTERNS 2 #define NTIMES 10 #define MIN(x,y) ((x)<(y)?(x):(y)) #define MAX(x,y) ((x)>(y)?(x):(y)) void printResults(void); //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving accessing the elements with a stride of one 32-bit word // the access pattern causes no bank conflicts // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemAccessWithStride1(void) { __shared__ __attribute__((unused)) float array[BLOCK_SIZE]; int idx = threadIdx.x; float scalar = 5.0f; array[idx] = scalar; } //////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving reading the same word by all the threads // the access pattern is conflict-free // since the word gets broadcasted to all the threads // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemReadSameWord(void) { __shared__ float array[BLOCK_SIZE]; int idx = 3; float __attribute__((unused)) scalar ; scalar = array[idx]; } /////////////////////////////////////////////////////////////////////////////////////////////////// static double avgtime[NO_OF_PATTERNS] = {0}, maxtime[NO_OF_PATTERNS] = {0}, mintime[NO_OF_PATTERNS]; static float bandWidths[NO_OF_PATTERNS] = {0}; //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // the main routene // for timing the two kernals // finding the corresponding band widths // //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// int main(int argc, char* argv[]) { float elapsedTimes[NO_OF_PATTERNS][NTIMES]; cudaEvent_t start,stop; double bytes = sizeof(float) * TRANSFERED_DATA_SIZE; // event creation, which will be used for timing the code cudaEventCreate(&start); cudaEventCreate(&stop); //finding the 1D grid size int gridSize = TRANSFERED_DATA_SIZE/BLOCK_SIZE; if( TRANSFERED_DATA_SIZE % BLOCK_SIZE != 0 ) gridSize += 1; // running each pattern NTIMES for(int k=0; k < NTIMES; k++) { // timing the kernels corresponding to different access patterns // PATTERN 1 cudaEventRecord(start,0); sharedMemAccessWithStride1 <<< gridSize,BLOCK_SIZE >>> (); cudaEventRecord(stop,0); cudaEventSynchronize(stop); cudaEventElapsedTime(&elapsedTimes[0][k],start,stop); // PATTERN 2 cudaEventRecord(start,0); sharedMemReadSameWord <<< gridSize,BLOCK_SIZE >>> (); cudaEventRecord(stop,0); cudaEventSynchronize(stop); cudaEventElapsedTime(&elapsedTimes[1][k],start,stop); } // end of the for loop involving NTIMES // intializing the mintime array for(int i=0; i < NO_OF_PATTERNS;i++) mintime[i] = FLT_MAX; for (int k=1; k < NTIMES; k++) // skiping the first iteration { for (int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i] + elapsedTimes[i][k]; mintime[i] = MIN(mintime[i],elapsedTimes[i][k]); maxtime[i] = MAX(maxtime[i], elapsedTimes[i][k]); } } // calculation of the different band widths that are achieved by different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i]/(double)(NTIMES-1); // finding the average time bandWidths[i] = bytes/mintime[i]; } printResults(); printf("\n\n**** successful termination of the program ****\n\n"); cudaEventDestroy(start); cudaEventDestroy(stop); return 0; } //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // prints the results containig the minimum, maximum, average times taken by the two kernels // the associated maximum bandwidth achieved by the different kernels // /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// void printResults() { printf("Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts\n"); printf("The transfered data size (Bytes): %ld\n",TRANSFERED_DATA_SIZE*sizeof(float)); printf("\n-------------------------------------------------------------------------------------------------------------------------------\n"); printf("Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)\n"); printf("-------------------------------------------------------------------------------------------------------------------------------\n"); // printing the results for different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { switch(i) { case 0: printf("Stride of one 32-bit word "); break; case 1: printf("Reading the same 32-bit word"); break; } printf("\t %.6f \t\t %f \t\t %f \t\t %f\n",bandWidths[i]/1000000,avgtime[i],mintime[i],maxtime[i]); } printf("\n ------------------------------------------------------------------------------------------------------------------------------\n"); } ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
.file "tmpxft_001402d6_00000000-6_sharedMemoryReadingSameWord.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2061: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2061: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC0: .string "Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts\n" .align 8 .LC1: .string "The transfered data size (Bytes): %ld\n" .align 8 .LC2: .string "\n-------------------------------------------------------------------------------------------------------------------------------\n" .align 8 .LC3: .string "Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)\n" .align 8 .LC4: .string "-------------------------------------------------------------------------------------------------------------------------------\n" .section .rodata.str1.1,"aMS",@progbits,1 .LC5: .string "Stride of one 32-bit word " .LC7: .string "\t %.6f \t\t %f \t\t %f \t\t %f\n" .LC8: .string "Reading the same 32-bit word" .section .rodata.str1.8 .align 8 .LC9: .string "\n ------------------------------------------------------------------------------------------------------------------------------\n" .text .globl _Z12printResultsv .type _Z12printResultsv, @function _Z12printResultsv: .LFB2058: .cfi_startproc endbr64 pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset 3, -16 leaq .LC0(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl $8000000, %edx leaq .LC1(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq .LC2(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq .LC3(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq .LC4(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq .LC5(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movss _ZL10bandWidths(%rip), %xmm0 divss .LC6(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 movsd _ZL7maxtime(%rip), %xmm3 movsd _ZL7mintime(%rip), %xmm2 movsd _ZL7avgtime(%rip), %xmm1 leaq .LC7(%rip), %rbx movq %rbx, %rsi movl $2, %edi movl $4, %eax call __printf_chk@PLT leaq .LC8(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movss 4+_ZL10bandWidths(%rip), %xmm0 divss .LC6(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 movsd 8+_ZL7maxtime(%rip), %xmm3 movsd 8+_ZL7mintime(%rip), %xmm2 movsd 8+_ZL7avgtime(%rip), %xmm1 movq %rbx, %rsi movl $2, %edi movl $4, %eax call __printf_chk@PLT leaq .LC9(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT popq %rbx .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2058: .size _Z12printResultsv, .-_Z12printResultsv .globl _Z45__device_stub__Z26sharedMemAccessWithStride1vv .type _Z45__device_stub__Z26sharedMemAccessWithStride1vv, @function _Z45__device_stub__Z26sharedMemAccessWithStride1vv: .LFB2083: .cfi_startproc endbr64 subq $88, %rsp .cfi_def_cfa_offset 96 movq %fs:40, %rax movq %rax, 72(%rsp) xorl %eax, %eax movl $1, 16(%rsp) movl $1, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) leaq 8(%rsp), %rcx movq %rsp, %rdx leaq 28(%rsp), %rsi leaq 16(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L9 .L5: movq 72(%rsp), %rax subq %fs:40, %rax jne .L10 addq $88, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L9: .cfi_restore_state pushq 8(%rsp) .cfi_def_cfa_offset 104 pushq 8(%rsp) .cfi_def_cfa_offset 112 leaq 80(%rsp), %r9 movq 44(%rsp), %rcx movl 52(%rsp), %r8d movq 32(%rsp), %rsi movl 40(%rsp), %edx leaq _Z26sharedMemAccessWithStride1v(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 96 jmp .L5 .L10: call __stack_chk_fail@PLT .cfi_endproc .LFE2083: .size _Z45__device_stub__Z26sharedMemAccessWithStride1vv, .-_Z45__device_stub__Z26sharedMemAccessWithStride1vv .globl _Z26sharedMemAccessWithStride1v .type _Z26sharedMemAccessWithStride1v, @function _Z26sharedMemAccessWithStride1v: .LFB2084: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z45__device_stub__Z26sharedMemAccessWithStride1vv addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2084: .size _Z26sharedMemAccessWithStride1v, .-_Z26sharedMemAccessWithStride1v .globl _Z40__device_stub__Z21sharedMemReadSameWordvv .type _Z40__device_stub__Z21sharedMemReadSameWordvv, @function _Z40__device_stub__Z21sharedMemReadSameWordvv: .LFB2085: .cfi_startproc endbr64 subq $88, %rsp .cfi_def_cfa_offset 96 movq %fs:40, %rax movq %rax, 72(%rsp) xorl %eax, %eax movl $1, 16(%rsp) movl $1, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) leaq 8(%rsp), %rcx movq %rsp, %rdx leaq 28(%rsp), %rsi leaq 16(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L17 .L13: movq 72(%rsp), %rax subq %fs:40, %rax jne .L18 addq $88, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L17: .cfi_restore_state pushq 8(%rsp) .cfi_def_cfa_offset 104 pushq 8(%rsp) .cfi_def_cfa_offset 112 leaq 80(%rsp), %r9 movq 44(%rsp), %rcx movl 52(%rsp), %r8d movq 32(%rsp), %rsi movl 40(%rsp), %edx leaq _Z21sharedMemReadSameWordv(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 96 jmp .L13 .L18: call __stack_chk_fail@PLT .cfi_endproc .LFE2085: .size _Z40__device_stub__Z21sharedMemReadSameWordvv, .-_Z40__device_stub__Z21sharedMemReadSameWordvv .globl _Z21sharedMemReadSameWordv .type _Z21sharedMemReadSameWordv, @function _Z21sharedMemReadSameWordv: .LFB2086: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z40__device_stub__Z21sharedMemReadSameWordvv addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2086: .size _Z21sharedMemReadSameWordv, .-_Z21sharedMemReadSameWordv .section .rodata.str1.8 .align 8 .LC13: .string "\n\n**** successful termination of the program ****\n\n" .text .globl main .type main, @function main: .LFB2057: .cfi_startproc endbr64 pushq %r12 .cfi_def_cfa_offset 16 .cfi_offset 12, -16 pushq %rbp .cfi_def_cfa_offset 24 .cfi_offset 6, -24 pushq %rbx .cfi_def_cfa_offset 32 .cfi_offset 3, -32 subq $144, %rsp .cfi_def_cfa_offset 176 movq %fs:40, %rax movq %rax, 136(%rsp) xorl %eax, %eax leaq 8(%rsp), %rdi call cudaEventCreate@PLT leaq 16(%rsp), %rdi call cudaEventCreate@PLT leaq 48(%rsp), %rbp leaq 88(%rsp), %rbx leaq 128(%rsp), %r12 jmp .L24 .L38: call _Z45__device_stub__Z26sharedMemAccessWithStride1vv jmp .L22 .L23: movl $0, %esi movq 16(%rsp), %rdi call cudaEventRecord@PLT movq 16(%rsp), %rdi call cudaEventSynchronize@PLT movq 16(%rsp), %rdx movq 8(%rsp), %rsi movq %rbx, %rdi call cudaEventElapsedTime@PLT addq $4, %rbx cmpq %rbx, %r12 je .L37 .L24: movl $0, %esi movq 8(%rsp), %rdi call cudaEventRecord@PLT movl $128, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $15625, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $0, %r9d movl $0, %r8d movq 36(%rsp), %rdx movl $1, %ecx movq 24(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L38 .L22: movl $0, %esi movq 16(%rsp), %rdi call cudaEventRecord@PLT movq 16(%rsp), %rdi call cudaEventSynchronize@PLT leaq -40(%rbx), %rdi movq 16(%rsp), %rdx movq 8(%rsp), %rsi call cudaEventElapsedTime@PLT movl $0, %esi movq 8(%rsp), %rdi call cudaEventRecord@PLT movl $128, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $15625, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $0, %r9d movl $0, %r8d movq 36(%rsp), %rdx movl $1, %ecx movq 24(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax jne .L23 call _Z40__device_stub__Z21sharedMemReadSameWordvv jmp .L23 .L37: leaq 52(%rsp), %rax leaq 40(%rbp), %rdx movsd _ZL7avgtime(%rip), %xmm2 movsd _ZL7maxtime(%rip), %xmm6 movsd 8+_ZL7avgtime(%rip), %xmm1 movsd 8+_ZL7maxtime(%rip), %xmm5 movsd .LC10(%rip), %xmm3 movapd %xmm3, %xmm4 .L25: pxor %xmm0, %xmm0 cvtss2sd (%rax), %xmm0 addsd %xmm0, %xmm2 minsd %xmm0, %xmm3 maxsd %xmm0, %xmm6 pxor %xmm0, %xmm0 cvtss2sd 40(%rax), %xmm0 addsd %xmm0, %xmm1 minsd %xmm0, %xmm4 maxsd %xmm0, %xmm5 addq $4, %rax cmpq %rax, %rdx jne .L25 movsd %xmm3, _ZL7mintime(%rip) movsd %xmm6, _ZL7maxtime(%rip) movsd %xmm4, 8+_ZL7mintime(%rip) movsd %xmm5, 8+_ZL7maxtime(%rip) movsd .LC11(%rip), %xmm5 divsd %xmm5, %xmm2 movsd %xmm2, _ZL7avgtime(%rip) movsd .LC12(%rip), %xmm0 movapd %xmm0, %xmm2 divsd %xmm3, %xmm2 cvtsd2ss %xmm2, %xmm2 movss %xmm2, _ZL10bandWidths(%rip) divsd %xmm5, %xmm1 movsd %xmm1, 8+_ZL7avgtime(%rip) divsd %xmm4, %xmm0 cvtsd2ss %xmm0, %xmm0 movss %xmm0, 4+_ZL10bandWidths(%rip) call _Z12printResultsv leaq .LC13(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq 8(%rsp), %rdi call cudaEventDestroy@PLT movq 16(%rsp), %rdi call cudaEventDestroy@PLT movq 136(%rsp), %rax subq %fs:40, %rax jne .L39 movl $0, %eax addq $144, %rsp .cfi_remember_state .cfi_def_cfa_offset 32 popq %rbx .cfi_def_cfa_offset 24 popq %rbp .cfi_def_cfa_offset 16 popq %r12 .cfi_def_cfa_offset 8 ret .L39: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1 .LC14: .string "_Z21sharedMemReadSameWordv" .section .rodata.str1.8 .align 8 .LC15: .string "_Z26sharedMemAccessWithStride1v" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2088: .cfi_startproc endbr64 pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset 3, -16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rbx movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC14(%rip), %rdx movq %rdx, %rcx leaq _Z21sharedMemReadSameWordv(%rip), %rsi movq %rax, %rdi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC15(%rip), %rdx movq %rdx, %rcx leaq _Z26sharedMemAccessWithStride1v(%rip), %rsi movq %rbx, %rdi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT popq %rbx .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2088: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .local _ZL10bandWidths .comm _ZL10bandWidths,8,8 .local _ZL7mintime .comm _ZL7mintime,16,16 .local _ZL7maxtime .comm _ZL7maxtime,16,16 .local _ZL7avgtime .comm _ZL7avgtime,16,16 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC6: .long 1232348160 .section .rodata.cst8,"aM",@progbits,8 .align 8 .LC10: .long -536870912 .long 1206910975 .align 8 .LC11: .long 0 .long 1075970048 .align 8 .LC12: .long 0 .long 1096713344 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
/********************************************************************************************** Source Code : sharedMemoryReadingSameWord.cu Objective : Example code to demonstrate the different access patterns of float3 array in the global memory the corresponding advantages in terms of the bandwidth that is achievable Descritpion : Example code to demonstrate that while reading the same word by all the threads, there will not be any serialization since all threads are accessing from the same bank the 32-bit word gets broadcasted to all the threads -- hence bandwidth can be comparable to the value got when there were no bank conflicts output: The different bandwidths of the shared memory that are achieved while accessing the same word and the access pattern without any bank conflicts Modified : Aug 2011 Author : RarchK ****************************************************************************************/ #include <stdio.h> #include <cuda.h> #include <float.h> #define BLOCK_SIZE 128 // 128 threads per block #define TRANSFERED_DATA_SIZE 2000000 //2 MB #define NO_OF_PATTERNS 2 #define NTIMES 10 #define MIN(x,y) ((x)<(y)?(x):(y)) #define MAX(x,y) ((x)>(y)?(x):(y)) void printResults(void); //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving accessing the elements with a stride of one 32-bit word // the access pattern causes no bank conflicts // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemAccessWithStride1(void) { __shared__ __attribute__((unused)) float array[BLOCK_SIZE]; int idx = threadIdx.x; float scalar = 5.0f; array[idx] = scalar; } //////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving reading the same word by all the threads // the access pattern is conflict-free // since the word gets broadcasted to all the threads // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemReadSameWord(void) { __shared__ float array[BLOCK_SIZE]; int idx = 3; float __attribute__((unused)) scalar ; scalar = array[idx]; } /////////////////////////////////////////////////////////////////////////////////////////////////// static double avgtime[NO_OF_PATTERNS] = {0}, maxtime[NO_OF_PATTERNS] = {0}, mintime[NO_OF_PATTERNS]; static float bandWidths[NO_OF_PATTERNS] = {0}; //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // the main routene // for timing the two kernals // finding the corresponding band widths // //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// int main(int argc, char* argv[]) { float elapsedTimes[NO_OF_PATTERNS][NTIMES]; cudaEvent_t start,stop; double bytes = sizeof(float) * TRANSFERED_DATA_SIZE; // event creation, which will be used for timing the code cudaEventCreate(&start); cudaEventCreate(&stop); //finding the 1D grid size int gridSize = TRANSFERED_DATA_SIZE/BLOCK_SIZE; if( TRANSFERED_DATA_SIZE % BLOCK_SIZE != 0 ) gridSize += 1; // running each pattern NTIMES for(int k=0; k < NTIMES; k++) { // timing the kernels corresponding to different access patterns // PATTERN 1 cudaEventRecord(start,0); sharedMemAccessWithStride1 <<< gridSize,BLOCK_SIZE >>> (); cudaEventRecord(stop,0); cudaEventSynchronize(stop); cudaEventElapsedTime(&elapsedTimes[0][k],start,stop); // PATTERN 2 cudaEventRecord(start,0); sharedMemReadSameWord <<< gridSize,BLOCK_SIZE >>> (); cudaEventRecord(stop,0); cudaEventSynchronize(stop); cudaEventElapsedTime(&elapsedTimes[1][k],start,stop); } // end of the for loop involving NTIMES // intializing the mintime array for(int i=0; i < NO_OF_PATTERNS;i++) mintime[i] = FLT_MAX; for (int k=1; k < NTIMES; k++) // skiping the first iteration { for (int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i] + elapsedTimes[i][k]; mintime[i] = MIN(mintime[i],elapsedTimes[i][k]); maxtime[i] = MAX(maxtime[i], elapsedTimes[i][k]); } } // calculation of the different band widths that are achieved by different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i]/(double)(NTIMES-1); // finding the average time bandWidths[i] = bytes/mintime[i]; } printResults(); printf("\n\n**** successful termination of the program ****\n\n"); cudaEventDestroy(start); cudaEventDestroy(stop); return 0; } //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // prints the results containig the minimum, maximum, average times taken by the two kernels // the associated maximum bandwidth achieved by the different kernels // /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// void printResults() { printf("Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts\n"); printf("The transfered data size (Bytes): %ld\n",TRANSFERED_DATA_SIZE*sizeof(float)); printf("\n-------------------------------------------------------------------------------------------------------------------------------\n"); printf("Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)\n"); printf("-------------------------------------------------------------------------------------------------------------------------------\n"); // printing the results for different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { switch(i) { case 0: printf("Stride of one 32-bit word "); break; case 1: printf("Reading the same 32-bit word"); break; } printf("\t %.6f \t\t %f \t\t %f \t\t %f\n",bandWidths[i]/1000000,avgtime[i],mintime[i],maxtime[i]); } printf("\n ------------------------------------------------------------------------------------------------------------------------------\n"); } ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
/********************************************************************************************** Source Code : sharedMemoryReadingSameWord.cu Objective : Example code to demonstrate the different access patterns of float3 array in the global memory the corresponding advantages in terms of the bandwidth that is achievable Descritpion : Example code to demonstrate that while reading the same word by all the threads, there will not be any serialization since all threads are accessing from the same bank the 32-bit word gets broadcasted to all the threads -- hence bandwidth can be comparable to the value got when there were no bank conflicts output: The different bandwidths of the shared memory that are achieved while accessing the same word and the access pattern without any bank conflicts Modified : Aug 2011 Author : RarchK ****************************************************************************************/ #include <stdio.h> #include <hip/hip_runtime.h> #include <float.h> #define BLOCK_SIZE 128 // 128 threads per block #define TRANSFERED_DATA_SIZE 2000000 //2 MB #define NO_OF_PATTERNS 2 #define NTIMES 10 #define MIN(x,y) ((x)<(y)?(x):(y)) #define MAX(x,y) ((x)>(y)?(x):(y)) void printResults(void); //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving accessing the elements with a stride of one 32-bit word // the access pattern causes no bank conflicts // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemAccessWithStride1(void) { __shared__ __attribute__((unused)) float array[BLOCK_SIZE]; int idx = threadIdx.x; float scalar = 5.0f; array[idx] = scalar; } //////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving reading the same word by all the threads // the access pattern is conflict-free // since the word gets broadcasted to all the threads // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemReadSameWord(void) { __shared__ float array[BLOCK_SIZE]; int idx = 3; float __attribute__((unused)) scalar ; scalar = array[idx]; } /////////////////////////////////////////////////////////////////////////////////////////////////// static double avgtime[NO_OF_PATTERNS] = {0}, maxtime[NO_OF_PATTERNS] = {0}, mintime[NO_OF_PATTERNS]; static float bandWidths[NO_OF_PATTERNS] = {0}; //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // the main routene // for timing the two kernals // finding the corresponding band widths // //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// int main(int argc, char* argv[]) { float elapsedTimes[NO_OF_PATTERNS][NTIMES]; hipEvent_t start,stop; double bytes = sizeof(float) * TRANSFERED_DATA_SIZE; // event creation, which will be used for timing the code hipEventCreate(&start); hipEventCreate(&stop); //finding the 1D grid size int gridSize = TRANSFERED_DATA_SIZE/BLOCK_SIZE; if( TRANSFERED_DATA_SIZE % BLOCK_SIZE != 0 ) gridSize += 1; // running each pattern NTIMES for(int k=0; k < NTIMES; k++) { // timing the kernels corresponding to different access patterns // PATTERN 1 hipEventRecord(start,0); sharedMemAccessWithStride1 <<< gridSize,BLOCK_SIZE >>> (); hipEventRecord(stop,0); hipEventSynchronize(stop); hipEventElapsedTime(&elapsedTimes[0][k],start,stop); // PATTERN 2 hipEventRecord(start,0); sharedMemReadSameWord <<< gridSize,BLOCK_SIZE >>> (); hipEventRecord(stop,0); hipEventSynchronize(stop); hipEventElapsedTime(&elapsedTimes[1][k],start,stop); } // end of the for loop involving NTIMES // intializing the mintime array for(int i=0; i < NO_OF_PATTERNS;i++) mintime[i] = FLT_MAX; for (int k=1; k < NTIMES; k++) // skiping the first iteration { for (int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i] + elapsedTimes[i][k]; mintime[i] = MIN(mintime[i],elapsedTimes[i][k]); maxtime[i] = MAX(maxtime[i], elapsedTimes[i][k]); } } // calculation of the different band widths that are achieved by different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i]/(double)(NTIMES-1); // finding the average time bandWidths[i] = bytes/mintime[i]; } printResults(); printf("\n\n**** successful termination of the program ****\n\n"); hipEventDestroy(start); hipEventDestroy(stop); return 0; } //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // prints the results containig the minimum, maximum, average times taken by the two kernels // the associated maximum bandwidth achieved by the different kernels // /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// void printResults() { printf("Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts\n"); printf("The transfered data size (Bytes): %ld\n",TRANSFERED_DATA_SIZE*sizeof(float)); printf("\n-------------------------------------------------------------------------------------------------------------------------------\n"); printf("Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)\n"); printf("-------------------------------------------------------------------------------------------------------------------------------\n"); // printing the results for different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { switch(i) { case 0: printf("Stride of one 32-bit word "); break; case 1: printf("Reading the same 32-bit word"); break; } printf("\t %.6f \t\t %f \t\t %f \t\t %f\n",bandWidths[i]/1000000,avgtime[i],mintime[i],maxtime[i]); } printf("\n ------------------------------------------------------------------------------------------------------------------------------\n"); } ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
/********************************************************************************************** Source Code : sharedMemoryReadingSameWord.cu Objective : Example code to demonstrate the different access patterns of float3 array in the global memory the corresponding advantages in terms of the bandwidth that is achievable Descritpion : Example code to demonstrate that while reading the same word by all the threads, there will not be any serialization since all threads are accessing from the same bank the 32-bit word gets broadcasted to all the threads -- hence bandwidth can be comparable to the value got when there were no bank conflicts output: The different bandwidths of the shared memory that are achieved while accessing the same word and the access pattern without any bank conflicts Modified : Aug 2011 Author : RarchK ****************************************************************************************/ #include <stdio.h> #include <hip/hip_runtime.h> #include <float.h> #define BLOCK_SIZE 128 // 128 threads per block #define TRANSFERED_DATA_SIZE 2000000 //2 MB #define NO_OF_PATTERNS 2 #define NTIMES 10 #define MIN(x,y) ((x)<(y)?(x):(y)) #define MAX(x,y) ((x)>(y)?(x):(y)) void printResults(void); //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving accessing the elements with a stride of one 32-bit word // the access pattern causes no bank conflicts // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemAccessWithStride1(void) { __shared__ __attribute__((unused)) float array[BLOCK_SIZE]; int idx = threadIdx.x; float scalar = 5.0f; array[idx] = scalar; } //////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving reading the same word by all the threads // the access pattern is conflict-free // since the word gets broadcasted to all the threads // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemReadSameWord(void) { __shared__ float array[BLOCK_SIZE]; int idx = 3; float __attribute__((unused)) scalar ; scalar = array[idx]; } /////////////////////////////////////////////////////////////////////////////////////////////////// static double avgtime[NO_OF_PATTERNS] = {0}, maxtime[NO_OF_PATTERNS] = {0}, mintime[NO_OF_PATTERNS]; static float bandWidths[NO_OF_PATTERNS] = {0}; //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // the main routene // for timing the two kernals // finding the corresponding band widths // //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// int main(int argc, char* argv[]) { float elapsedTimes[NO_OF_PATTERNS][NTIMES]; hipEvent_t start,stop; double bytes = sizeof(float) * TRANSFERED_DATA_SIZE; // event creation, which will be used for timing the code hipEventCreate(&start); hipEventCreate(&stop); //finding the 1D grid size int gridSize = TRANSFERED_DATA_SIZE/BLOCK_SIZE; if( TRANSFERED_DATA_SIZE % BLOCK_SIZE != 0 ) gridSize += 1; // running each pattern NTIMES for(int k=0; k < NTIMES; k++) { // timing the kernels corresponding to different access patterns // PATTERN 1 hipEventRecord(start,0); sharedMemAccessWithStride1 <<< gridSize,BLOCK_SIZE >>> (); hipEventRecord(stop,0); hipEventSynchronize(stop); hipEventElapsedTime(&elapsedTimes[0][k],start,stop); // PATTERN 2 hipEventRecord(start,0); sharedMemReadSameWord <<< gridSize,BLOCK_SIZE >>> (); hipEventRecord(stop,0); hipEventSynchronize(stop); hipEventElapsedTime(&elapsedTimes[1][k],start,stop); } // end of the for loop involving NTIMES // intializing the mintime array for(int i=0; i < NO_OF_PATTERNS;i++) mintime[i] = FLT_MAX; for (int k=1; k < NTIMES; k++) // skiping the first iteration { for (int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i] + elapsedTimes[i][k]; mintime[i] = MIN(mintime[i],elapsedTimes[i][k]); maxtime[i] = MAX(maxtime[i], elapsedTimes[i][k]); } } // calculation of the different band widths that are achieved by different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i]/(double)(NTIMES-1); // finding the average time bandWidths[i] = bytes/mintime[i]; } printResults(); printf("\n\n**** successful termination of the program ****\n\n"); hipEventDestroy(start); hipEventDestroy(stop); return 0; } //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // prints the results containig the minimum, maximum, average times taken by the two kernels // the associated maximum bandwidth achieved by the different kernels // /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// void printResults() { printf("Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts\n"); printf("The transfered data size (Bytes): %ld\n",TRANSFERED_DATA_SIZE*sizeof(float)); printf("\n-------------------------------------------------------------------------------------------------------------------------------\n"); printf("Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)\n"); printf("-------------------------------------------------------------------------------------------------------------------------------\n"); // printing the results for different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { switch(i) { case 0: printf("Stride of one 32-bit word "); break; case 1: printf("Reading the same 32-bit word"); break; } printf("\t %.6f \t\t %f \t\t %f \t\t %f\n",bandWidths[i]/1000000,avgtime[i],mintime[i],maxtime[i]); } printf("\n ------------------------------------------------------------------------------------------------------------------------------\n"); } ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z26sharedMemAccessWithStride1v .globl _Z26sharedMemAccessWithStride1v .p2align 8 .type _Z26sharedMemAccessWithStride1v,@function _Z26sharedMemAccessWithStride1v: s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z26sharedMemAccessWithStride1v .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 0 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 0 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 1 .amdhsa_next_free_sgpr 1 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z26sharedMemAccessWithStride1v, .Lfunc_end0-_Z26sharedMemAccessWithStride1v .section .AMDGPU.csdata,"",@progbits .text .protected _Z21sharedMemReadSameWordv .globl _Z21sharedMemReadSameWordv .p2align 8 .type _Z21sharedMemReadSameWordv,@function _Z21sharedMemReadSameWordv: s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z21sharedMemReadSameWordv .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 0 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 0 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 1 .amdhsa_next_free_sgpr 1 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end1: .size _Z21sharedMemReadSameWordv, .Lfunc_end1-_Z21sharedMemReadSameWordv .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: [] .group_segment_fixed_size: 0 .kernarg_segment_align: 4 .kernarg_segment_size: 0 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z26sharedMemAccessWithStride1v .private_segment_fixed_size: 0 .sgpr_count: 0 .sgpr_spill_count: 0 .symbol: _Z26sharedMemAccessWithStride1v.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 0 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 - .args: [] .group_segment_fixed_size: 0 .kernarg_segment_align: 4 .kernarg_segment_size: 0 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z21sharedMemReadSameWordv .private_segment_fixed_size: 0 .sgpr_count: 0 .sgpr_spill_count: 0 .symbol: _Z21sharedMemReadSameWordv.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 0 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
/********************************************************************************************** Source Code : sharedMemoryReadingSameWord.cu Objective : Example code to demonstrate the different access patterns of float3 array in the global memory the corresponding advantages in terms of the bandwidth that is achievable Descritpion : Example code to demonstrate that while reading the same word by all the threads, there will not be any serialization since all threads are accessing from the same bank the 32-bit word gets broadcasted to all the threads -- hence bandwidth can be comparable to the value got when there were no bank conflicts output: The different bandwidths of the shared memory that are achieved while accessing the same word and the access pattern without any bank conflicts Modified : Aug 2011 Author : RarchK ****************************************************************************************/ #include <stdio.h> #include <hip/hip_runtime.h> #include <float.h> #define BLOCK_SIZE 128 // 128 threads per block #define TRANSFERED_DATA_SIZE 2000000 //2 MB #define NO_OF_PATTERNS 2 #define NTIMES 10 #define MIN(x,y) ((x)<(y)?(x):(y)) #define MAX(x,y) ((x)>(y)?(x):(y)) void printResults(void); //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving accessing the elements with a stride of one 32-bit word // the access pattern causes no bank conflicts // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemAccessWithStride1(void) { __shared__ __attribute__((unused)) float array[BLOCK_SIZE]; int idx = threadIdx.x; float scalar = 5.0f; array[idx] = scalar; } //////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////// // // access pattern involving reading the same word by all the threads // the access pattern is conflict-free // since the word gets broadcasted to all the threads // ///////////////////////////////////////////////////////////////////////////////////////////////// __global__ void sharedMemReadSameWord(void) { __shared__ float array[BLOCK_SIZE]; int idx = 3; float __attribute__((unused)) scalar ; scalar = array[idx]; } /////////////////////////////////////////////////////////////////////////////////////////////////// static double avgtime[NO_OF_PATTERNS] = {0}, maxtime[NO_OF_PATTERNS] = {0}, mintime[NO_OF_PATTERNS]; static float bandWidths[NO_OF_PATTERNS] = {0}; //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // the main routene // for timing the two kernals // finding the corresponding band widths // //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// int main(int argc, char* argv[]) { float elapsedTimes[NO_OF_PATTERNS][NTIMES]; hipEvent_t start,stop; double bytes = sizeof(float) * TRANSFERED_DATA_SIZE; // event creation, which will be used for timing the code hipEventCreate(&start); hipEventCreate(&stop); //finding the 1D grid size int gridSize = TRANSFERED_DATA_SIZE/BLOCK_SIZE; if( TRANSFERED_DATA_SIZE % BLOCK_SIZE != 0 ) gridSize += 1; // running each pattern NTIMES for(int k=0; k < NTIMES; k++) { // timing the kernels corresponding to different access patterns // PATTERN 1 hipEventRecord(start,0); sharedMemAccessWithStride1 <<< gridSize,BLOCK_SIZE >>> (); hipEventRecord(stop,0); hipEventSynchronize(stop); hipEventElapsedTime(&elapsedTimes[0][k],start,stop); // PATTERN 2 hipEventRecord(start,0); sharedMemReadSameWord <<< gridSize,BLOCK_SIZE >>> (); hipEventRecord(stop,0); hipEventSynchronize(stop); hipEventElapsedTime(&elapsedTimes[1][k],start,stop); } // end of the for loop involving NTIMES // intializing the mintime array for(int i=0; i < NO_OF_PATTERNS;i++) mintime[i] = FLT_MAX; for (int k=1; k < NTIMES; k++) // skiping the first iteration { for (int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i] + elapsedTimes[i][k]; mintime[i] = MIN(mintime[i],elapsedTimes[i][k]); maxtime[i] = MAX(maxtime[i], elapsedTimes[i][k]); } } // calculation of the different band widths that are achieved by different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { avgtime[i] = avgtime[i]/(double)(NTIMES-1); // finding the average time bandWidths[i] = bytes/mintime[i]; } printResults(); printf("\n\n**** successful termination of the program ****\n\n"); hipEventDestroy(start); hipEventDestroy(stop); return 0; } //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// // // prints the results containig the minimum, maximum, average times taken by the two kernels // the associated maximum bandwidth achieved by the different kernels // /////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// void printResults() { printf("Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts\n"); printf("The transfered data size (Bytes): %ld\n",TRANSFERED_DATA_SIZE*sizeof(float)); printf("\n-------------------------------------------------------------------------------------------------------------------------------\n"); printf("Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)\n"); printf("-------------------------------------------------------------------------------------------------------------------------------\n"); // printing the results for different access patterns for(int i=0; i < NO_OF_PATTERNS; i++) { switch(i) { case 0: printf("Stride of one 32-bit word "); break; case 1: printf("Reading the same 32-bit word"); break; } printf("\t %.6f \t\t %f \t\t %f \t\t %f\n",bandWidths[i]/1000000,avgtime[i],mintime[i],maxtime[i]); } printf("\n ------------------------------------------------------------------------------------------------------------------------------\n"); } ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
.text .file "sharedMemoryReadingSameWord.hip" .globl _Z41__device_stub__sharedMemAccessWithStride1v # -- Begin function _Z41__device_stub__sharedMemAccessWithStride1v .p2align 4, 0x90 .type _Z41__device_stub__sharedMemAccessWithStride1v,@function _Z41__device_stub__sharedMemAccessWithStride1v: # @_Z41__device_stub__sharedMemAccessWithStride1v .cfi_startproc # %bb.0: subq $56, %rsp .cfi_def_cfa_offset 64 leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 48(%rsp), %r9 movl $_Z26sharedMemAccessWithStride1v, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $72, %rsp .cfi_adjust_cfa_offset -72 retq .Lfunc_end0: .size _Z41__device_stub__sharedMemAccessWithStride1v, .Lfunc_end0-_Z41__device_stub__sharedMemAccessWithStride1v .cfi_endproc # -- End function .globl _Z36__device_stub__sharedMemReadSameWordv # -- Begin function _Z36__device_stub__sharedMemReadSameWordv .p2align 4, 0x90 .type _Z36__device_stub__sharedMemReadSameWordv,@function _Z36__device_stub__sharedMemReadSameWordv: # @_Z36__device_stub__sharedMemReadSameWordv .cfi_startproc # %bb.0: subq $56, %rsp .cfi_def_cfa_offset 64 leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 48(%rsp), %r9 movl $_Z21sharedMemReadSameWordv, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $72, %rsp .cfi_adjust_cfa_offset -72 retq .Lfunc_end1: .size _Z36__device_stub__sharedMemReadSameWordv, .Lfunc_end1-_Z36__device_stub__sharedMemReadSameWordv .cfi_endproc # -- End function .section .rodata.cst8,"aM",@progbits,8 .p2align 3, 0x0 # -- Begin function main .LCPI2_0: .quad 0x4022000000000000 # double 9 .LCPI2_1: .quad 0x415e848000000000 # double 8.0E+6 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $168, %rsp .cfi_def_cfa_offset 224 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movabsq $4294967424, %rbx # imm = 0x100000080 leaq 8(%rsp), %rdi callq hipEventCreate movq %rsp, %rdi callq hipEventCreate leaq 80(%rsp), %r14 movl $10, %ebp leaq 15497(%rbx), %r15 leaq 56(%rsp), %r12 leaq 64(%rsp), %r13 jmp .LBB2_1 .p2align 4, 0x90 .LBB2_5: # in Loop: Header=BB2_1 Depth=1 movq (%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq (%rsp), %rdi callq hipEventSynchronize leaq 40(%r14), %rdi movq 8(%rsp), %rsi movq (%rsp), %rdx callq hipEventElapsedTime addq $4, %r14 decq %rbp je .LBB2_6 .LBB2_1: # =>This Inner Loop Header: Depth=1 movq 8(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq %r15, %rdi movl $1, %esi movq %rbx, %rdx movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB2_3 # %bb.2: # in Loop: Header=BB2_1 Depth=1 leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx movq %r12, %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d movl $_Z26sharedMemAccessWithStride1v, %edi movq %r13, %r9 pushq 56(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB2_3: # in Loop: Header=BB2_1 Depth=1 movq (%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq (%rsp), %rdi callq hipEventSynchronize movq 8(%rsp), %rsi movq (%rsp), %rdx movq %r14, %rdi callq hipEventElapsedTime movq 8(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq %r15, %rdi movl $1, %esi movq %rbx, %rdx movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB2_5 # %bb.4: # in Loop: Header=BB2_1 Depth=1 leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx movq %r12, %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d movl $_Z21sharedMemReadSameWordv, %edi movq %r13, %r9 pushq 56(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 jmp .LBB2_5 .LBB2_6: # %.preheader70.preheader movq $-8, %rax movabsq $5183643170566569984, %rcx # imm = 0x47EFFFFFE0000000 .p2align 4, 0x90 .LBB2_7: # %.preheader70 # =>This Inner Loop Header: Depth=1 movq %rcx, _ZL7mintime+8(%rax) addq $8, %rax je .LBB2_7 # %bb.8: # %.preheader68.preheader movl $1, %eax leaq 84(%rsp), %rcx .p2align 4, 0x90 .LBB2_9: # %.preheader68 # =>This Loop Header: Depth=1 # Child Loop BB2_10 Depth 2 movq $-8, %rdx movq %rcx, %rsi .p2align 4, 0x90 .LBB2_10: # Parent Loop BB2_9 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rsi), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movsd _ZL7avgtime+8(%rdx), %xmm1 # xmm1 = mem[0],zero addsd %xmm0, %xmm1 movsd %xmm1, _ZL7avgtime+8(%rdx) movsd _ZL7mintime+8(%rdx), %xmm1 # xmm1 = mem[0],zero minsd %xmm0, %xmm1 movsd %xmm1, _ZL7mintime+8(%rdx) movsd _ZL7maxtime+8(%rdx), %xmm1 # xmm1 = mem[0],zero maxsd %xmm0, %xmm1 movsd %xmm1, _ZL7maxtime+8(%rdx) addq $40, %rsi addq $8, %rdx je .LBB2_10 # %bb.11: # in Loop: Header=BB2_9 Depth=1 incq %rax addq $4, %rcx cmpq $10, %rax jne .LBB2_9 # %bb.12: # %.preheader.preheader movq $-4, %rax movsd .LCPI2_0(%rip), %xmm0 # xmm0 = mem[0],zero movsd .LCPI2_1(%rip), %xmm1 # xmm1 = mem[0],zero .p2align 4, 0x90 .LBB2_13: # %.preheader # =>This Inner Loop Header: Depth=1 movsd _ZL7avgtime+8(%rax,%rax), %xmm2 # xmm2 = mem[0],zero divsd %xmm0, %xmm2 movsd %xmm2, _ZL7avgtime+8(%rax,%rax) movapd %xmm1, %xmm2 divsd _ZL7mintime+8(%rax,%rax), %xmm2 cvtsd2ss %xmm2, %xmm2 movss %xmm2, _ZL10bandWidths+4(%rax) addq $4, %rax je .LBB2_13 # %bb.14: callq _Z12printResultsv movl $.Lstr, %edi callq puts@PLT movq 8(%rsp), %rdi callq hipEventDestroy movq (%rsp), %rdi callq hipEventDestroy xorl %eax, %eax addq $168, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end2: .size main, .Lfunc_end2-main .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function _Z12printResultsv .LCPI3_0: .long 0x49742400 # float 1.0E+6 .text .globl _Z12printResultsv .p2align 4, 0x90 .type _Z12printResultsv,@function _Z12printResultsv: # @_Z12printResultsv .cfi_startproc # %bb.0: pushq %r14 .cfi_def_cfa_offset 16 pushq %rbx .cfi_def_cfa_offset 24 pushq %rax .cfi_def_cfa_offset 32 .cfi_offset %rbx, -24 .cfi_offset %r14, -16 movl $.Lstr.1, %edi callq puts@PLT movl $.L.str.2, %edi movl $8000000, %esi # imm = 0x7A1200 xorl %eax, %eax callq printf movl $.Lstr.2, %edi callq puts@PLT movl $.Lstr.3, %edi callq puts@PLT movl $.Lstr.4, %edi callq puts@PLT movq $-4, %rbx movl $.L.str.6, %r14d .p2align 4, 0x90 .LBB3_1: # =>This Inner Loop Header: Depth=1 cmpq $-4, %rbx movl $.L.str.7, %edi cmoveq %r14, %rdi xorl %eax, %eax callq printf movss _ZL10bandWidths+4(%rbx), %xmm0 # xmm0 = mem[0],zero,zero,zero divss .LCPI3_0(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 movsd _ZL7avgtime+8(%rbx,%rbx), %xmm1 # xmm1 = mem[0],zero movsd _ZL7mintime+8(%rbx,%rbx), %xmm2 # xmm2 = mem[0],zero movsd _ZL7maxtime+8(%rbx,%rbx), %xmm3 # xmm3 = mem[0],zero movl $.L.str.8, %edi movb $4, %al callq printf addq $4, %rbx je .LBB3_1 # %bb.2: movl $.Lstr.5, %edi addq $8, %rsp .cfi_def_cfa_offset 24 popq %rbx .cfi_def_cfa_offset 16 popq %r14 .cfi_def_cfa_offset 8 jmp puts@PLT # TAILCALL .Lfunc_end3: .size _Z12printResultsv, .Lfunc_end3-_Z12printResultsv .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 subq $32, %rsp .cfi_def_cfa_offset 48 .cfi_offset %rbx, -16 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB4_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB4_2: movq __hip_gpubin_handle(%rip), %rbx xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z26sharedMemAccessWithStride1v, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z21sharedMemReadSameWordv, %esi movl $.L__unnamed_2, %edx movl $.L__unnamed_2, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $32, %rsp .cfi_def_cfa_offset 16 popq %rbx .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end4: .size __hip_module_ctor, .Lfunc_end4-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB5_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB5_2: retq .Lfunc_end5: .size __hip_module_dtor, .Lfunc_end5-__hip_module_dtor .cfi_endproc # -- End function .type _Z26sharedMemAccessWithStride1v,@object # @_Z26sharedMemAccessWithStride1v .section .rodata,"a",@progbits .globl _Z26sharedMemAccessWithStride1v .p2align 3, 0x0 _Z26sharedMemAccessWithStride1v: .quad _Z41__device_stub__sharedMemAccessWithStride1v .size _Z26sharedMemAccessWithStride1v, 8 .type _Z21sharedMemReadSameWordv,@object # @_Z21sharedMemReadSameWordv .globl _Z21sharedMemReadSameWordv .p2align 3, 0x0 _Z21sharedMemReadSameWordv: .quad _Z36__device_stub__sharedMemReadSameWordv .size _Z21sharedMemReadSameWordv, 8 .type _ZL7mintime,@object # @_ZL7mintime .local _ZL7mintime .comm _ZL7mintime,16,16 .type _ZL7avgtime,@object # @_ZL7avgtime .local _ZL7avgtime .comm _ZL7avgtime,16,16 .type _ZL7maxtime,@object # @_ZL7maxtime .local _ZL7maxtime .comm _ZL7maxtime,16,16 .type _ZL10bandWidths,@object # @_ZL10bandWidths .local _ZL10bandWidths .comm _ZL10bandWidths,8,4 .type .L.str.2,@object # @.str.2 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.2: .asciz "The transfered data size (Bytes): %ld\n" .size .L.str.2, 39 .type .L.str.6,@object # @.str.6 .L.str.6: .asciz "Stride of one 32-bit word " .size .L.str.6, 29 .type .L.str.7,@object # @.str.7 .L.str.7: .asciz "Reading the same 32-bit word" .size .L.str.7, 29 .type .L.str.8,@object # @.str.8 .L.str.8: .asciz "\t %.6f \t\t %f \t\t %f \t\t %f\n" .size .L.str.8, 28 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z26sharedMemAccessWithStride1v" .size .L__unnamed_1, 32 .type .L__unnamed_2,@object # @1 .L__unnamed_2: .asciz "_Z21sharedMemReadSameWordv" .size .L__unnamed_2, 27 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "\n\n**** successful termination of the program ****\n" .size .Lstr, 52 .type .Lstr.1,@object # @str.1 .Lstr.1: .asciz "Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts" .size .Lstr.1, 89 .type .Lstr.2,@object # @str.2 .Lstr.2: .asciz "\n-------------------------------------------------------------------------------------------------------------------------------" .size .Lstr.2, 129 .type .Lstr.3,@object # @str.3 .Lstr.3: .asciz "Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)" .size .Lstr.3, 81 .type .Lstr.4,@object # @str.4 .Lstr.4: .asciz "-------------------------------------------------------------------------------------------------------------------------------" .size .Lstr.4, 128 .type .Lstr.5,@object # @str.5 .Lstr.5: .asciz "\n ------------------------------------------------------------------------------------------------------------------------------" .size .Lstr.5, 129 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z41__device_stub__sharedMemAccessWithStride1v .addrsig_sym _Z36__device_stub__sharedMemReadSameWordv .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z26sharedMemAccessWithStride1v .addrsig_sym _Z21sharedMemReadSameWordv .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z21sharedMemReadSameWordv .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0020*/ BRA 0x20; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0030*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0040*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0050*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0060*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0070*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0080*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0090*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ .......... Function : _Z26sharedMemAccessWithStride1v .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0020*/ BRA 0x20; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0030*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0040*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0050*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0060*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0070*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0080*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0090*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z26sharedMemAccessWithStride1v .globl _Z26sharedMemAccessWithStride1v .p2align 8 .type _Z26sharedMemAccessWithStride1v,@function _Z26sharedMemAccessWithStride1v: s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z26sharedMemAccessWithStride1v .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 0 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 0 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 1 .amdhsa_next_free_sgpr 1 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z26sharedMemAccessWithStride1v, .Lfunc_end0-_Z26sharedMemAccessWithStride1v .section .AMDGPU.csdata,"",@progbits .text .protected _Z21sharedMemReadSameWordv .globl _Z21sharedMemReadSameWordv .p2align 8 .type _Z21sharedMemReadSameWordv,@function _Z21sharedMemReadSameWordv: s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z21sharedMemReadSameWordv .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 0 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 0 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 1 .amdhsa_next_free_sgpr 1 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end1: .size _Z21sharedMemReadSameWordv, .Lfunc_end1-_Z21sharedMemReadSameWordv .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: [] .group_segment_fixed_size: 0 .kernarg_segment_align: 4 .kernarg_segment_size: 0 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z26sharedMemAccessWithStride1v .private_segment_fixed_size: 0 .sgpr_count: 0 .sgpr_spill_count: 0 .symbol: _Z26sharedMemAccessWithStride1v.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 0 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 - .args: [] .group_segment_fixed_size: 0 .kernarg_segment_align: 4 .kernarg_segment_size: 0 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z21sharedMemReadSameWordv .private_segment_fixed_size: 0 .sgpr_count: 0 .sgpr_spill_count: 0 .symbol: _Z21sharedMemReadSameWordv.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 0 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_001402d6_00000000-6_sharedMemoryReadingSameWord.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2061: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2061: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC0: .string "Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts\n" .align 8 .LC1: .string "The transfered data size (Bytes): %ld\n" .align 8 .LC2: .string "\n-------------------------------------------------------------------------------------------------------------------------------\n" .align 8 .LC3: .string "Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)\n" .align 8 .LC4: .string "-------------------------------------------------------------------------------------------------------------------------------\n" .section .rodata.str1.1,"aMS",@progbits,1 .LC5: .string "Stride of one 32-bit word " .LC7: .string "\t %.6f \t\t %f \t\t %f \t\t %f\n" .LC8: .string "Reading the same 32-bit word" .section .rodata.str1.8 .align 8 .LC9: .string "\n ------------------------------------------------------------------------------------------------------------------------------\n" .text .globl _Z12printResultsv .type _Z12printResultsv, @function _Z12printResultsv: .LFB2058: .cfi_startproc endbr64 pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset 3, -16 leaq .LC0(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl $8000000, %edx leaq .LC1(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq .LC2(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq .LC3(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq .LC4(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq .LC5(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movss _ZL10bandWidths(%rip), %xmm0 divss .LC6(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 movsd _ZL7maxtime(%rip), %xmm3 movsd _ZL7mintime(%rip), %xmm2 movsd _ZL7avgtime(%rip), %xmm1 leaq .LC7(%rip), %rbx movq %rbx, %rsi movl $2, %edi movl $4, %eax call __printf_chk@PLT leaq .LC8(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movss 4+_ZL10bandWidths(%rip), %xmm0 divss .LC6(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 movsd 8+_ZL7maxtime(%rip), %xmm3 movsd 8+_ZL7mintime(%rip), %xmm2 movsd 8+_ZL7avgtime(%rip), %xmm1 movq %rbx, %rsi movl $2, %edi movl $4, %eax call __printf_chk@PLT leaq .LC9(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT popq %rbx .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2058: .size _Z12printResultsv, .-_Z12printResultsv .globl _Z45__device_stub__Z26sharedMemAccessWithStride1vv .type _Z45__device_stub__Z26sharedMemAccessWithStride1vv, @function _Z45__device_stub__Z26sharedMemAccessWithStride1vv: .LFB2083: .cfi_startproc endbr64 subq $88, %rsp .cfi_def_cfa_offset 96 movq %fs:40, %rax movq %rax, 72(%rsp) xorl %eax, %eax movl $1, 16(%rsp) movl $1, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) leaq 8(%rsp), %rcx movq %rsp, %rdx leaq 28(%rsp), %rsi leaq 16(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L9 .L5: movq 72(%rsp), %rax subq %fs:40, %rax jne .L10 addq $88, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L9: .cfi_restore_state pushq 8(%rsp) .cfi_def_cfa_offset 104 pushq 8(%rsp) .cfi_def_cfa_offset 112 leaq 80(%rsp), %r9 movq 44(%rsp), %rcx movl 52(%rsp), %r8d movq 32(%rsp), %rsi movl 40(%rsp), %edx leaq _Z26sharedMemAccessWithStride1v(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 96 jmp .L5 .L10: call __stack_chk_fail@PLT .cfi_endproc .LFE2083: .size _Z45__device_stub__Z26sharedMemAccessWithStride1vv, .-_Z45__device_stub__Z26sharedMemAccessWithStride1vv .globl _Z26sharedMemAccessWithStride1v .type _Z26sharedMemAccessWithStride1v, @function _Z26sharedMemAccessWithStride1v: .LFB2084: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z45__device_stub__Z26sharedMemAccessWithStride1vv addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2084: .size _Z26sharedMemAccessWithStride1v, .-_Z26sharedMemAccessWithStride1v .globl _Z40__device_stub__Z21sharedMemReadSameWordvv .type _Z40__device_stub__Z21sharedMemReadSameWordvv, @function _Z40__device_stub__Z21sharedMemReadSameWordvv: .LFB2085: .cfi_startproc endbr64 subq $88, %rsp .cfi_def_cfa_offset 96 movq %fs:40, %rax movq %rax, 72(%rsp) xorl %eax, %eax movl $1, 16(%rsp) movl $1, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) leaq 8(%rsp), %rcx movq %rsp, %rdx leaq 28(%rsp), %rsi leaq 16(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L17 .L13: movq 72(%rsp), %rax subq %fs:40, %rax jne .L18 addq $88, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L17: .cfi_restore_state pushq 8(%rsp) .cfi_def_cfa_offset 104 pushq 8(%rsp) .cfi_def_cfa_offset 112 leaq 80(%rsp), %r9 movq 44(%rsp), %rcx movl 52(%rsp), %r8d movq 32(%rsp), %rsi movl 40(%rsp), %edx leaq _Z21sharedMemReadSameWordv(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 96 jmp .L13 .L18: call __stack_chk_fail@PLT .cfi_endproc .LFE2085: .size _Z40__device_stub__Z21sharedMemReadSameWordvv, .-_Z40__device_stub__Z21sharedMemReadSameWordvv .globl _Z21sharedMemReadSameWordv .type _Z21sharedMemReadSameWordv, @function _Z21sharedMemReadSameWordv: .LFB2086: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z40__device_stub__Z21sharedMemReadSameWordvv addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2086: .size _Z21sharedMemReadSameWordv, .-_Z21sharedMemReadSameWordv .section .rodata.str1.8 .align 8 .LC13: .string "\n\n**** successful termination of the program ****\n\n" .text .globl main .type main, @function main: .LFB2057: .cfi_startproc endbr64 pushq %r12 .cfi_def_cfa_offset 16 .cfi_offset 12, -16 pushq %rbp .cfi_def_cfa_offset 24 .cfi_offset 6, -24 pushq %rbx .cfi_def_cfa_offset 32 .cfi_offset 3, -32 subq $144, %rsp .cfi_def_cfa_offset 176 movq %fs:40, %rax movq %rax, 136(%rsp) xorl %eax, %eax leaq 8(%rsp), %rdi call cudaEventCreate@PLT leaq 16(%rsp), %rdi call cudaEventCreate@PLT leaq 48(%rsp), %rbp leaq 88(%rsp), %rbx leaq 128(%rsp), %r12 jmp .L24 .L38: call _Z45__device_stub__Z26sharedMemAccessWithStride1vv jmp .L22 .L23: movl $0, %esi movq 16(%rsp), %rdi call cudaEventRecord@PLT movq 16(%rsp), %rdi call cudaEventSynchronize@PLT movq 16(%rsp), %rdx movq 8(%rsp), %rsi movq %rbx, %rdi call cudaEventElapsedTime@PLT addq $4, %rbx cmpq %rbx, %r12 je .L37 .L24: movl $0, %esi movq 8(%rsp), %rdi call cudaEventRecord@PLT movl $128, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $15625, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $0, %r9d movl $0, %r8d movq 36(%rsp), %rdx movl $1, %ecx movq 24(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L38 .L22: movl $0, %esi movq 16(%rsp), %rdi call cudaEventRecord@PLT movq 16(%rsp), %rdi call cudaEventSynchronize@PLT leaq -40(%rbx), %rdi movq 16(%rsp), %rdx movq 8(%rsp), %rsi call cudaEventElapsedTime@PLT movl $0, %esi movq 8(%rsp), %rdi call cudaEventRecord@PLT movl $128, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $15625, 24(%rsp) movl $1, 28(%rsp) movl $1, 32(%rsp) movl $0, %r9d movl $0, %r8d movq 36(%rsp), %rdx movl $1, %ecx movq 24(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax jne .L23 call _Z40__device_stub__Z21sharedMemReadSameWordvv jmp .L23 .L37: leaq 52(%rsp), %rax leaq 40(%rbp), %rdx movsd _ZL7avgtime(%rip), %xmm2 movsd _ZL7maxtime(%rip), %xmm6 movsd 8+_ZL7avgtime(%rip), %xmm1 movsd 8+_ZL7maxtime(%rip), %xmm5 movsd .LC10(%rip), %xmm3 movapd %xmm3, %xmm4 .L25: pxor %xmm0, %xmm0 cvtss2sd (%rax), %xmm0 addsd %xmm0, %xmm2 minsd %xmm0, %xmm3 maxsd %xmm0, %xmm6 pxor %xmm0, %xmm0 cvtss2sd 40(%rax), %xmm0 addsd %xmm0, %xmm1 minsd %xmm0, %xmm4 maxsd %xmm0, %xmm5 addq $4, %rax cmpq %rax, %rdx jne .L25 movsd %xmm3, _ZL7mintime(%rip) movsd %xmm6, _ZL7maxtime(%rip) movsd %xmm4, 8+_ZL7mintime(%rip) movsd %xmm5, 8+_ZL7maxtime(%rip) movsd .LC11(%rip), %xmm5 divsd %xmm5, %xmm2 movsd %xmm2, _ZL7avgtime(%rip) movsd .LC12(%rip), %xmm0 movapd %xmm0, %xmm2 divsd %xmm3, %xmm2 cvtsd2ss %xmm2, %xmm2 movss %xmm2, _ZL10bandWidths(%rip) divsd %xmm5, %xmm1 movsd %xmm1, 8+_ZL7avgtime(%rip) divsd %xmm4, %xmm0 cvtsd2ss %xmm0, %xmm0 movss %xmm0, 4+_ZL10bandWidths(%rip) call _Z12printResultsv leaq .LC13(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq 8(%rsp), %rdi call cudaEventDestroy@PLT movq 16(%rsp), %rdi call cudaEventDestroy@PLT movq 136(%rsp), %rax subq %fs:40, %rax jne .L39 movl $0, %eax addq $144, %rsp .cfi_remember_state .cfi_def_cfa_offset 32 popq %rbx .cfi_def_cfa_offset 24 popq %rbp .cfi_def_cfa_offset 16 popq %r12 .cfi_def_cfa_offset 8 ret .L39: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1 .LC14: .string "_Z21sharedMemReadSameWordv" .section .rodata.str1.8 .align 8 .LC15: .string "_Z26sharedMemAccessWithStride1v" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2088: .cfi_startproc endbr64 pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset 3, -16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rbx movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC14(%rip), %rdx movq %rdx, %rcx leaq _Z21sharedMemReadSameWordv(%rip), %rsi movq %rax, %rdi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC15(%rip), %rdx movq %rdx, %rcx leaq _Z26sharedMemAccessWithStride1v(%rip), %rsi movq %rbx, %rdi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT popq %rbx .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2088: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .local _ZL10bandWidths .comm _ZL10bandWidths,8,8 .local _ZL7mintime .comm _ZL7mintime,16,16 .local _ZL7maxtime .comm _ZL7maxtime,16,16 .local _ZL7avgtime .comm _ZL7avgtime,16,16 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC6: .long 1232348160 .section .rodata.cst8,"aM",@progbits,8 .align 8 .LC10: .long -536870912 .long 1206910975 .align 8 .LC11: .long 0 .long 1075970048 .align 8 .LC12: .long 0 .long 1096713344 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "sharedMemoryReadingSameWord.hip" .globl _Z41__device_stub__sharedMemAccessWithStride1v # -- Begin function _Z41__device_stub__sharedMemAccessWithStride1v .p2align 4, 0x90 .type _Z41__device_stub__sharedMemAccessWithStride1v,@function _Z41__device_stub__sharedMemAccessWithStride1v: # @_Z41__device_stub__sharedMemAccessWithStride1v .cfi_startproc # %bb.0: subq $56, %rsp .cfi_def_cfa_offset 64 leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 48(%rsp), %r9 movl $_Z26sharedMemAccessWithStride1v, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $72, %rsp .cfi_adjust_cfa_offset -72 retq .Lfunc_end0: .size _Z41__device_stub__sharedMemAccessWithStride1v, .Lfunc_end0-_Z41__device_stub__sharedMemAccessWithStride1v .cfi_endproc # -- End function .globl _Z36__device_stub__sharedMemReadSameWordv # -- Begin function _Z36__device_stub__sharedMemReadSameWordv .p2align 4, 0x90 .type _Z36__device_stub__sharedMemReadSameWordv,@function _Z36__device_stub__sharedMemReadSameWordv: # @_Z36__device_stub__sharedMemReadSameWordv .cfi_startproc # %bb.0: subq $56, %rsp .cfi_def_cfa_offset 64 leaq 32(%rsp), %rdi leaq 16(%rsp), %rsi leaq 8(%rsp), %rdx movq %rsp, %rcx callq __hipPopCallConfiguration movq 32(%rsp), %rsi movl 40(%rsp), %edx movq 16(%rsp), %rcx movl 24(%rsp), %r8d leaq 48(%rsp), %r9 movl $_Z21sharedMemReadSameWordv, %edi pushq (%rsp) .cfi_adjust_cfa_offset 8 pushq 16(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $72, %rsp .cfi_adjust_cfa_offset -72 retq .Lfunc_end1: .size _Z36__device_stub__sharedMemReadSameWordv, .Lfunc_end1-_Z36__device_stub__sharedMemReadSameWordv .cfi_endproc # -- End function .section .rodata.cst8,"aM",@progbits,8 .p2align 3, 0x0 # -- Begin function main .LCPI2_0: .quad 0x4022000000000000 # double 9 .LCPI2_1: .quad 0x415e848000000000 # double 8.0E+6 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $168, %rsp .cfi_def_cfa_offset 224 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movabsq $4294967424, %rbx # imm = 0x100000080 leaq 8(%rsp), %rdi callq hipEventCreate movq %rsp, %rdi callq hipEventCreate leaq 80(%rsp), %r14 movl $10, %ebp leaq 15497(%rbx), %r15 leaq 56(%rsp), %r12 leaq 64(%rsp), %r13 jmp .LBB2_1 .p2align 4, 0x90 .LBB2_5: # in Loop: Header=BB2_1 Depth=1 movq (%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq (%rsp), %rdi callq hipEventSynchronize leaq 40(%r14), %rdi movq 8(%rsp), %rsi movq (%rsp), %rdx callq hipEventElapsedTime addq $4, %r14 decq %rbp je .LBB2_6 .LBB2_1: # =>This Inner Loop Header: Depth=1 movq 8(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq %r15, %rdi movl $1, %esi movq %rbx, %rdx movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB2_3 # %bb.2: # in Loop: Header=BB2_1 Depth=1 leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx movq %r12, %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d movl $_Z26sharedMemAccessWithStride1v, %edi movq %r13, %r9 pushq 56(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB2_3: # in Loop: Header=BB2_1 Depth=1 movq (%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq (%rsp), %rdi callq hipEventSynchronize movq 8(%rsp), %rsi movq (%rsp), %rdx movq %r14, %rdi callq hipEventElapsedTime movq 8(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq %r15, %rdi movl $1, %esi movq %rbx, %rdx movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB2_5 # %bb.4: # in Loop: Header=BB2_1 Depth=1 leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx movq %r12, %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d movl $_Z21sharedMemReadSameWordv, %edi movq %r13, %r9 pushq 56(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 jmp .LBB2_5 .LBB2_6: # %.preheader70.preheader movq $-8, %rax movabsq $5183643170566569984, %rcx # imm = 0x47EFFFFFE0000000 .p2align 4, 0x90 .LBB2_7: # %.preheader70 # =>This Inner Loop Header: Depth=1 movq %rcx, _ZL7mintime+8(%rax) addq $8, %rax je .LBB2_7 # %bb.8: # %.preheader68.preheader movl $1, %eax leaq 84(%rsp), %rcx .p2align 4, 0x90 .LBB2_9: # %.preheader68 # =>This Loop Header: Depth=1 # Child Loop BB2_10 Depth 2 movq $-8, %rdx movq %rcx, %rsi .p2align 4, 0x90 .LBB2_10: # Parent Loop BB2_9 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rsi), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movsd _ZL7avgtime+8(%rdx), %xmm1 # xmm1 = mem[0],zero addsd %xmm0, %xmm1 movsd %xmm1, _ZL7avgtime+8(%rdx) movsd _ZL7mintime+8(%rdx), %xmm1 # xmm1 = mem[0],zero minsd %xmm0, %xmm1 movsd %xmm1, _ZL7mintime+8(%rdx) movsd _ZL7maxtime+8(%rdx), %xmm1 # xmm1 = mem[0],zero maxsd %xmm0, %xmm1 movsd %xmm1, _ZL7maxtime+8(%rdx) addq $40, %rsi addq $8, %rdx je .LBB2_10 # %bb.11: # in Loop: Header=BB2_9 Depth=1 incq %rax addq $4, %rcx cmpq $10, %rax jne .LBB2_9 # %bb.12: # %.preheader.preheader movq $-4, %rax movsd .LCPI2_0(%rip), %xmm0 # xmm0 = mem[0],zero movsd .LCPI2_1(%rip), %xmm1 # xmm1 = mem[0],zero .p2align 4, 0x90 .LBB2_13: # %.preheader # =>This Inner Loop Header: Depth=1 movsd _ZL7avgtime+8(%rax,%rax), %xmm2 # xmm2 = mem[0],zero divsd %xmm0, %xmm2 movsd %xmm2, _ZL7avgtime+8(%rax,%rax) movapd %xmm1, %xmm2 divsd _ZL7mintime+8(%rax,%rax), %xmm2 cvtsd2ss %xmm2, %xmm2 movss %xmm2, _ZL10bandWidths+4(%rax) addq $4, %rax je .LBB2_13 # %bb.14: callq _Z12printResultsv movl $.Lstr, %edi callq puts@PLT movq 8(%rsp), %rdi callq hipEventDestroy movq (%rsp), %rdi callq hipEventDestroy xorl %eax, %eax addq $168, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end2: .size main, .Lfunc_end2-main .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function _Z12printResultsv .LCPI3_0: .long 0x49742400 # float 1.0E+6 .text .globl _Z12printResultsv .p2align 4, 0x90 .type _Z12printResultsv,@function _Z12printResultsv: # @_Z12printResultsv .cfi_startproc # %bb.0: pushq %r14 .cfi_def_cfa_offset 16 pushq %rbx .cfi_def_cfa_offset 24 pushq %rax .cfi_def_cfa_offset 32 .cfi_offset %rbx, -24 .cfi_offset %r14, -16 movl $.Lstr.1, %edi callq puts@PLT movl $.L.str.2, %edi movl $8000000, %esi # imm = 0x7A1200 xorl %eax, %eax callq printf movl $.Lstr.2, %edi callq puts@PLT movl $.Lstr.3, %edi callq puts@PLT movl $.Lstr.4, %edi callq puts@PLT movq $-4, %rbx movl $.L.str.6, %r14d .p2align 4, 0x90 .LBB3_1: # =>This Inner Loop Header: Depth=1 cmpq $-4, %rbx movl $.L.str.7, %edi cmoveq %r14, %rdi xorl %eax, %eax callq printf movss _ZL10bandWidths+4(%rbx), %xmm0 # xmm0 = mem[0],zero,zero,zero divss .LCPI3_0(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 movsd _ZL7avgtime+8(%rbx,%rbx), %xmm1 # xmm1 = mem[0],zero movsd _ZL7mintime+8(%rbx,%rbx), %xmm2 # xmm2 = mem[0],zero movsd _ZL7maxtime+8(%rbx,%rbx), %xmm3 # xmm3 = mem[0],zero movl $.L.str.8, %edi movb $4, %al callq printf addq $4, %rbx je .LBB3_1 # %bb.2: movl $.Lstr.5, %edi addq $8, %rsp .cfi_def_cfa_offset 24 popq %rbx .cfi_def_cfa_offset 16 popq %r14 .cfi_def_cfa_offset 8 jmp puts@PLT # TAILCALL .Lfunc_end3: .size _Z12printResultsv, .Lfunc_end3-_Z12printResultsv .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 subq $32, %rsp .cfi_def_cfa_offset 48 .cfi_offset %rbx, -16 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB4_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB4_2: movq __hip_gpubin_handle(%rip), %rbx xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z26sharedMemAccessWithStride1v, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z21sharedMemReadSameWordv, %esi movl $.L__unnamed_2, %edx movl $.L__unnamed_2, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $32, %rsp .cfi_def_cfa_offset 16 popq %rbx .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end4: .size __hip_module_ctor, .Lfunc_end4-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB5_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB5_2: retq .Lfunc_end5: .size __hip_module_dtor, .Lfunc_end5-__hip_module_dtor .cfi_endproc # -- End function .type _Z26sharedMemAccessWithStride1v,@object # @_Z26sharedMemAccessWithStride1v .section .rodata,"a",@progbits .globl _Z26sharedMemAccessWithStride1v .p2align 3, 0x0 _Z26sharedMemAccessWithStride1v: .quad _Z41__device_stub__sharedMemAccessWithStride1v .size _Z26sharedMemAccessWithStride1v, 8 .type _Z21sharedMemReadSameWordv,@object # @_Z21sharedMemReadSameWordv .globl _Z21sharedMemReadSameWordv .p2align 3, 0x0 _Z21sharedMemReadSameWordv: .quad _Z36__device_stub__sharedMemReadSameWordv .size _Z21sharedMemReadSameWordv, 8 .type _ZL7mintime,@object # @_ZL7mintime .local _ZL7mintime .comm _ZL7mintime,16,16 .type _ZL7avgtime,@object # @_ZL7avgtime .local _ZL7avgtime .comm _ZL7avgtime,16,16 .type _ZL7maxtime,@object # @_ZL7maxtime .local _ZL7maxtime .comm _ZL7maxtime,16,16 .type _ZL10bandWidths,@object # @_ZL10bandWidths .local _ZL10bandWidths .comm _ZL10bandWidths,8,4 .type .L.str.2,@object # @.str.2 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.2: .asciz "The transfered data size (Bytes): %ld\n" .size .L.str.2, 39 .type .L.str.6,@object # @.str.6 .L.str.6: .asciz "Stride of one 32-bit word " .size .L.str.6, 29 .type .L.str.7,@object # @.str.7 .L.str.7: .asciz "Reading the same 32-bit word" .size .L.str.7, 29 .type .L.str.8,@object # @.str.8 .L.str.8: .asciz "\t %.6f \t\t %f \t\t %f \t\t %f\n" .size .L.str.8, 28 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z26sharedMemAccessWithStride1v" .size .L__unnamed_1, 32 .type .L__unnamed_2,@object # @1 .L__unnamed_2: .asciz "_Z21sharedMemReadSameWordv" .size .L__unnamed_2, 27 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "\n\n**** successful termination of the program ****\n" .size .Lstr, 52 .type .Lstr.1,@object # @str.1 .Lstr.1: .asciz "Demonstrating that reading same 32-bit word by all threads will not cause bank conflicts" .size .Lstr.1, 89 .type .Lstr.2,@object # @str.2 .Lstr.2: .asciz "\n-------------------------------------------------------------------------------------------------------------------------------" .size .Lstr.2, 129 .type .Lstr.3,@object # @str.3 .Lstr.3: .asciz "Pattern \t\t\t Bandwidth (GB/sec) \t Avg time (ms) \t Min time (ms) \t Max time (ms)" .size .Lstr.3, 81 .type .Lstr.4,@object # @str.4 .Lstr.4: .asciz "-------------------------------------------------------------------------------------------------------------------------------" .size .Lstr.4, 128 .type .Lstr.5,@object # @str.5 .Lstr.5: .asciz "\n ------------------------------------------------------------------------------------------------------------------------------" .size .Lstr.5, 129 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z41__device_stub__sharedMemAccessWithStride1v .addrsig_sym _Z36__device_stub__sharedMemReadSameWordv .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z26sharedMemAccessWithStride1v .addrsig_sym _Z21sharedMemReadSameWordv .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include <cstdlib> #include <stdio.h> #include <cassert> #include <cuda_runtime.h> /* Naive implementation. Allocate one thread for one element in result matrix, processing dot(Arow, Bcol); */ __global__ void kMatrixMul0 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2){ int x = blockDim.x*blockIdx.x + threadIdx.x; int y = blockDim.y*blockIdx.y + threadIdx.y; if (x >= n2 || y >= m1){return;} float sum=0.0; for (int i=0; i<m2; i++){ // if (blockIdx.x == 0 && threadIdx.x == 0 && blockIdx.y == 0 && threadIdx.y == 0) { // printf("sum:%f mat1:%f mat2:%f i:%d\n", sum, d_mat1[y*m2+i], d_mat2[i*n2+x], i); // } sum += d_mat1[y*m2+i] * d_mat2[i*n2+x]; } d_res[n2*y+x] = sum; }; /* Using shared memory */ template<int BLOCK_SIZE> __global__ void kMatrixMul1 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data As[ty][tx] = d_mat1[a + m2 * ty + tx]; Bs[ty][tx] = d_mat2[b + n2 * ty + tx]; __syncthreads(); for (int k = 0; k<blockDim.x; k++){ c += As[ty][k] * Bs[k][tx]; } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } /* Using register blocking */ template<int BLOCK_SIZE, int REGT_SIZE> __global__ void kMatrixMul2 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; int reg_tiles = BLOCK_SIZE / REGT_SIZE; float acc[REGT_SIZE][REGT_SIZE]; float Ar; float Br[REGT_SIZE]; #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ acc[i][j] = 0.0f; } } //block 32x32, each thread work on 2x2 for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ As[ty+i][tx+j] = d_mat1[a + m2 * (ty + i) + tx + j]; Bs[ty+i][tx+j] = d_mat2[b + n2 * (ty + i) + tx + j]; } } __syncthreads(); for (int k=0; k<BLOCK_SIZE; k+=REGT_SIZE){ #pragma unroll for (int i=0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ int ax = k + j; int ay = k + i; acc[ty+i][tx+j] += As[ty][k] * Bs[k][tx]; } } } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } float* MatrixMultGPU0(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ float *d_res, *d_mat1, *d_mat2; cudaEvent_t start, end; cudaError_t error; error = cudaEventCreate(&start); error = cudaEventCreate(&end); //malloc the device memory for matrices cudaError_t result = cudaMalloc((void**)&d_res, sizeof(float)*m1*n2); result = cudaMalloc((void**)&d_mat1, sizeof(float)*m1*m2); assert (result == cudaSuccess); result = cudaMalloc((void**)&d_mat2, sizeof(float)*n1*n2); assert (result == cudaSuccess); //init source matrices in device memory result = cudaMemcpy(d_mat1, mat1, sizeof(float)*m1*m2, cudaMemcpyHostToDevice); assert (result == cudaSuccess); result = cudaMemcpy(d_mat2, mat2, sizeof(float)*n1*n2, cudaMemcpyHostToDevice); assert (result == cudaSuccess); cudaEventRecord(start, NULL); int N = 64; dim3 block_size(N, N); //grid width in blocks int grid_wib = ceil(float(n2)/float(N)); //grid height in blocks int grid_hib = ceil(float(m1)/float(N)); dim3 grid_size(grid_wib, grid_hib); //naive version //kMatrixMul0<<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //sharedMem version kMatrixMul1<64><<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //copy back the multiplication result float* res = new float[m1*n2]; result = cudaMemcpy(res, d_res, sizeof(float)*m1*n2, cudaMemcpyDeviceToHost); assert (result == cudaSuccess); cudaEventRecord(end, NULL); error = cudaEventSynchronize(end); float msecTotal = 0.0f; error = cudaEventElapsedTime(&msecTotal, start, end); printf("calculation Time:%f ms\n", msecTotal); cudaFree(d_res); cudaFree(d_mat1); cudaFree(d_mat2); return res; } float* MatrixMulCPU(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ assert(m2 == n1 && "matrx a's cols != matrix b's rows"); float* res = new float[m1*n2] ; for (int i = 0; i < m1; i++) { for (int j = 0; j < n2; j++) { res[i*m1 + j] = 0; for (int x = 0; x < m2; x++) { res[i*m1+j] += mat1[i*m1+x] * mat2[x*n1+j]; } } } return res; } float CompareMatrix(float mat1[], float mat2[], int m1, int m2){ float err = 0; for (int x=0; x<m1; x++){ for (int y=0; y<m2; y++){ err += mat1[m2*y+x] - mat2[m2*y+x]; } } return err; /* if (err > 0.1){ printf("matrix comparison failed.error:%f\n", err); return false; } return true; */ } void FillMatrix(float mat[], int m1, int m2, float d=1.0){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ if (d<0.0) { mat[i*m1+j] = static_cast<float>(rand())/static_cast<float>(RAND_MAX) * 10.0; } else { mat[i*m1+j] = d; } } } } void PrintMatrix(float mat[], int m1, int m2){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ printf("%f,", mat[i*m1+j]); } printf("\n"); } } int main(int argc, char *argv[]){ int m1 = 1024; int m2; int n1 = m2 = 1024; int n2 = 1024; float *matrix_a = new float[m1*m2]; float *matrix_b = new float[n1*n2]; FillMatrix((float*)matrix_a, m1, m2); FillMatrix((float*)matrix_b, n1, n2); //float *ref = MatrixMulCPU((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); float *res = MatrixMultGPU0((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); //printf("error:%f\n", CompareMatrix(ref, res, m1, n2)); //PrintMatrix(res, m1, m2); delete[] matrix_a; delete[] matrix_b; delete res; //delete ref; }
.file "tmpxft_000499f5_00000000-6_matrixmult.cudafe1.cpp" .text #APP #NO_APP .type _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii, @function _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii: .LFB2093: .cfi_startproc subq $184, %rsp .cfi_def_cfa_offset 192 movq %rdi, 40(%rsp) movq %rsi, 32(%rsp) movl %edx, 28(%rsp) movl %ecx, 24(%rsp) movq %r8, 16(%rsp) movl %r9d, 12(%rsp) movq %fs:40, %rax movq %rax, 168(%rsp) xorl %eax, %eax leaq 40(%rsp), %rax movq %rax, 112(%rsp) leaq 32(%rsp), %rax movq %rax, 120(%rsp) leaq 28(%rsp), %rax movq %rax, 128(%rsp) leaq 24(%rsp), %rax movq %rax, 136(%rsp) leaq 16(%rsp), %rax movq %rax, 144(%rsp) leaq 12(%rsp), %rax movq %rax, 152(%rsp) leaq 192(%rsp), %rax movq %rax, 160(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) movl $1, 72(%rsp) movl $1, 76(%rsp) movl $1, 80(%rsp) movl $1, 84(%rsp) leaq 56(%rsp), %rcx leaq 48(%rsp), %rdx leaq 76(%rsp), %rsi leaq 64(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L5 .L1: movq 168(%rsp), %rax subq %fs:40, %rax jne .L6 addq $184, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L5: .cfi_restore_state pushq 56(%rsp) .cfi_def_cfa_offset 200 pushq 56(%rsp) .cfi_def_cfa_offset 208 leaq 128(%rsp), %r9 movq 92(%rsp), %rcx movl 100(%rsp), %r8d movq 80(%rsp), %rsi movl 88(%rsp), %edx leaq _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 192 jmp .L1 .L6: call __stack_chk_fail@PLT .cfi_endproc .LFE2093: .size _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii, .-_ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii .section .text._Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,"axG",@progbits,_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,comdat .weak _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .type _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, @function _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii: .LFB2142: .cfi_startproc endbr64 subq $16, %rsp .cfi_def_cfa_offset 24 movl 24(%rsp), %eax pushq %rax .cfi_def_cfa_offset 32 call _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii addq $24, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2142: .size _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, .-_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .text .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2069: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2069: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .section .rodata.str1.1,"aMS",@progbits,1 .LC5: .string "calculation Time:%f ms\n" .text .globl _Z14MatrixMultGPU0PfiiS_ii .type _Z14MatrixMultGPU0PfiiS_ii, @function _Z14MatrixMultGPU0PfiiS_ii: .LFB2061: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $120, %rsp .cfi_def_cfa_offset 176 movq %rdi, 16(%rsp) movl %esi, %r13d movl %edx, 8(%rsp) movq %rcx, 24(%rsp) movl %r8d, 12(%rsp) movl %r9d, %r12d movq %fs:40, %rax movq %rax, 104(%rsp) xorl %eax, %eax leaq 64(%rsp), %rdi call cudaEventCreate@PLT leaq 72(%rsp), %rdi call cudaEventCreate@PLT movslq %r13d, %rbp movslq %r12d, %rbx movq %rbp, %r15 imulq %rbx, %r15 leaq 0(,%r15,4), %r14 leaq 40(%rsp), %rdi movq %r14, %rsi call cudaMalloc@PLT movslq 8(%rsp), %rax imulq %rax, %rbp salq $2, %rbp leaq 48(%rsp), %rdi movq %rbp, %rsi call cudaMalloc@PLT movslq 12(%rsp), %rax imulq %rax, %rbx salq $2, %rbx leaq 56(%rsp), %rdi movq %rbx, %rsi call cudaMalloc@PLT movl $1, %ecx movq %rbp, %rdx movq 16(%rsp), %rsi movq 48(%rsp), %rdi call cudaMemcpy@PLT movl $1, %ecx movq %rbx, %rdx movq 24(%rsp), %rsi movq 56(%rsp), %rdi call cudaMemcpy@PLT movl $0, %esi movq 64(%rsp), %rdi call cudaEventRecord@PLT movl $64, 80(%rsp) movl $64, 84(%rsp) movl $1, 88(%rsp) pxor %xmm0, %xmm0 cvtsi2ssl %r12d, %xmm0 mulss .LC0(%rip), %xmm0 movaps %xmm0, %xmm1 movss .LC6(%rip), %xmm3 movaps %xmm0, %xmm2 andps %xmm3, %xmm2 movss .LC1(%rip), %xmm4 ucomiss %xmm2, %xmm4 jbe .L12 cvttss2sil %xmm0, %eax pxor %xmm2, %xmm2 cvtsi2ssl %eax, %xmm2 cmpnless %xmm2, %xmm1 movss .LC3(%rip), %xmm4 andps %xmm4, %xmm1 addss %xmm2, %xmm1 andnps %xmm0, %xmm3 orps %xmm3, %xmm1 .L12: pxor %xmm0, %xmm0 cvtsi2ssl %r13d, %xmm0 mulss .LC0(%rip), %xmm0 movaps %xmm0, %xmm4 movss .LC6(%rip), %xmm3 movaps %xmm0, %xmm2 andps %xmm3, %xmm2 movss .LC1(%rip), %xmm5 ucomiss %xmm2, %xmm5 jbe .L13 cvttss2sil %xmm0, %eax pxor %xmm2, %xmm2 cvtsi2ssl %eax, %xmm2 cmpnless %xmm2, %xmm4 movss .LC3(%rip), %xmm5 andps %xmm5, %xmm4 addss %xmm2, %xmm4 andnps %xmm0, %xmm3 orps %xmm3, %xmm4 .L13: cvttss2sil %xmm1, %eax movl %eax, 92(%rsp) cvttss2sil %xmm4, %eax movl %eax, 96(%rsp) movl 88(%rsp), %ecx movl $0, %r9d movl $0, %r8d movq 80(%rsp), %rdx movq 92(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L20 .L14: movabsq $2305843009213693950, %rax cmpq %r15, %rax jb .L15 movq %r14, %rdi call _Znam@PLT movq %rax, %rbx movl $2, %ecx movq %r14, %rdx movq 40(%rsp), %rsi movq %rax, %rdi call cudaMemcpy@PLT movl $0, %esi movq 72(%rsp), %rdi call cudaEventRecord@PLT movq 72(%rsp), %rdi call cudaEventSynchronize@PLT movl $0x00000000, 36(%rsp) leaq 36(%rsp), %rdi movq 72(%rsp), %rdx movq 64(%rsp), %rsi call cudaEventElapsedTime@PLT pxor %xmm0, %xmm0 cvtss2sd 36(%rsp), %xmm0 leaq .LC5(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movq 40(%rsp), %rdi call cudaFree@PLT movq 48(%rsp), %rdi call cudaFree@PLT movq 56(%rsp), %rdi call cudaFree@PLT movq 104(%rsp), %rax subq %fs:40, %rax jne .L21 movq %rbx, %rax addq $120, %rsp .cfi_remember_state .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .L20: .cfi_restore_state subq $8, %rsp .cfi_def_cfa_offset 184 pushq %r12 .cfi_def_cfa_offset 192 movl 28(%rsp), %r9d movq 72(%rsp), %r8 movl 24(%rsp), %ecx movl %r13d, %edx movq 64(%rsp), %rsi movq 56(%rsp), %rdi call _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii addq $16, %rsp .cfi_def_cfa_offset 176 jmp .L14 .L15: movq 104(%rsp), %rax subq %fs:40, %rax je .L16 call __stack_chk_fail@PLT .L16: call __cxa_throw_bad_array_new_length@PLT .L21: call __stack_chk_fail@PLT .cfi_endproc .LFE2061: .size _Z14MatrixMultGPU0PfiiS_ii, .-_Z14MatrixMultGPU0PfiiS_ii .globl _Z12MatrixMulCPUPfiiS_ii .type _Z12MatrixMulCPUPfiiS_ii, @function _Z12MatrixMulCPUPfiiS_ii: .LFB2062: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $24, %rsp .cfi_def_cfa_offset 80 movl %edx, %ebp movq %rcx, 8(%rsp) movl %esi, %eax imull %r9d, %eax cltq movabsq $2305843009213693950, %rdx cmpq %rax, %rdx jb .L23 movq %rdi, %r13 movl %esi, %r12d movl %r8d, %r15d movl %r9d, %ebx leaq 0(,%rax,4), %rdi call _Znam@PLT movq %rax, %r14 testl %r12d, %r12d jle .L22 movslq %r15d, %rsi salq $2, %rsi movl $0, %edx movl $0, %eax movslq %ebp, %r15 movq %r14, %rdi movq %r15, %rcx jmp .L25 .L23: call __cxa_throw_bad_array_new_length@PLT .L29: movslq %edx, %r8 leaq 0(,%r8,4), %r15 leaq (%rdi,%r15), %r9 movq 8(%rsp), %r11 addq %r13, %r15 addq %rcx, %r8 leaq 0(%r13,%r8,4), %r8 movl $0, %r10d movl %eax, (%rsp) movl %edx, 4(%rsp) .L28: movq %r9, %r14 movl $0x00000000, (%r9) testl %ebp, %ebp jle .L26 movq %r11, %rdx movq %r15, %rax pxor %xmm1, %xmm1 .L27: movss (%rax), %xmm0 mulss (%rdx), %xmm0 addss %xmm0, %xmm1 addq $4, %rax addq %rsi, %rdx cmpq %r8, %rax jne .L27 movss %xmm1, (%r14) .L26: addl $1, %r10d addq $4, %r9 addq $4, %r11 cmpl %r10d, %ebx jne .L28 movl (%rsp), %eax movl 4(%rsp), %edx .L30: addl $1, %eax addl %r12d, %edx cmpl %eax, %r12d je .L33 .L25: testl %ebx, %ebx jg .L29 jmp .L30 .L33: movq %rdi, %r14 .L22: movq %r14, %rax addq $24, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2062: .size _Z12MatrixMulCPUPfiiS_ii, .-_Z12MatrixMulCPUPfiiS_ii .globl _Z13CompareMatrixPfS_ii .type _Z13CompareMatrixPfS_ii, @function _Z13CompareMatrixPfS_ii: .LFB2063: .cfi_startproc endbr64 testl %edx, %edx jle .L41 movslq %edx, %r10 movslq %ecx, %r8 salq $2, %r8 movl $0, %r9d pxor %xmm1, %xmm1 jmp .L37 .L38: movss (%rdi,%rax), %xmm0 subss (%rsi,%rax), %xmm0 addss %xmm0, %xmm1 addl $1, %edx addq %r8, %rax cmpl %edx, %ecx jne .L38 .L40: addq $1, %r9 cmpq %r10, %r9 je .L35 .L37: leaq 0(,%r9,4), %rax movl $0, %edx testl %ecx, %ecx jg .L38 jmp .L40 .L41: pxor %xmm1, %xmm1 .L35: movaps %xmm1, %xmm0 ret .cfi_endproc .LFE2063: .size _Z13CompareMatrixPfS_ii, .-_Z13CompareMatrixPfS_ii .globl _Z10FillMatrixPfiif .type _Z10FillMatrixPfiif, @function _Z10FillMatrixPfiif: .LFB2064: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $24, %rsp .cfi_def_cfa_offset 80 movl %edx, 12(%rsp) testl %esi, %esi jle .L44 movq %rdi, %r13 movl %esi, %r15d movd %xmm0, %ebp movl %edx, %r12d movl $0, 8(%rsp) movl $0, %r14d jmp .L46 .L58: call rand@PLT movl %eax, %edx movslq %ebx, %rax pxor %xmm0, %xmm0 cvtsi2ssl %edx, %xmm0 mulss .LC7(%rip), %xmm0 mulss .LC8(%rip), %xmm0 movss %xmm0, 0(%r13,%rax,4) .L49: addl $1, %ebx cmpl %r12d, %ebx je .L51 .L50: pxor %xmm1, %xmm1 movd %ebp, %xmm2 comiss %xmm2, %xmm1 ja .L58 movslq %ebx, %rax movl %ebp, 0(%r13,%rax,4) jmp .L49 .L51: addl $1, %r14d addl %r15d, %r12d addl %r15d, 8(%rsp) cmpl %r14d, %r15d je .L44 .L46: movl 8(%rsp), %ebx cmpl $0, 12(%rsp) jg .L50 jmp .L51 .L44: addq $24, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2064: .size _Z10FillMatrixPfiif, .-_Z10FillMatrixPfiif .section .rodata.str1.1 .LC9: .string "%f," .LC10: .string "\n" .text .globl _Z11PrintMatrixPfii .type _Z11PrintMatrixPfii, @function _Z11PrintMatrixPfii: .LFB2065: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $40, %rsp .cfi_def_cfa_offset 96 movq %rdi, 16(%rsp) movl %edx, 12(%rsp) testl %esi, %esi jle .L59 movl %esi, %r15d movl $0, %r14d movl $0, %r13d movl %edx, %eax cltq movq %rax, 24(%rsp) leaq .LC9(%rip), %r12 jmp .L61 .L63: movslq %r14d, %rax movq 16(%rsp), %rcx leaq (%rcx,%rax,4), %rbx movq 24(%rsp), %rdx addq %rdx, %rax leaq (%rcx,%rax,4), %rbp .L62: pxor %xmm0, %xmm0 cvtss2sd (%rbx), %xmm0 movq %r12, %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT addq $4, %rbx cmpq %rbp, %rbx jne .L62 .L64: leaq .LC10(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addl $1, %r13d addl %r15d, %r14d cmpl %r13d, %r15d je .L59 .L61: cmpl $0, 12(%rsp) jg .L63 jmp .L64 .L59: addq $40, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2065: .size _Z11PrintMatrixPfii, .-_Z11PrintMatrixPfii .globl main .type main, @function main: .LFB2066: .cfi_startproc endbr64 pushq %r12 .cfi_def_cfa_offset 16 .cfi_offset 12, -16 pushq %rbp .cfi_def_cfa_offset 24 .cfi_offset 6, -24 pushq %rbx .cfi_def_cfa_offset 32 .cfi_offset 3, -32 movl $4194304, %edi call _Znam@PLT movq %rax, %rbp movl $4194304, %edi call _Znam@PLT movq %rax, %rbx movss .LC3(%rip), %xmm0 movl $1024, %edx movl $1024, %esi movq %rbp, %rdi call _Z10FillMatrixPfiif movss .LC3(%rip), %xmm0 movl $1024, %edx movl $1024, %esi movq %rbx, %rdi call _Z10FillMatrixPfiif movl $1024, %r9d movl $1024, %r8d movq %rbx, %rcx movl $1024, %edx movl $1024, %esi movq %rbp, %rdi call _Z14MatrixMultGPU0PfiiS_ii movq %rax, %r12 movq %rbp, %rdi call _ZdaPv@PLT movq %rbx, %rdi call _ZdaPv@PLT testq %r12, %r12 je .L68 movl $4, %esi movq %r12, %rdi call _ZdlPvm@PLT .L68: movl $0, %eax popq %rbx .cfi_def_cfa_offset 24 popq %rbp .cfi_def_cfa_offset 16 popq %r12 .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2066: .size main, .-main .globl _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii .type _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii, @function _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii: .LFB2091: .cfi_startproc endbr64 subq $184, %rsp .cfi_def_cfa_offset 192 movq %rdi, 40(%rsp) movq %rsi, 32(%rsp) movl %edx, 28(%rsp) movl %ecx, 24(%rsp) movq %r8, 16(%rsp) movl %r9d, 12(%rsp) movq %fs:40, %rax movq %rax, 168(%rsp) xorl %eax, %eax leaq 40(%rsp), %rax movq %rax, 112(%rsp) leaq 32(%rsp), %rax movq %rax, 120(%rsp) leaq 28(%rsp), %rax movq %rax, 128(%rsp) leaq 24(%rsp), %rax movq %rax, 136(%rsp) leaq 16(%rsp), %rax movq %rax, 144(%rsp) leaq 12(%rsp), %rax movq %rax, 152(%rsp) leaq 192(%rsp), %rax movq %rax, 160(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) movl $1, 72(%rsp) movl $1, 76(%rsp) movl $1, 80(%rsp) movl $1, 84(%rsp) leaq 56(%rsp), %rcx leaq 48(%rsp), %rdx leaq 76(%rsp), %rsi leaq 64(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L74 .L70: movq 168(%rsp), %rax subq %fs:40, %rax jne .L75 addq $184, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L74: .cfi_restore_state pushq 56(%rsp) .cfi_def_cfa_offset 200 pushq 56(%rsp) .cfi_def_cfa_offset 208 leaq 128(%rsp), %r9 movq 92(%rsp), %rcx movl 100(%rsp), %r8d movq 80(%rsp), %rsi movl 88(%rsp), %edx leaq _Z11kMatrixMul0PfS_iiS_ii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 192 jmp .L70 .L75: call __stack_chk_fail@PLT .cfi_endproc .LFE2091: .size _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii, .-_Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii .globl _Z11kMatrixMul0PfS_iiS_ii .type _Z11kMatrixMul0PfS_iiS_ii, @function _Z11kMatrixMul0PfS_iiS_ii: .LFB2092: .cfi_startproc endbr64 subq $16, %rsp .cfi_def_cfa_offset 24 movl 24(%rsp), %eax pushq %rax .cfi_def_cfa_offset 32 call _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii addq $24, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2092: .size _Z11kMatrixMul0PfS_iiS_ii, .-_Z11kMatrixMul0PfS_iiS_ii .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC11: .string "_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii" .section .rodata.str1.1 .LC12: .string "_Z11kMatrixMul0PfS_iiS_ii" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2096: .cfi_startproc endbr64 pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset 3, -16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rbx movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC11(%rip), %rdx movq %rdx, %rcx leaq _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii(%rip), %rsi movq %rax, %rdi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC12(%rip), %rdx movq %rdx, %rcx leaq _Z11kMatrixMul0PfS_iiS_ii(%rip), %rsi movq %rbx, %rdi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT popq %rbx .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2096: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC0: .long 1015021568 .align 4 .LC1: .long 1258291200 .align 4 .LC3: .long 1065353216 .align 4 .LC6: .long 2147483647 .align 4 .LC7: .long 805306368 .align 4 .LC8: .long 1092616192 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include <cstdlib> #include <stdio.h> #include <cassert> #include <cuda_runtime.h> /* Naive implementation. Allocate one thread for one element in result matrix, processing dot(Arow, Bcol); */ __global__ void kMatrixMul0 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2){ int x = blockDim.x*blockIdx.x + threadIdx.x; int y = blockDim.y*blockIdx.y + threadIdx.y; if (x >= n2 || y >= m1){return;} float sum=0.0; for (int i=0; i<m2; i++){ // if (blockIdx.x == 0 && threadIdx.x == 0 && blockIdx.y == 0 && threadIdx.y == 0) { // printf("sum:%f mat1:%f mat2:%f i:%d\n", sum, d_mat1[y*m2+i], d_mat2[i*n2+x], i); // } sum += d_mat1[y*m2+i] * d_mat2[i*n2+x]; } d_res[n2*y+x] = sum; }; /* Using shared memory */ template<int BLOCK_SIZE> __global__ void kMatrixMul1 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data As[ty][tx] = d_mat1[a + m2 * ty + tx]; Bs[ty][tx] = d_mat2[b + n2 * ty + tx]; __syncthreads(); for (int k = 0; k<blockDim.x; k++){ c += As[ty][k] * Bs[k][tx]; } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } /* Using register blocking */ template<int BLOCK_SIZE, int REGT_SIZE> __global__ void kMatrixMul2 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; int reg_tiles = BLOCK_SIZE / REGT_SIZE; float acc[REGT_SIZE][REGT_SIZE]; float Ar; float Br[REGT_SIZE]; #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ acc[i][j] = 0.0f; } } //block 32x32, each thread work on 2x2 for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ As[ty+i][tx+j] = d_mat1[a + m2 * (ty + i) + tx + j]; Bs[ty+i][tx+j] = d_mat2[b + n2 * (ty + i) + tx + j]; } } __syncthreads(); for (int k=0; k<BLOCK_SIZE; k+=REGT_SIZE){ #pragma unroll for (int i=0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ int ax = k + j; int ay = k + i; acc[ty+i][tx+j] += As[ty][k] * Bs[k][tx]; } } } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } float* MatrixMultGPU0(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ float *d_res, *d_mat1, *d_mat2; cudaEvent_t start, end; cudaError_t error; error = cudaEventCreate(&start); error = cudaEventCreate(&end); //malloc the device memory for matrices cudaError_t result = cudaMalloc((void**)&d_res, sizeof(float)*m1*n2); result = cudaMalloc((void**)&d_mat1, sizeof(float)*m1*m2); assert (result == cudaSuccess); result = cudaMalloc((void**)&d_mat2, sizeof(float)*n1*n2); assert (result == cudaSuccess); //init source matrices in device memory result = cudaMemcpy(d_mat1, mat1, sizeof(float)*m1*m2, cudaMemcpyHostToDevice); assert (result == cudaSuccess); result = cudaMemcpy(d_mat2, mat2, sizeof(float)*n1*n2, cudaMemcpyHostToDevice); assert (result == cudaSuccess); cudaEventRecord(start, NULL); int N = 64; dim3 block_size(N, N); //grid width in blocks int grid_wib = ceil(float(n2)/float(N)); //grid height in blocks int grid_hib = ceil(float(m1)/float(N)); dim3 grid_size(grid_wib, grid_hib); //naive version //kMatrixMul0<<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //sharedMem version kMatrixMul1<64><<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //copy back the multiplication result float* res = new float[m1*n2]; result = cudaMemcpy(res, d_res, sizeof(float)*m1*n2, cudaMemcpyDeviceToHost); assert (result == cudaSuccess); cudaEventRecord(end, NULL); error = cudaEventSynchronize(end); float msecTotal = 0.0f; error = cudaEventElapsedTime(&msecTotal, start, end); printf("calculation Time:%f ms\n", msecTotal); cudaFree(d_res); cudaFree(d_mat1); cudaFree(d_mat2); return res; } float* MatrixMulCPU(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ assert(m2 == n1 && "matrx a's cols != matrix b's rows"); float* res = new float[m1*n2] ; for (int i = 0; i < m1; i++) { for (int j = 0; j < n2; j++) { res[i*m1 + j] = 0; for (int x = 0; x < m2; x++) { res[i*m1+j] += mat1[i*m1+x] * mat2[x*n1+j]; } } } return res; } float CompareMatrix(float mat1[], float mat2[], int m1, int m2){ float err = 0; for (int x=0; x<m1; x++){ for (int y=0; y<m2; y++){ err += mat1[m2*y+x] - mat2[m2*y+x]; } } return err; /* if (err > 0.1){ printf("matrix comparison failed.error:%f\n", err); return false; } return true; */ } void FillMatrix(float mat[], int m1, int m2, float d=1.0){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ if (d<0.0) { mat[i*m1+j] = static_cast<float>(rand())/static_cast<float>(RAND_MAX) * 10.0; } else { mat[i*m1+j] = d; } } } } void PrintMatrix(float mat[], int m1, int m2){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ printf("%f,", mat[i*m1+j]); } printf("\n"); } } int main(int argc, char *argv[]){ int m1 = 1024; int m2; int n1 = m2 = 1024; int n2 = 1024; float *matrix_a = new float[m1*m2]; float *matrix_b = new float[n1*n2]; FillMatrix((float*)matrix_a, m1, m2); FillMatrix((float*)matrix_b, n1, n2); //float *ref = MatrixMulCPU((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); float *res = MatrixMultGPU0((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); //printf("error:%f\n", CompareMatrix(ref, res, m1, n2)); //PrintMatrix(res, m1, m2); delete[] matrix_a; delete[] matrix_b; delete res; //delete ref; }
#include <hip/hip_runtime.h> #include <cstdlib> #include <stdio.h> #include <cassert> #include <cuda_runtime.h> /* Naive implementation. Allocate one thread for one element in result matrix, processing dot(Arow, Bcol); */ __global__ void kMatrixMul0 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2){ int x = blockDim.x*blockIdx.x + threadIdx.x; int y = blockDim.y*blockIdx.y + threadIdx.y; if (x >= n2 || y >= m1){return;} float sum=0.0; for (int i=0; i<m2; i++){ // if (blockIdx.x == 0 && threadIdx.x == 0 && blockIdx.y == 0 && threadIdx.y == 0) { // printf("sum:%f mat1:%f mat2:%f i:%d\n", sum, d_mat1[y*m2+i], d_mat2[i*n2+x], i); // } sum += d_mat1[y*m2+i] * d_mat2[i*n2+x]; } d_res[n2*y+x] = sum; }; /* Using shared memory */ template<int BLOCK_SIZE> __global__ void kMatrixMul1 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data As[ty][tx] = d_mat1[a + m2 * ty + tx]; Bs[ty][tx] = d_mat2[b + n2 * ty + tx]; __syncthreads(); for (int k = 0; k<blockDim.x; k++){ c += As[ty][k] * Bs[k][tx]; } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } /* Using register blocking */ template<int BLOCK_SIZE, int REGT_SIZE> __global__ void kMatrixMul2 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; int reg_tiles = BLOCK_SIZE / REGT_SIZE; float acc[REGT_SIZE][REGT_SIZE]; float Ar; float Br[REGT_SIZE]; #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ acc[i][j] = 0.0f; } } //block 32x32, each thread work on 2x2 for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ As[ty+i][tx+j] = d_mat1[a + m2 * (ty + i) + tx + j]; Bs[ty+i][tx+j] = d_mat2[b + n2 * (ty + i) + tx + j]; } } __syncthreads(); for (int k=0; k<BLOCK_SIZE; k+=REGT_SIZE){ #pragma unroll for (int i=0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ int ax = k + j; int ay = k + i; acc[ty+i][tx+j] += As[ty][k] * Bs[k][tx]; } } } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } float* MatrixMultGPU0(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ float *d_res, *d_mat1, *d_mat2; hipEvent_t start, end; hipError_t error; error = hipEventCreate(&start); error = hipEventCreate(&end); //malloc the device memory for matrices hipError_t result = hipMalloc((void**)&d_res, sizeof(float)*m1*n2); result = hipMalloc((void**)&d_mat1, sizeof(float)*m1*m2); assert (result == hipSuccess); result = hipMalloc((void**)&d_mat2, sizeof(float)*n1*n2); assert (result == hipSuccess); //init source matrices in device memory result = hipMemcpy(d_mat1, mat1, sizeof(float)*m1*m2, hipMemcpyHostToDevice); assert (result == hipSuccess); result = hipMemcpy(d_mat2, mat2, sizeof(float)*n1*n2, hipMemcpyHostToDevice); assert (result == hipSuccess); hipEventRecord(start, NULL); int N = 64; dim3 block_size(N, N); //grid width in blocks int grid_wib = ceil(float(n2)/float(N)); //grid height in blocks int grid_hib = ceil(float(m1)/float(N)); dim3 grid_size(grid_wib, grid_hib); //naive version //kMatrixMul0<<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //sharedMem version kMatrixMul1<64><<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //copy back the multiplication result float* res = new float[m1*n2]; result = hipMemcpy(res, d_res, sizeof(float)*m1*n2, hipMemcpyDeviceToHost); assert (result == hipSuccess); hipEventRecord(end, NULL); error = hipEventSynchronize(end); float msecTotal = 0.0f; error = hipEventElapsedTime(&msecTotal, start, end); printf("calculation Time:%f ms\n", msecTotal); hipFree(d_res); hipFree(d_mat1); hipFree(d_mat2); return res; } float* MatrixMulCPU(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ assert(m2 == n1 && "matrx a's cols != matrix b's rows"); float* res = new float[m1*n2] ; for (int i = 0; i < m1; i++) { for (int j = 0; j < n2; j++) { res[i*m1 + j] = 0; for (int x = 0; x < m2; x++) { res[i*m1+j] += mat1[i*m1+x] * mat2[x*n1+j]; } } } return res; } float CompareMatrix(float mat1[], float mat2[], int m1, int m2){ float err = 0; for (int x=0; x<m1; x++){ for (int y=0; y<m2; y++){ err += mat1[m2*y+x] - mat2[m2*y+x]; } } return err; /* if (err > 0.1){ printf("matrix comparison failed.error:%f\n", err); return false; } return true; */ } void FillMatrix(float mat[], int m1, int m2, float d=1.0){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ if (d<0.0) { mat[i*m1+j] = static_cast<float>(rand())/static_cast<float>(RAND_MAX) * 10.0; } else { mat[i*m1+j] = d; } } } } void PrintMatrix(float mat[], int m1, int m2){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ printf("%f,", mat[i*m1+j]); } printf("\n"); } } int main(int argc, char *argv[]){ int m1 = 1024; int m2; int n1 = m2 = 1024; int n2 = 1024; float *matrix_a = new float[m1*m2]; float *matrix_b = new float[n1*n2]; FillMatrix((float*)matrix_a, m1, m2); FillMatrix((float*)matrix_b, n1, n2); //float *ref = MatrixMulCPU((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); float *res = MatrixMultGPU0((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); //printf("error:%f\n", CompareMatrix(ref, res, m1, n2)); //PrintMatrix(res, m1, m2); delete[] matrix_a; delete[] matrix_b; delete res; //delete ref; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> #include <cstdlib> #include <stdio.h> #include <cassert> #include <cuda_runtime.h> /* Naive implementation. Allocate one thread for one element in result matrix, processing dot(Arow, Bcol); */ __global__ void kMatrixMul0 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2){ int x = blockDim.x*blockIdx.x + threadIdx.x; int y = blockDim.y*blockIdx.y + threadIdx.y; if (x >= n2 || y >= m1){return;} float sum=0.0; for (int i=0; i<m2; i++){ // if (blockIdx.x == 0 && threadIdx.x == 0 && blockIdx.y == 0 && threadIdx.y == 0) { // printf("sum:%f mat1:%f mat2:%f i:%d\n", sum, d_mat1[y*m2+i], d_mat2[i*n2+x], i); // } sum += d_mat1[y*m2+i] * d_mat2[i*n2+x]; } d_res[n2*y+x] = sum; }; /* Using shared memory */ template<int BLOCK_SIZE> __global__ void kMatrixMul1 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data As[ty][tx] = d_mat1[a + m2 * ty + tx]; Bs[ty][tx] = d_mat2[b + n2 * ty + tx]; __syncthreads(); for (int k = 0; k<blockDim.x; k++){ c += As[ty][k] * Bs[k][tx]; } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } /* Using register blocking */ template<int BLOCK_SIZE, int REGT_SIZE> __global__ void kMatrixMul2 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; int reg_tiles = BLOCK_SIZE / REGT_SIZE; float acc[REGT_SIZE][REGT_SIZE]; float Ar; float Br[REGT_SIZE]; #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ acc[i][j] = 0.0f; } } //block 32x32, each thread work on 2x2 for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ As[ty+i][tx+j] = d_mat1[a + m2 * (ty + i) + tx + j]; Bs[ty+i][tx+j] = d_mat2[b + n2 * (ty + i) + tx + j]; } } __syncthreads(); for (int k=0; k<BLOCK_SIZE; k+=REGT_SIZE){ #pragma unroll for (int i=0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ int ax = k + j; int ay = k + i; acc[ty+i][tx+j] += As[ty][k] * Bs[k][tx]; } } } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } float* MatrixMultGPU0(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ float *d_res, *d_mat1, *d_mat2; hipEvent_t start, end; hipError_t error; error = hipEventCreate(&start); error = hipEventCreate(&end); //malloc the device memory for matrices hipError_t result = hipMalloc((void**)&d_res, sizeof(float)*m1*n2); result = hipMalloc((void**)&d_mat1, sizeof(float)*m1*m2); assert (result == hipSuccess); result = hipMalloc((void**)&d_mat2, sizeof(float)*n1*n2); assert (result == hipSuccess); //init source matrices in device memory result = hipMemcpy(d_mat1, mat1, sizeof(float)*m1*m2, hipMemcpyHostToDevice); assert (result == hipSuccess); result = hipMemcpy(d_mat2, mat2, sizeof(float)*n1*n2, hipMemcpyHostToDevice); assert (result == hipSuccess); hipEventRecord(start, NULL); int N = 64; dim3 block_size(N, N); //grid width in blocks int grid_wib = ceil(float(n2)/float(N)); //grid height in blocks int grid_hib = ceil(float(m1)/float(N)); dim3 grid_size(grid_wib, grid_hib); //naive version //kMatrixMul0<<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //sharedMem version kMatrixMul1<64><<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //copy back the multiplication result float* res = new float[m1*n2]; result = hipMemcpy(res, d_res, sizeof(float)*m1*n2, hipMemcpyDeviceToHost); assert (result == hipSuccess); hipEventRecord(end, NULL); error = hipEventSynchronize(end); float msecTotal = 0.0f; error = hipEventElapsedTime(&msecTotal, start, end); printf("calculation Time:%f ms\n", msecTotal); hipFree(d_res); hipFree(d_mat1); hipFree(d_mat2); return res; } float* MatrixMulCPU(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ assert(m2 == n1 && "matrx a's cols != matrix b's rows"); float* res = new float[m1*n2] ; for (int i = 0; i < m1; i++) { for (int j = 0; j < n2; j++) { res[i*m1 + j] = 0; for (int x = 0; x < m2; x++) { res[i*m1+j] += mat1[i*m1+x] * mat2[x*n1+j]; } } } return res; } float CompareMatrix(float mat1[], float mat2[], int m1, int m2){ float err = 0; for (int x=0; x<m1; x++){ for (int y=0; y<m2; y++){ err += mat1[m2*y+x] - mat2[m2*y+x]; } } return err; /* if (err > 0.1){ printf("matrix comparison failed.error:%f\n", err); return false; } return true; */ } void FillMatrix(float mat[], int m1, int m2, float d=1.0){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ if (d<0.0) { mat[i*m1+j] = static_cast<float>(rand())/static_cast<float>(RAND_MAX) * 10.0; } else { mat[i*m1+j] = d; } } } } void PrintMatrix(float mat[], int m1, int m2){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ printf("%f,", mat[i*m1+j]); } printf("\n"); } } int main(int argc, char *argv[]){ int m1 = 1024; int m2; int n1 = m2 = 1024; int n2 = 1024; float *matrix_a = new float[m1*m2]; float *matrix_b = new float[n1*n2]; FillMatrix((float*)matrix_a, m1, m2); FillMatrix((float*)matrix_b, n1, n2); //float *ref = MatrixMulCPU((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); float *res = MatrixMultGPU0((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); //printf("error:%f\n", CompareMatrix(ref, res, m1, n2)); //PrintMatrix(res, m1, m2); delete[] matrix_a; delete[] matrix_b; delete res; //delete ref; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z11kMatrixMul0PfS_iiS_ii .globl _Z11kMatrixMul0PfS_iiS_ii .p2align 8 .type _Z11kMatrixMul0PfS_iiS_ii,@function _Z11kMatrixMul0PfS_iiS_ii: s_clause 0x2 s_load_b32 s2, s[0:1], 0x34 s_load_b32 s4, s[0:1], 0x24 s_load_b32 s3, s[0:1], 0x10 v_and_b32_e32 v2, 0x3ff, v0 v_bfe_u32 v3, v0, 10, 10 s_waitcnt lgkmcnt(0) s_and_b32 s5, s2, 0xffff s_lshr_b32 s2, s2, 16 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_mad_u64_u32 v[0:1], null, s14, s5, v[2:3] v_mad_u64_u32 v[1:2], null, s15, s2, v[3:4] v_cmp_gt_i32_e32 vcc_lo, s4, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_gt_i32_e64 s2, s3, v1 s_and_b32 s2, vcc_lo, s2 s_delay_alu instid0(SALU_CYCLE_1) s_and_saveexec_b32 s3, s2 s_cbranch_execz .LBB0_6 s_load_b32 s5, s[0:1], 0x14 s_waitcnt lgkmcnt(0) s_cmp_lt_i32 s5, 1 s_cbranch_scc1 .LBB0_4 s_clause 0x1 s_load_b64 s[6:7], s[0:1], 0x8 s_load_b64 s[2:3], s[0:1], 0x18 v_mul_lo_u32 v2, v1, s5 v_mov_b32_e32 v6, 0 v_mov_b32_e32 v4, v0 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_ashrrev_i32_e32 v3, 31, v2 v_lshlrev_b64 v[2:3], 2, v[2:3] s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v2, vcc_lo, s6, v2 v_add_co_ci_u32_e32 v3, vcc_lo, s7, v3, vcc_lo .p2align 6 .LBB0_3: v_ashrrev_i32_e32 v5, 31, v4 s_add_i32 s5, s5, -1 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) s_cmp_eq_u32 s5, 0 v_lshlrev_b64 v[7:8], 2, v[4:5] v_add_nc_u32_e32 v4, s4, v4 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v7, vcc_lo, s2, v7 v_add_co_ci_u32_e32 v8, vcc_lo, s3, v8, vcc_lo global_load_b32 v5, v[2:3], off global_load_b32 v7, v[7:8], off v_add_co_u32 v2, vcc_lo, v2, 4 v_add_co_ci_u32_e32 v3, vcc_lo, 0, v3, vcc_lo s_waitcnt vmcnt(0) v_fmac_f32_e32 v6, v5, v7 s_cbranch_scc0 .LBB0_3 s_branch .LBB0_5 .LBB0_4: v_mov_b32_e32 v6, 0 .LBB0_5: s_load_b64 s[0:1], s[0:1], 0x0 v_mad_u64_u32 v[2:3], null, v1, s4, v[0:1] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_ashrrev_i32_e32 v3, 31, v2 v_lshlrev_b64 v[0:1], 2, v[2:3] s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v0, vcc_lo, s0, v0 v_add_co_ci_u32_e32 v1, vcc_lo, s1, v1, vcc_lo global_store_b32 v[0:1], v6, off .LBB0_6: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z11kMatrixMul0PfS_iiS_ii .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 296 .amdhsa_user_sgpr_count 14 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 1 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 1 .amdhsa_next_free_vgpr 9 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z11kMatrixMul0PfS_iiS_ii, .Lfunc_end0-_Z11kMatrixMul0PfS_iiS_ii .section .AMDGPU.csdata,"",@progbits .section .text._Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,"axG",@progbits,_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,comdat .protected _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .globl _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .p2align 8 .type _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,@function _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii: s_clause 0x2 s_load_b32 s2, s[0:1], 0x34 s_load_b32 s9, s[0:1], 0x14 s_load_b32 s3, s[0:1], 0x24 v_dual_mov_b32 v4, 0 :: v_dual_and_b32 v1, 0x3ff, v0 v_bfe_u32 v0, v0, 10, 10 s_waitcnt lgkmcnt(0) s_lshr_b32 s13, s2, 16 s_and_b32 s12, s2, 0xffff s_mul_i32 s8, s15, s13 s_mul_i32 s10, s14, s12 s_mul_i32 s11, s8, s9 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_ge_i32 s11, s9 s_cbranch_scc1 .LBB1_5 s_clause 0x1 s_load_b64 s[4:5], s[0:1], 0x8 s_load_b64 s[6:7], s[0:1], 0x18 v_mad_u64_u32 v[2:3], null, v0, s9, v[1:2] v_lshlrev_b32_e32 v4, 2, v1 s_cmp_lg_u32 s12, 0 v_lshlrev_b32_e32 v5, 8, v0 s_cselect_b32 s2, -1, 0 s_mul_i32 s13, s13, s3 v_or_b32_e32 v7, 0x4000, v4 v_cndmask_b32_e64 v9, 0, 1, s2 v_add_nc_u32_e32 v6, v5, v4 v_mad_u64_u32 v[3:4], null, v0, s3, v[1:2] v_mov_b32_e32 v4, 0 v_add_nc_u32_e32 v8, v7, v5 v_cmp_ne_u32_e64 s2, 1, v9 s_mov_b32 s14, s10 s_set_inst_prefetch_distance 0x1 s_branch .LBB1_3 .p2align 6 .LBB1_2: s_add_i32 s11, s11, s12 s_add_i32 s14, s14, s13 s_cmp_ge_i32 s11, s9 s_barrier buffer_gl0_inv s_cbranch_scc1 .LBB1_5 .LBB1_3: v_add_nc_u32_e32 v9, s11, v2 v_add_nc_u32_e32 v11, s14, v3 s_mov_b32 s15, s12 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v10, 31, v9 v_ashrrev_i32_e32 v12, 31, v11 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[9:10], 2, v[9:10] v_lshlrev_b64 v[11:12], 2, v[11:12] s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v9, vcc_lo, s4, v9 v_add_co_ci_u32_e32 v10, vcc_lo, s5, v10, vcc_lo s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_4) v_add_co_u32 v11, vcc_lo, s6, v11 v_add_co_ci_u32_e32 v12, vcc_lo, s7, v12, vcc_lo s_and_b32 vcc_lo, exec_lo, s2 global_load_b32 v13, v[9:10], off global_load_b32 v11, v[11:12], off v_dual_mov_b32 v10, v5 :: v_dual_mov_b32 v9, v7 s_waitcnt vmcnt(1) ds_store_b32 v6, v13 s_waitcnt vmcnt(0) ds_store_b32 v8, v11 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_cbranch_vccnz .LBB1_2 .LBB1_4: ds_load_b32 v11, v10 ds_load_b32 v12, v9 v_add_nc_u32_e32 v10, 4, v10 v_add_nc_u32_e32 v9, 0x100, v9 s_add_i32 s15, s15, -1 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_eq_u32 s15, 0 s_waitcnt lgkmcnt(0) v_fmac_f32_e32 v4, v11, v12 s_cbranch_scc0 .LBB1_4 s_branch .LBB1_2 .LBB1_5: s_set_inst_prefetch_distance 0x2 v_add_nc_u32_e32 v0, s8, v0 s_load_b64 s[0:1], s[0:1], 0x0 v_mov_b32_e32 v2, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_lo_u32 v0, v0, s3 v_add3_u32 v1, s10, v1, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[0:1], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v0, vcc_lo, s0, v0 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v1, vcc_lo, s1, v1, vcc_lo global_store_b32 v[0:1], v4, off s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .amdhsa_group_segment_fixed_size 32768 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 296 .amdhsa_user_sgpr_count 14 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 1 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 1 .amdhsa_next_free_vgpr 14 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .section .text._Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,"axG",@progbits,_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,comdat .Lfunc_end1: .size _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, .Lfunc_end1-_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .offset: 16 .size: 4 .value_kind: by_value - .offset: 20 .size: 4 .value_kind: by_value - .address_space: global .offset: 24 .size: 8 .value_kind: global_buffer - .offset: 32 .size: 4 .value_kind: by_value - .offset: 36 .size: 4 .value_kind: by_value - .offset: 40 .size: 4 .value_kind: hidden_block_count_x - .offset: 44 .size: 4 .value_kind: hidden_block_count_y - .offset: 48 .size: 4 .value_kind: hidden_block_count_z - .offset: 52 .size: 2 .value_kind: hidden_group_size_x - .offset: 54 .size: 2 .value_kind: hidden_group_size_y - .offset: 56 .size: 2 .value_kind: hidden_group_size_z - .offset: 58 .size: 2 .value_kind: hidden_remainder_x - .offset: 60 .size: 2 .value_kind: hidden_remainder_y - .offset: 62 .size: 2 .value_kind: hidden_remainder_z - .offset: 80 .size: 8 .value_kind: hidden_global_offset_x - .offset: 88 .size: 8 .value_kind: hidden_global_offset_y - .offset: 96 .size: 8 .value_kind: hidden_global_offset_z - .offset: 104 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 296 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z11kMatrixMul0PfS_iiS_ii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z11kMatrixMul0PfS_iiS_ii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 9 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .offset: 16 .size: 4 .value_kind: by_value - .offset: 20 .size: 4 .value_kind: by_value - .address_space: global .offset: 24 .size: 8 .value_kind: global_buffer - .offset: 32 .size: 4 .value_kind: by_value - .offset: 36 .size: 4 .value_kind: by_value - .offset: 40 .size: 4 .value_kind: hidden_block_count_x - .offset: 44 .size: 4 .value_kind: hidden_block_count_y - .offset: 48 .size: 4 .value_kind: hidden_block_count_z - .offset: 52 .size: 2 .value_kind: hidden_group_size_x - .offset: 54 .size: 2 .value_kind: hidden_group_size_y - .offset: 56 .size: 2 .value_kind: hidden_group_size_z - .offset: 58 .size: 2 .value_kind: hidden_remainder_x - .offset: 60 .size: 2 .value_kind: hidden_remainder_y - .offset: 62 .size: 2 .value_kind: hidden_remainder_z - .offset: 80 .size: 8 .value_kind: hidden_global_offset_x - .offset: 88 .size: 8 .value_kind: hidden_global_offset_y - .offset: 96 .size: 8 .value_kind: hidden_global_offset_z - .offset: 104 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 32768 .kernarg_segment_align: 8 .kernarg_segment_size: 296 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 14 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <hip/hip_runtime.h> #include <cstdlib> #include <stdio.h> #include <cassert> #include <cuda_runtime.h> /* Naive implementation. Allocate one thread for one element in result matrix, processing dot(Arow, Bcol); */ __global__ void kMatrixMul0 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2){ int x = blockDim.x*blockIdx.x + threadIdx.x; int y = blockDim.y*blockIdx.y + threadIdx.y; if (x >= n2 || y >= m1){return;} float sum=0.0; for (int i=0; i<m2; i++){ // if (blockIdx.x == 0 && threadIdx.x == 0 && blockIdx.y == 0 && threadIdx.y == 0) { // printf("sum:%f mat1:%f mat2:%f i:%d\n", sum, d_mat1[y*m2+i], d_mat2[i*n2+x], i); // } sum += d_mat1[y*m2+i] * d_mat2[i*n2+x]; } d_res[n2*y+x] = sum; }; /* Using shared memory */ template<int BLOCK_SIZE> __global__ void kMatrixMul1 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data As[ty][tx] = d_mat1[a + m2 * ty + tx]; Bs[ty][tx] = d_mat2[b + n2 * ty + tx]; __syncthreads(); for (int k = 0; k<blockDim.x; k++){ c += As[ty][k] * Bs[k][tx]; } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } /* Using register blocking */ template<int BLOCK_SIZE, int REGT_SIZE> __global__ void kMatrixMul2 (float *d_res, float *d_mat1, int m1, int m2, float *d_mat2, int n1, int n2) { //assume squre block __shared__ float As[BLOCK_SIZE][BLOCK_SIZE]; __shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE]; //trivial opt: calculate these in CPU, shared by all blocks int aStart = blockDim.y*blockIdx.y * m2; int bStart = blockDim.x*blockIdx.x; int bStep = blockDim.y*n2; int tx = threadIdx.x; int ty = threadIdx.y; float c = 0.0f; int reg_tiles = BLOCK_SIZE / REGT_SIZE; float acc[REGT_SIZE][REGT_SIZE]; float Ar; float Br[REGT_SIZE]; #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ acc[i][j] = 0.0f; } } //block 32x32, each thread work on 2x2 for (int a = aStart, b = bStart; a < m2; a += blockDim.x, b += bStep) { //load data #pragma unroll for (int i = 0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ As[ty+i][tx+j] = d_mat1[a + m2 * (ty + i) + tx + j]; Bs[ty+i][tx+j] = d_mat2[b + n2 * (ty + i) + tx + j]; } } __syncthreads(); for (int k=0; k<BLOCK_SIZE; k+=REGT_SIZE){ #pragma unroll for (int i=0; i<REGT_SIZE; i++){ #pragma unroll for (int j = 0; j<REGT_SIZE; j++){ int ax = k + j; int ay = k + i; acc[ty+i][tx+j] += As[ty][k] * Bs[k][tx]; } } } __syncthreads(); } d_res[(blockDim.y*blockIdx.y+ty)*n2+blockDim.x*blockIdx.x+tx] = c; } float* MatrixMultGPU0(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ float *d_res, *d_mat1, *d_mat2; hipEvent_t start, end; hipError_t error; error = hipEventCreate(&start); error = hipEventCreate(&end); //malloc the device memory for matrices hipError_t result = hipMalloc((void**)&d_res, sizeof(float)*m1*n2); result = hipMalloc((void**)&d_mat1, sizeof(float)*m1*m2); assert (result == hipSuccess); result = hipMalloc((void**)&d_mat2, sizeof(float)*n1*n2); assert (result == hipSuccess); //init source matrices in device memory result = hipMemcpy(d_mat1, mat1, sizeof(float)*m1*m2, hipMemcpyHostToDevice); assert (result == hipSuccess); result = hipMemcpy(d_mat2, mat2, sizeof(float)*n1*n2, hipMemcpyHostToDevice); assert (result == hipSuccess); hipEventRecord(start, NULL); int N = 64; dim3 block_size(N, N); //grid width in blocks int grid_wib = ceil(float(n2)/float(N)); //grid height in blocks int grid_hib = ceil(float(m1)/float(N)); dim3 grid_size(grid_wib, grid_hib); //naive version //kMatrixMul0<<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //sharedMem version kMatrixMul1<64><<<grid_size, block_size>>>(d_res, d_mat1, m1, m2, d_mat2, n1, n2); //copy back the multiplication result float* res = new float[m1*n2]; result = hipMemcpy(res, d_res, sizeof(float)*m1*n2, hipMemcpyDeviceToHost); assert (result == hipSuccess); hipEventRecord(end, NULL); error = hipEventSynchronize(end); float msecTotal = 0.0f; error = hipEventElapsedTime(&msecTotal, start, end); printf("calculation Time:%f ms\n", msecTotal); hipFree(d_res); hipFree(d_mat1); hipFree(d_mat2); return res; } float* MatrixMulCPU(float *mat1, int m1, int m2, float *mat2, int n1, int n2){ assert(m2 == n1 && "matrx a's cols != matrix b's rows"); float* res = new float[m1*n2] ; for (int i = 0; i < m1; i++) { for (int j = 0; j < n2; j++) { res[i*m1 + j] = 0; for (int x = 0; x < m2; x++) { res[i*m1+j] += mat1[i*m1+x] * mat2[x*n1+j]; } } } return res; } float CompareMatrix(float mat1[], float mat2[], int m1, int m2){ float err = 0; for (int x=0; x<m1; x++){ for (int y=0; y<m2; y++){ err += mat1[m2*y+x] - mat2[m2*y+x]; } } return err; /* if (err > 0.1){ printf("matrix comparison failed.error:%f\n", err); return false; } return true; */ } void FillMatrix(float mat[], int m1, int m2, float d=1.0){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ if (d<0.0) { mat[i*m1+j] = static_cast<float>(rand())/static_cast<float>(RAND_MAX) * 10.0; } else { mat[i*m1+j] = d; } } } } void PrintMatrix(float mat[], int m1, int m2){ for (int i=0; i<m1; i++){ for (int j=0; j<m2; j++){ printf("%f,", mat[i*m1+j]); } printf("\n"); } } int main(int argc, char *argv[]){ int m1 = 1024; int m2; int n1 = m2 = 1024; int n2 = 1024; float *matrix_a = new float[m1*m2]; float *matrix_b = new float[n1*n2]; FillMatrix((float*)matrix_a, m1, m2); FillMatrix((float*)matrix_b, n1, n2); //float *ref = MatrixMulCPU((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); float *res = MatrixMultGPU0((float*)matrix_a, m1, m2, (float*)matrix_b, n1, n2); //printf("error:%f\n", CompareMatrix(ref, res, m1, n2)); //PrintMatrix(res, m1, m2); delete[] matrix_a; delete[] matrix_b; delete res; //delete ref; }
.text .file "matrixmult.hip" .globl _Z26__device_stub__kMatrixMul0PfS_iiS_ii # -- Begin function _Z26__device_stub__kMatrixMul0PfS_iiS_ii .p2align 4, 0x90 .type _Z26__device_stub__kMatrixMul0PfS_iiS_ii,@function _Z26__device_stub__kMatrixMul0PfS_iiS_ii: # @_Z26__device_stub__kMatrixMul0PfS_iiS_ii .cfi_startproc # %bb.0: subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 88(%rsp) movq %rsi, 80(%rsp) movl %edx, 20(%rsp) movl %ecx, 16(%rsp) movq %r8, 72(%rsp) movl %r9d, 12(%rsp) leaq 88(%rsp), %rax movq %rax, 96(%rsp) leaq 80(%rsp), %rax movq %rax, 104(%rsp) leaq 20(%rsp), %rax movq %rax, 112(%rsp) leaq 16(%rsp), %rax movq %rax, 120(%rsp) leaq 72(%rsp), %rax movq %rax, 128(%rsp) leaq 12(%rsp), %rax movq %rax, 136(%rsp) leaq 160(%rsp), %rax movq %rax, 144(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 96(%rsp), %r9 movl $_Z11kMatrixMul0PfS_iiS_ii, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $168, %rsp .cfi_adjust_cfa_offset -168 retq .Lfunc_end0: .size _Z26__device_stub__kMatrixMul0PfS_iiS_ii, .Lfunc_end0-_Z26__device_stub__kMatrixMul0PfS_iiS_ii .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function _Z14MatrixMultGPU0PfiiS_ii .LCPI1_0: .long 0x3c800000 # float 0.015625 .text .globl _Z14MatrixMultGPU0PfiiS_ii .p2align 4, 0x90 .type _Z14MatrixMultGPU0PfiiS_ii,@function _Z14MatrixMultGPU0PfiiS_ii: # @_Z14MatrixMultGPU0PfiiS_ii .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $232, %rsp .cfi_def_cfa_offset 288 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %r9d, %r15d movl %r8d, %r13d movq %rcx, 80(%rsp) # 8-byte Spill movl %edx, %r12d movl %esi, %ebx movq %rdi, 72(%rsp) # 8-byte Spill leaq 64(%rsp), %rdi callq hipEventCreate leaq 8(%rsp), %rdi callq hipEventCreate movslq %ebx, %rax movq %rax, 88(%rsp) # 8-byte Spill leaq (,%rax,4), %rbp movslq %r15d, %r14 movq %rbp, %rsi imulq %r14, %rsi leaq 32(%rsp), %rdi movq %rsi, 96(%rsp) # 8-byte Spill callq hipMalloc movl %r12d, 40(%rsp) # 4-byte Spill movslq %r12d, %r12 imulq %rbp, %r12 leaq 24(%rsp), %rdi movq %r12, %rsi callq hipMalloc movl %r13d, 44(%rsp) # 4-byte Spill movslq %r13d, %rbp imulq %r14, %rbp shlq $2, %rbp leaq 16(%rsp), %rdi movq %rbp, %rsi callq hipMalloc movq 24(%rsp), %rdi movq 72(%rsp), %rsi # 8-byte Reload movq %r12, %rdx movl $1, %ecx callq hipMemcpy movq 16(%rsp), %rdi movq 80(%rsp), %rsi # 8-byte Reload movq %rbp, %rdx movl $1, %ecx callq hipMemcpy movq 64(%rsp), %rdi xorl %esi, %esi callq hipEventRecord cvtsi2ss %r14d, %xmm0 mulss .LCPI1_0(%rip), %xmm0 callq ceilf@PLT cvttss2si %xmm0, %r14d xorps %xmm0, %xmm0 cvtsi2ssl 88(%rsp), %xmm0 # 4-byte Folded Reload mulss .LCPI1_0(%rip), %xmm0 callq ceilf@PLT cvttss2si %xmm0, %edi shlq $32, %rdi orq %r14, %rdi movabsq $274877907008, %rdx # imm = 0x4000000040 movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_2 # %bb.1: movq 32(%rsp), %rax movq 24(%rsp), %rcx movq 16(%rsp), %rdx movq %rax, 168(%rsp) movq %rcx, 160(%rsp) movl %ebx, 60(%rsp) movl 40(%rsp), %eax # 4-byte Reload movl %eax, 56(%rsp) movq %rdx, 152(%rsp) movl 44(%rsp), %eax # 4-byte Reload movl %eax, 52(%rsp) movl %r15d, 48(%rsp) leaq 168(%rsp), %rax movq %rax, 176(%rsp) leaq 160(%rsp), %rax movq %rax, 184(%rsp) leaq 60(%rsp), %rax movq %rax, 192(%rsp) leaq 56(%rsp), %rax movq %rax, 200(%rsp) leaq 152(%rsp), %rax movq %rax, 208(%rsp) leaq 52(%rsp), %rax movq %rax, 216(%rsp) leaq 48(%rsp), %rax movq %rax, 224(%rsp) leaq 136(%rsp), %rdi leaq 120(%rsp), %rsi leaq 112(%rsp), %rdx leaq 104(%rsp), %rcx callq __hipPopCallConfiguration movq 136(%rsp), %rsi movl 144(%rsp), %edx movq 120(%rsp), %rcx movl 128(%rsp), %r8d leaq 176(%rsp), %r9 movl $_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, %edi pushq 104(%rsp) .cfi_adjust_cfa_offset 8 pushq 120(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_2: imull %ebx, %r15d movslq %r15d, %rax leaq (,%rax,4), %rcx testl %eax, %eax movq $-1, %rdi cmovnsq %rcx, %rdi callq _Znam movq %rax, %rbx movq 32(%rsp), %rsi movq %rax, %rdi movq 96(%rsp), %rdx # 8-byte Reload movl $2, %ecx callq hipMemcpy movq 8(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq 8(%rsp), %rdi callq hipEventSynchronize movl $0, 176(%rsp) movq 64(%rsp), %rsi movq 8(%rsp), %rdx leaq 176(%rsp), %rdi callq hipEventElapsedTime movss 176(%rsp), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str, %edi movb $1, %al callq printf movq 32(%rsp), %rdi callq hipFree movq 24(%rsp), %rdi callq hipFree movq 16(%rsp), %rdi callq hipFree movq %rbx, %rax addq $232, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size _Z14MatrixMultGPU0PfiiS_ii, .Lfunc_end1-_Z14MatrixMultGPU0PfiiS_ii .cfi_endproc # -- End function .section .text._Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii,"axG",@progbits,_Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii,comdat .weak _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii # -- Begin function _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .p2align 4, 0x90 .type _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii,@function _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii: # @_Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .cfi_startproc # %bb.0: subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 88(%rsp) movq %rsi, 80(%rsp) movl %edx, 20(%rsp) movl %ecx, 16(%rsp) movq %r8, 72(%rsp) movl %r9d, 12(%rsp) leaq 88(%rsp), %rax movq %rax, 96(%rsp) leaq 80(%rsp), %rax movq %rax, 104(%rsp) leaq 20(%rsp), %rax movq %rax, 112(%rsp) leaq 16(%rsp), %rax movq %rax, 120(%rsp) leaq 72(%rsp), %rax movq %rax, 128(%rsp) leaq 12(%rsp), %rax movq %rax, 136(%rsp) leaq 160(%rsp), %rax movq %rax, 144(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 96(%rsp), %r9 movl $_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $168, %rsp .cfi_adjust_cfa_offset -168 retq .Lfunc_end2: .size _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii, .Lfunc_end2-_Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .cfi_endproc # -- End function .text .globl _Z12MatrixMulCPUPfiiS_ii # -- Begin function _Z12MatrixMulCPUPfiiS_ii .p2align 4, 0x90 .type _Z12MatrixMulCPUPfiiS_ii,@function _Z12MatrixMulCPUPfiiS_ii: # @_Z12MatrixMulCPUPfiiS_ii .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 pushq %rax .cfi_def_cfa_offset 64 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %r9d, %ebx movl %r8d, %r12d movq %rcx, (%rsp) # 8-byte Spill movl %edx, %ebp movl %esi, %r13d movq %rdi, %r15 movl %r9d, %eax imull %esi, %eax cltq leaq (,%rax,4), %rcx testl %eax, %eax movq $-1, %rdi cmovnsq %rcx, %rdi callq _Znam testl %r13d, %r13d jle .LBB3_10 # %bb.1: # %.preheader.lr.ph movslq %r12d, %rcx movl %r13d, %edx movl %ebx, %esi movl %ebp, %edi leaq (,%rdx,4), %r8 shlq $2, %rcx xorl %r9d, %r9d jmp .LBB3_2 .p2align 4, 0x90 .LBB3_9: # %._crit_edge33 # in Loop: Header=BB3_2 Depth=1 incq %r9 addq %r8, %r15 cmpq %rdx, %r9 je .LBB3_10 .LBB3_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB3_4 Depth 2 # Child Loop BB3_6 Depth 3 testl %ebx, %ebx jle .LBB3_9 # %bb.3: # %.lr.ph32 # in Loop: Header=BB3_2 Depth=1 movq %r9, %r10 imulq %rdx, %r10 leaq (%rax,%r10,4), %r10 movq (%rsp), %r11 # 8-byte Reload xorl %r12d, %r12d jmp .LBB3_4 .p2align 4, 0x90 .LBB3_8: # in Loop: Header=BB3_4 Depth=2 incq %r12 addq $4, %r11 cmpq %rsi, %r12 je .LBB3_9 .LBB3_4: # Parent Loop BB3_2 Depth=1 # => This Loop Header: Depth=2 # Child Loop BB3_6 Depth 3 movl $0, (%r10,%r12,4) testl %ebp, %ebp jle .LBB3_8 # %bb.5: # %.lr.ph # in Loop: Header=BB3_4 Depth=2 movss (%r10,%r12,4), %xmm0 # xmm0 = mem[0],zero,zero,zero movq %r11, %r13 xorl %r14d, %r14d .p2align 4, 0x90 .LBB3_6: # Parent Loop BB3_2 Depth=1 # Parent Loop BB3_4 Depth=2 # => This Inner Loop Header: Depth=3 movss (%r15,%r14,4), %xmm1 # xmm1 = mem[0],zero,zero,zero mulss (%r13), %xmm1 addss %xmm1, %xmm0 incq %r14 addq %rcx, %r13 cmpq %r14, %rdi jne .LBB3_6 # %bb.7: # %._crit_edge # in Loop: Header=BB3_4 Depth=2 movss %xmm0, (%r10,%r12,4) jmp .LBB3_8 .LBB3_10: # %._crit_edge35 addq $8, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end3: .size _Z12MatrixMulCPUPfiiS_ii, .Lfunc_end3-_Z12MatrixMulCPUPfiiS_ii .cfi_endproc # -- End function .globl _Z13CompareMatrixPfS_ii # -- Begin function _Z13CompareMatrixPfS_ii .p2align 4, 0x90 .type _Z13CompareMatrixPfS_ii,@function _Z13CompareMatrixPfS_ii: # @_Z13CompareMatrixPfS_ii .cfi_startproc # %bb.0: testl %edx, %edx jle .LBB4_1 # %bb.3: # %.preheader.lr.ph movslq %ecx, %rax movl %edx, %edx movl %eax, %r8d shlq $2, %rax xorps %xmm0, %xmm0 xorl %r9d, %r9d jmp .LBB4_4 .p2align 4, 0x90 .LBB4_7: # %._crit_edge # in Loop: Header=BB4_4 Depth=1 incq %r9 addq $4, %rsi addq $4, %rdi cmpq %rdx, %r9 je .LBB4_2 .LBB4_4: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB4_6 Depth 2 testl %ecx, %ecx jle .LBB4_7 # %bb.5: # %.lr.ph.preheader # in Loop: Header=BB4_4 Depth=1 movq %r8, %r10 xorl %r11d, %r11d .p2align 4, 0x90 .LBB4_6: # %.lr.ph # Parent Loop BB4_4 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rdi,%r11), %xmm1 # xmm1 = mem[0],zero,zero,zero subss (%rsi,%r11), %xmm1 addss %xmm1, %xmm0 addq %rax, %r11 decq %r10 jne .LBB4_6 jmp .LBB4_7 .LBB4_1: xorps %xmm0, %xmm0 .LBB4_2: # %._crit_edge22 retq .Lfunc_end4: .size _Z13CompareMatrixPfS_ii, .Lfunc_end4-_Z13CompareMatrixPfS_ii .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function _Z10FillMatrixPfiif .LCPI5_0: .long 0x30000000 # float 4.65661287E-10 .LCPI5_1: .long 0x41200000 # float 10 .text .globl _Z10FillMatrixPfiif .p2align 4, 0x90 .type _Z10FillMatrixPfiif,@function _Z10FillMatrixPfiif: # @_Z10FillMatrixPfiif .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $24, %rsp .cfi_def_cfa_offset 80 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %edx, (%rsp) # 4-byte Spill movq %rdi, 8(%rsp) # 8-byte Spill testl %esi, %esi jle .LBB5_8 # %bb.1: # %.preheader.lr.ph movl %esi, %ebp movl %esi, %eax movq %rax, 16(%rsp) # 8-byte Spill movl (%rsp), %r12d # 4-byte Reload xorl %r13d, %r13d xorps %xmm2, %xmm2 xorl %r14d, %r14d movss %xmm0, 4(%rsp) # 4-byte Spill jmp .LBB5_2 .p2align 4, 0x90 .LBB5_7: # %._crit_edge # in Loop: Header=BB5_2 Depth=1 incq %r14 addl %ebp, %r13d cmpq 16(%rsp), %r14 # 8-byte Folded Reload je .LBB5_8 .LBB5_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB5_4 Depth 2 cmpl $0, (%rsp) # 4-byte Folded Reload jle .LBB5_7 # %bb.3: # %.lr.ph # in Loop: Header=BB5_2 Depth=1 movl %r13d, %eax movq 8(%rsp), %rcx # 8-byte Reload leaq (%rcx,%rax,4), %rbx xorl %r15d, %r15d jmp .LBB5_4 .p2align 4, 0x90 .LBB5_6: # in Loop: Header=BB5_4 Depth=2 movss %xmm1, (%rbx,%r15,4) incq %r15 cmpq %r15, %r12 je .LBB5_7 .LBB5_4: # Parent Loop BB5_2 Depth=1 # => This Inner Loop Header: Depth=2 ucomiss %xmm0, %xmm2 movaps %xmm0, %xmm1 jbe .LBB5_6 # %bb.5: # in Loop: Header=BB5_4 Depth=2 callq rand xorps %xmm2, %xmm2 movss 4(%rsp), %xmm0 # 4-byte Reload # xmm0 = mem[0],zero,zero,zero xorps %xmm1, %xmm1 cvtsi2ss %eax, %xmm1 mulss .LCPI5_0(%rip), %xmm1 mulss .LCPI5_1(%rip), %xmm1 jmp .LBB5_6 .LBB5_8: # %._crit_edge19 addq $24, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end5: .size _Z10FillMatrixPfiif, .Lfunc_end5-_Z10FillMatrixPfiif .cfi_endproc # -- End function .globl _Z11PrintMatrixPfii # -- Begin function _Z11PrintMatrixPfii .p2align 4, 0x90 .type _Z11PrintMatrixPfii,@function _Z11PrintMatrixPfii: # @_Z11PrintMatrixPfii .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $24, %rsp .cfi_def_cfa_offset 80 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %edx, 4(%rsp) # 4-byte Spill movq %rdi, 8(%rsp) # 8-byte Spill testl %esi, %esi jle .LBB6_6 # %bb.1: # %.preheader.lr.ph movl %esi, %ebp movl %esi, %eax movq %rax, 16(%rsp) # 8-byte Spill movl 4(%rsp), %r12d # 4-byte Reload xorl %r13d, %r13d xorl %r14d, %r14d jmp .LBB6_2 .p2align 4, 0x90 .LBB6_5: # %._crit_edge # in Loop: Header=BB6_2 Depth=1 movl $10, %edi callq putchar@PLT incq %r14 addl %ebp, %r13d cmpq 16(%rsp), %r14 # 8-byte Folded Reload je .LBB6_6 .LBB6_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB6_4 Depth 2 cmpl $0, 4(%rsp) # 4-byte Folded Reload jle .LBB6_5 # %bb.3: # %.lr.ph # in Loop: Header=BB6_2 Depth=1 movl %r13d, %eax movq 8(%rsp), %rcx # 8-byte Reload leaq (%rcx,%rax,4), %rbx xorl %r15d, %r15d .p2align 4, 0x90 .LBB6_4: # Parent Loop BB6_2 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rbx,%r15,4), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str.1, %edi movb $1, %al callq printf incq %r15 cmpq %r15, %r12 jne .LBB6_4 jmp .LBB6_5 .LBB6_6: # %._crit_edge13 addq $24, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end6: .size _Z11PrintMatrixPfii, .Lfunc_end6-_Z11PrintMatrixPfii .cfi_endproc # -- End function .globl main # -- Begin function main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %r15 .cfi_def_cfa_offset 16 pushq %r14 .cfi_def_cfa_offset 24 pushq %rbx .cfi_def_cfa_offset 32 .cfi_offset %rbx, -32 .cfi_offset %r14, -24 .cfi_offset %r15, -16 movl $4194304, %edi # imm = 0x400000 callq _Znam movq %rax, %rbx movl $4194304, %edi # imm = 0x400000 callq _Znam movq %rax, %r14 xorl %eax, %eax movq %rbx, %rcx .p2align 4, 0x90 .LBB7_1: # %.preheader.i # =>This Loop Header: Depth=1 # Child Loop BB7_2 Depth 2 xorl %edx, %edx .p2align 4, 0x90 .LBB7_2: # Parent Loop BB7_1 Depth=1 # => This Inner Loop Header: Depth=2 movl $1065353216, (%rcx,%rdx,4) # imm = 0x3F800000 incq %rdx cmpq $1024, %rdx # imm = 0x400 jne .LBB7_2 # %bb.3: # %._crit_edge.i # in Loop: Header=BB7_1 Depth=1 incq %rax addq $4096, %rcx # imm = 0x1000 cmpq $1024, %rax # imm = 0x400 jne .LBB7_1 # %bb.4: # %.preheader.i19.preheader xorl %eax, %eax movq %r14, %rcx .p2align 4, 0x90 .LBB7_5: # %.preheader.i19 # =>This Loop Header: Depth=1 # Child Loop BB7_6 Depth 2 xorl %edx, %edx .p2align 4, 0x90 .LBB7_6: # Parent Loop BB7_5 Depth=1 # => This Inner Loop Header: Depth=2 movl $1065353216, (%rcx,%rdx,4) # imm = 0x3F800000 incq %rdx cmpq $1024, %rdx # imm = 0x400 jne .LBB7_6 # %bb.7: # %._crit_edge.i24 # in Loop: Header=BB7_5 Depth=1 incq %rax addq $4096, %rcx # imm = 0x1000 cmpq $1024, %rax # imm = 0x400 jne .LBB7_5 # %bb.8: # %_Z10FillMatrixPfiif.exit27 movq %rbx, %rdi movl $1024, %esi # imm = 0x400 movl $1024, %edx # imm = 0x400 movq %r14, %rcx movl $1024, %r8d # imm = 0x400 movl $1024, %r9d # imm = 0x400 callq _Z14MatrixMultGPU0PfiiS_ii movq %rax, %r15 movq %rbx, %rdi callq _ZdaPv movq %r14, %rdi callq _ZdaPv movq %r15, %rdi callq _ZdlPv xorl %eax, %eax popq %rbx .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 retq .Lfunc_end7: .size main, .Lfunc_end7-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 subq $32, %rsp .cfi_def_cfa_offset 48 .cfi_offset %rbx, -16 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB8_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB8_2: movq __hip_gpubin_handle(%rip), %rbx xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z11kMatrixMul0PfS_iiS_ii, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, %esi movl $.L__unnamed_2, %edx movl $.L__unnamed_2, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $32, %rsp .cfi_def_cfa_offset 16 popq %rbx .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end8: .size __hip_module_ctor, .Lfunc_end8-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB9_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB9_2: retq .Lfunc_end9: .size __hip_module_dtor, .Lfunc_end9-__hip_module_dtor .cfi_endproc # -- End function .type _Z11kMatrixMul0PfS_iiS_ii,@object # @_Z11kMatrixMul0PfS_iiS_ii .section .rodata,"a",@progbits .globl _Z11kMatrixMul0PfS_iiS_ii .p2align 3, 0x0 _Z11kMatrixMul0PfS_iiS_ii: .quad _Z26__device_stub__kMatrixMul0PfS_iiS_ii .size _Z11kMatrixMul0PfS_iiS_ii, 8 .type _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,@object # @_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .section .rodata._Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,"aG",@progbits,_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,comdat .weak _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .p2align 3, 0x0 _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii: .quad _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .size _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "calculation Time:%f ms\n" .size .L.str, 24 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz "%f," .size .L.str.1, 4 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z11kMatrixMul0PfS_iiS_ii" .size .L__unnamed_1, 26 .type .L__unnamed_2,@object # @1 .L__unnamed_2: .asciz "_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii" .size .L__unnamed_2, 36 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z26__device_stub__kMatrixMul0PfS_iiS_ii .addrsig_sym _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z11kMatrixMul0PfS_iiS_ii .addrsig_sym _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_000499f5_00000000-6_matrixmult.cudafe1.cpp" .text #APP #NO_APP .type _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii, @function _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii: .LFB2093: .cfi_startproc subq $184, %rsp .cfi_def_cfa_offset 192 movq %rdi, 40(%rsp) movq %rsi, 32(%rsp) movl %edx, 28(%rsp) movl %ecx, 24(%rsp) movq %r8, 16(%rsp) movl %r9d, 12(%rsp) movq %fs:40, %rax movq %rax, 168(%rsp) xorl %eax, %eax leaq 40(%rsp), %rax movq %rax, 112(%rsp) leaq 32(%rsp), %rax movq %rax, 120(%rsp) leaq 28(%rsp), %rax movq %rax, 128(%rsp) leaq 24(%rsp), %rax movq %rax, 136(%rsp) leaq 16(%rsp), %rax movq %rax, 144(%rsp) leaq 12(%rsp), %rax movq %rax, 152(%rsp) leaq 192(%rsp), %rax movq %rax, 160(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) movl $1, 72(%rsp) movl $1, 76(%rsp) movl $1, 80(%rsp) movl $1, 84(%rsp) leaq 56(%rsp), %rcx leaq 48(%rsp), %rdx leaq 76(%rsp), %rsi leaq 64(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L5 .L1: movq 168(%rsp), %rax subq %fs:40, %rax jne .L6 addq $184, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L5: .cfi_restore_state pushq 56(%rsp) .cfi_def_cfa_offset 200 pushq 56(%rsp) .cfi_def_cfa_offset 208 leaq 128(%rsp), %r9 movq 92(%rsp), %rcx movl 100(%rsp), %r8d movq 80(%rsp), %rsi movl 88(%rsp), %edx leaq _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 192 jmp .L1 .L6: call __stack_chk_fail@PLT .cfi_endproc .LFE2093: .size _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii, .-_ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii .section .text._Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,"axG",@progbits,_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,comdat .weak _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .type _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, @function _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii: .LFB2142: .cfi_startproc endbr64 subq $16, %rsp .cfi_def_cfa_offset 24 movl 24(%rsp), %eax pushq %rax .cfi_def_cfa_offset 32 call _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii addq $24, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2142: .size _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, .-_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .text .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2069: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2069: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .section .rodata.str1.1,"aMS",@progbits,1 .LC5: .string "calculation Time:%f ms\n" .text .globl _Z14MatrixMultGPU0PfiiS_ii .type _Z14MatrixMultGPU0PfiiS_ii, @function _Z14MatrixMultGPU0PfiiS_ii: .LFB2061: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $120, %rsp .cfi_def_cfa_offset 176 movq %rdi, 16(%rsp) movl %esi, %r13d movl %edx, 8(%rsp) movq %rcx, 24(%rsp) movl %r8d, 12(%rsp) movl %r9d, %r12d movq %fs:40, %rax movq %rax, 104(%rsp) xorl %eax, %eax leaq 64(%rsp), %rdi call cudaEventCreate@PLT leaq 72(%rsp), %rdi call cudaEventCreate@PLT movslq %r13d, %rbp movslq %r12d, %rbx movq %rbp, %r15 imulq %rbx, %r15 leaq 0(,%r15,4), %r14 leaq 40(%rsp), %rdi movq %r14, %rsi call cudaMalloc@PLT movslq 8(%rsp), %rax imulq %rax, %rbp salq $2, %rbp leaq 48(%rsp), %rdi movq %rbp, %rsi call cudaMalloc@PLT movslq 12(%rsp), %rax imulq %rax, %rbx salq $2, %rbx leaq 56(%rsp), %rdi movq %rbx, %rsi call cudaMalloc@PLT movl $1, %ecx movq %rbp, %rdx movq 16(%rsp), %rsi movq 48(%rsp), %rdi call cudaMemcpy@PLT movl $1, %ecx movq %rbx, %rdx movq 24(%rsp), %rsi movq 56(%rsp), %rdi call cudaMemcpy@PLT movl $0, %esi movq 64(%rsp), %rdi call cudaEventRecord@PLT movl $64, 80(%rsp) movl $64, 84(%rsp) movl $1, 88(%rsp) pxor %xmm0, %xmm0 cvtsi2ssl %r12d, %xmm0 mulss .LC0(%rip), %xmm0 movaps %xmm0, %xmm1 movss .LC6(%rip), %xmm3 movaps %xmm0, %xmm2 andps %xmm3, %xmm2 movss .LC1(%rip), %xmm4 ucomiss %xmm2, %xmm4 jbe .L12 cvttss2sil %xmm0, %eax pxor %xmm2, %xmm2 cvtsi2ssl %eax, %xmm2 cmpnless %xmm2, %xmm1 movss .LC3(%rip), %xmm4 andps %xmm4, %xmm1 addss %xmm2, %xmm1 andnps %xmm0, %xmm3 orps %xmm3, %xmm1 .L12: pxor %xmm0, %xmm0 cvtsi2ssl %r13d, %xmm0 mulss .LC0(%rip), %xmm0 movaps %xmm0, %xmm4 movss .LC6(%rip), %xmm3 movaps %xmm0, %xmm2 andps %xmm3, %xmm2 movss .LC1(%rip), %xmm5 ucomiss %xmm2, %xmm5 jbe .L13 cvttss2sil %xmm0, %eax pxor %xmm2, %xmm2 cvtsi2ssl %eax, %xmm2 cmpnless %xmm2, %xmm4 movss .LC3(%rip), %xmm5 andps %xmm5, %xmm4 addss %xmm2, %xmm4 andnps %xmm0, %xmm3 orps %xmm3, %xmm4 .L13: cvttss2sil %xmm1, %eax movl %eax, 92(%rsp) cvttss2sil %xmm4, %eax movl %eax, 96(%rsp) movl 88(%rsp), %ecx movl $0, %r9d movl $0, %r8d movq 80(%rsp), %rdx movq 92(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L20 .L14: movabsq $2305843009213693950, %rax cmpq %r15, %rax jb .L15 movq %r14, %rdi call _Znam@PLT movq %rax, %rbx movl $2, %ecx movq %r14, %rdx movq 40(%rsp), %rsi movq %rax, %rdi call cudaMemcpy@PLT movl $0, %esi movq 72(%rsp), %rdi call cudaEventRecord@PLT movq 72(%rsp), %rdi call cudaEventSynchronize@PLT movl $0x00000000, 36(%rsp) leaq 36(%rsp), %rdi movq 72(%rsp), %rdx movq 64(%rsp), %rsi call cudaEventElapsedTime@PLT pxor %xmm0, %xmm0 cvtss2sd 36(%rsp), %xmm0 leaq .LC5(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movq 40(%rsp), %rdi call cudaFree@PLT movq 48(%rsp), %rdi call cudaFree@PLT movq 56(%rsp), %rdi call cudaFree@PLT movq 104(%rsp), %rax subq %fs:40, %rax jne .L21 movq %rbx, %rax addq $120, %rsp .cfi_remember_state .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .L20: .cfi_restore_state subq $8, %rsp .cfi_def_cfa_offset 184 pushq %r12 .cfi_def_cfa_offset 192 movl 28(%rsp), %r9d movq 72(%rsp), %r8 movl 24(%rsp), %ecx movl %r13d, %edx movq 64(%rsp), %rsi movq 56(%rsp), %rdi call _ZL49__device_stub__Z11kMatrixMul1ILi64EEvPfS0_iiS0_iiPfS_iiS_ii addq $16, %rsp .cfi_def_cfa_offset 176 jmp .L14 .L15: movq 104(%rsp), %rax subq %fs:40, %rax je .L16 call __stack_chk_fail@PLT .L16: call __cxa_throw_bad_array_new_length@PLT .L21: call __stack_chk_fail@PLT .cfi_endproc .LFE2061: .size _Z14MatrixMultGPU0PfiiS_ii, .-_Z14MatrixMultGPU0PfiiS_ii .globl _Z12MatrixMulCPUPfiiS_ii .type _Z12MatrixMulCPUPfiiS_ii, @function _Z12MatrixMulCPUPfiiS_ii: .LFB2062: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $24, %rsp .cfi_def_cfa_offset 80 movl %edx, %ebp movq %rcx, 8(%rsp) movl %esi, %eax imull %r9d, %eax cltq movabsq $2305843009213693950, %rdx cmpq %rax, %rdx jb .L23 movq %rdi, %r13 movl %esi, %r12d movl %r8d, %r15d movl %r9d, %ebx leaq 0(,%rax,4), %rdi call _Znam@PLT movq %rax, %r14 testl %r12d, %r12d jle .L22 movslq %r15d, %rsi salq $2, %rsi movl $0, %edx movl $0, %eax movslq %ebp, %r15 movq %r14, %rdi movq %r15, %rcx jmp .L25 .L23: call __cxa_throw_bad_array_new_length@PLT .L29: movslq %edx, %r8 leaq 0(,%r8,4), %r15 leaq (%rdi,%r15), %r9 movq 8(%rsp), %r11 addq %r13, %r15 addq %rcx, %r8 leaq 0(%r13,%r8,4), %r8 movl $0, %r10d movl %eax, (%rsp) movl %edx, 4(%rsp) .L28: movq %r9, %r14 movl $0x00000000, (%r9) testl %ebp, %ebp jle .L26 movq %r11, %rdx movq %r15, %rax pxor %xmm1, %xmm1 .L27: movss (%rax), %xmm0 mulss (%rdx), %xmm0 addss %xmm0, %xmm1 addq $4, %rax addq %rsi, %rdx cmpq %r8, %rax jne .L27 movss %xmm1, (%r14) .L26: addl $1, %r10d addq $4, %r9 addq $4, %r11 cmpl %r10d, %ebx jne .L28 movl (%rsp), %eax movl 4(%rsp), %edx .L30: addl $1, %eax addl %r12d, %edx cmpl %eax, %r12d je .L33 .L25: testl %ebx, %ebx jg .L29 jmp .L30 .L33: movq %rdi, %r14 .L22: movq %r14, %rax addq $24, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2062: .size _Z12MatrixMulCPUPfiiS_ii, .-_Z12MatrixMulCPUPfiiS_ii .globl _Z13CompareMatrixPfS_ii .type _Z13CompareMatrixPfS_ii, @function _Z13CompareMatrixPfS_ii: .LFB2063: .cfi_startproc endbr64 testl %edx, %edx jle .L41 movslq %edx, %r10 movslq %ecx, %r8 salq $2, %r8 movl $0, %r9d pxor %xmm1, %xmm1 jmp .L37 .L38: movss (%rdi,%rax), %xmm0 subss (%rsi,%rax), %xmm0 addss %xmm0, %xmm1 addl $1, %edx addq %r8, %rax cmpl %edx, %ecx jne .L38 .L40: addq $1, %r9 cmpq %r10, %r9 je .L35 .L37: leaq 0(,%r9,4), %rax movl $0, %edx testl %ecx, %ecx jg .L38 jmp .L40 .L41: pxor %xmm1, %xmm1 .L35: movaps %xmm1, %xmm0 ret .cfi_endproc .LFE2063: .size _Z13CompareMatrixPfS_ii, .-_Z13CompareMatrixPfS_ii .globl _Z10FillMatrixPfiif .type _Z10FillMatrixPfiif, @function _Z10FillMatrixPfiif: .LFB2064: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $24, %rsp .cfi_def_cfa_offset 80 movl %edx, 12(%rsp) testl %esi, %esi jle .L44 movq %rdi, %r13 movl %esi, %r15d movd %xmm0, %ebp movl %edx, %r12d movl $0, 8(%rsp) movl $0, %r14d jmp .L46 .L58: call rand@PLT movl %eax, %edx movslq %ebx, %rax pxor %xmm0, %xmm0 cvtsi2ssl %edx, %xmm0 mulss .LC7(%rip), %xmm0 mulss .LC8(%rip), %xmm0 movss %xmm0, 0(%r13,%rax,4) .L49: addl $1, %ebx cmpl %r12d, %ebx je .L51 .L50: pxor %xmm1, %xmm1 movd %ebp, %xmm2 comiss %xmm2, %xmm1 ja .L58 movslq %ebx, %rax movl %ebp, 0(%r13,%rax,4) jmp .L49 .L51: addl $1, %r14d addl %r15d, %r12d addl %r15d, 8(%rsp) cmpl %r14d, %r15d je .L44 .L46: movl 8(%rsp), %ebx cmpl $0, 12(%rsp) jg .L50 jmp .L51 .L44: addq $24, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2064: .size _Z10FillMatrixPfiif, .-_Z10FillMatrixPfiif .section .rodata.str1.1 .LC9: .string "%f," .LC10: .string "\n" .text .globl _Z11PrintMatrixPfii .type _Z11PrintMatrixPfii, @function _Z11PrintMatrixPfii: .LFB2065: .cfi_startproc endbr64 pushq %r15 .cfi_def_cfa_offset 16 .cfi_offset 15, -16 pushq %r14 .cfi_def_cfa_offset 24 .cfi_offset 14, -24 pushq %r13 .cfi_def_cfa_offset 32 .cfi_offset 13, -32 pushq %r12 .cfi_def_cfa_offset 40 .cfi_offset 12, -40 pushq %rbp .cfi_def_cfa_offset 48 .cfi_offset 6, -48 pushq %rbx .cfi_def_cfa_offset 56 .cfi_offset 3, -56 subq $40, %rsp .cfi_def_cfa_offset 96 movq %rdi, 16(%rsp) movl %edx, 12(%rsp) testl %esi, %esi jle .L59 movl %esi, %r15d movl $0, %r14d movl $0, %r13d movl %edx, %eax cltq movq %rax, 24(%rsp) leaq .LC9(%rip), %r12 jmp .L61 .L63: movslq %r14d, %rax movq 16(%rsp), %rcx leaq (%rcx,%rax,4), %rbx movq 24(%rsp), %rdx addq %rdx, %rax leaq (%rcx,%rax,4), %rbp .L62: pxor %xmm0, %xmm0 cvtss2sd (%rbx), %xmm0 movq %r12, %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT addq $4, %rbx cmpq %rbp, %rbx jne .L62 .L64: leaq .LC10(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addl $1, %r13d addl %r15d, %r14d cmpl %r13d, %r15d je .L59 .L61: cmpl $0, 12(%rsp) jg .L63 jmp .L64 .L59: addq $40, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %rbp .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2065: .size _Z11PrintMatrixPfii, .-_Z11PrintMatrixPfii .globl main .type main, @function main: .LFB2066: .cfi_startproc endbr64 pushq %r12 .cfi_def_cfa_offset 16 .cfi_offset 12, -16 pushq %rbp .cfi_def_cfa_offset 24 .cfi_offset 6, -24 pushq %rbx .cfi_def_cfa_offset 32 .cfi_offset 3, -32 movl $4194304, %edi call _Znam@PLT movq %rax, %rbp movl $4194304, %edi call _Znam@PLT movq %rax, %rbx movss .LC3(%rip), %xmm0 movl $1024, %edx movl $1024, %esi movq %rbp, %rdi call _Z10FillMatrixPfiif movss .LC3(%rip), %xmm0 movl $1024, %edx movl $1024, %esi movq %rbx, %rdi call _Z10FillMatrixPfiif movl $1024, %r9d movl $1024, %r8d movq %rbx, %rcx movl $1024, %edx movl $1024, %esi movq %rbp, %rdi call _Z14MatrixMultGPU0PfiiS_ii movq %rax, %r12 movq %rbp, %rdi call _ZdaPv@PLT movq %rbx, %rdi call _ZdaPv@PLT testq %r12, %r12 je .L68 movl $4, %esi movq %r12, %rdi call _ZdlPvm@PLT .L68: movl $0, %eax popq %rbx .cfi_def_cfa_offset 24 popq %rbp .cfi_def_cfa_offset 16 popq %r12 .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2066: .size main, .-main .globl _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii .type _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii, @function _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii: .LFB2091: .cfi_startproc endbr64 subq $184, %rsp .cfi_def_cfa_offset 192 movq %rdi, 40(%rsp) movq %rsi, 32(%rsp) movl %edx, 28(%rsp) movl %ecx, 24(%rsp) movq %r8, 16(%rsp) movl %r9d, 12(%rsp) movq %fs:40, %rax movq %rax, 168(%rsp) xorl %eax, %eax leaq 40(%rsp), %rax movq %rax, 112(%rsp) leaq 32(%rsp), %rax movq %rax, 120(%rsp) leaq 28(%rsp), %rax movq %rax, 128(%rsp) leaq 24(%rsp), %rax movq %rax, 136(%rsp) leaq 16(%rsp), %rax movq %rax, 144(%rsp) leaq 12(%rsp), %rax movq %rax, 152(%rsp) leaq 192(%rsp), %rax movq %rax, 160(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) movl $1, 72(%rsp) movl $1, 76(%rsp) movl $1, 80(%rsp) movl $1, 84(%rsp) leaq 56(%rsp), %rcx leaq 48(%rsp), %rdx leaq 76(%rsp), %rsi leaq 64(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L74 .L70: movq 168(%rsp), %rax subq %fs:40, %rax jne .L75 addq $184, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L74: .cfi_restore_state pushq 56(%rsp) .cfi_def_cfa_offset 200 pushq 56(%rsp) .cfi_def_cfa_offset 208 leaq 128(%rsp), %r9 movq 92(%rsp), %rcx movl 100(%rsp), %r8d movq 80(%rsp), %rsi movl 88(%rsp), %edx leaq _Z11kMatrixMul0PfS_iiS_ii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 192 jmp .L70 .L75: call __stack_chk_fail@PLT .cfi_endproc .LFE2091: .size _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii, .-_Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii .globl _Z11kMatrixMul0PfS_iiS_ii .type _Z11kMatrixMul0PfS_iiS_ii, @function _Z11kMatrixMul0PfS_iiS_ii: .LFB2092: .cfi_startproc endbr64 subq $16, %rsp .cfi_def_cfa_offset 24 movl 24(%rsp), %eax pushq %rax .cfi_def_cfa_offset 32 call _Z39__device_stub__Z11kMatrixMul0PfS_iiS_iiPfS_iiS_ii addq $24, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2092: .size _Z11kMatrixMul0PfS_iiS_ii, .-_Z11kMatrixMul0PfS_iiS_ii .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC11: .string "_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii" .section .rodata.str1.1 .LC12: .string "_Z11kMatrixMul0PfS_iiS_ii" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2096: .cfi_startproc endbr64 pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset 3, -16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rbx movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC11(%rip), %rdx movq %rdx, %rcx leaq _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii(%rip), %rsi movq %rax, %rdi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC12(%rip), %rdx movq %rdx, %rcx leaq _Z11kMatrixMul0PfS_iiS_ii(%rip), %rsi movq %rbx, %rdi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT popq %rbx .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2096: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC0: .long 1015021568 .align 4 .LC1: .long 1258291200 .align 4 .LC3: .long 1065353216 .align 4 .LC6: .long 2147483647 .align 4 .LC7: .long 805306368 .align 4 .LC8: .long 1092616192 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "matrixmult.hip" .globl _Z26__device_stub__kMatrixMul0PfS_iiS_ii # -- Begin function _Z26__device_stub__kMatrixMul0PfS_iiS_ii .p2align 4, 0x90 .type _Z26__device_stub__kMatrixMul0PfS_iiS_ii,@function _Z26__device_stub__kMatrixMul0PfS_iiS_ii: # @_Z26__device_stub__kMatrixMul0PfS_iiS_ii .cfi_startproc # %bb.0: subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 88(%rsp) movq %rsi, 80(%rsp) movl %edx, 20(%rsp) movl %ecx, 16(%rsp) movq %r8, 72(%rsp) movl %r9d, 12(%rsp) leaq 88(%rsp), %rax movq %rax, 96(%rsp) leaq 80(%rsp), %rax movq %rax, 104(%rsp) leaq 20(%rsp), %rax movq %rax, 112(%rsp) leaq 16(%rsp), %rax movq %rax, 120(%rsp) leaq 72(%rsp), %rax movq %rax, 128(%rsp) leaq 12(%rsp), %rax movq %rax, 136(%rsp) leaq 160(%rsp), %rax movq %rax, 144(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 96(%rsp), %r9 movl $_Z11kMatrixMul0PfS_iiS_ii, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $168, %rsp .cfi_adjust_cfa_offset -168 retq .Lfunc_end0: .size _Z26__device_stub__kMatrixMul0PfS_iiS_ii, .Lfunc_end0-_Z26__device_stub__kMatrixMul0PfS_iiS_ii .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function _Z14MatrixMultGPU0PfiiS_ii .LCPI1_0: .long 0x3c800000 # float 0.015625 .text .globl _Z14MatrixMultGPU0PfiiS_ii .p2align 4, 0x90 .type _Z14MatrixMultGPU0PfiiS_ii,@function _Z14MatrixMultGPU0PfiiS_ii: # @_Z14MatrixMultGPU0PfiiS_ii .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $232, %rsp .cfi_def_cfa_offset 288 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %r9d, %r15d movl %r8d, %r13d movq %rcx, 80(%rsp) # 8-byte Spill movl %edx, %r12d movl %esi, %ebx movq %rdi, 72(%rsp) # 8-byte Spill leaq 64(%rsp), %rdi callq hipEventCreate leaq 8(%rsp), %rdi callq hipEventCreate movslq %ebx, %rax movq %rax, 88(%rsp) # 8-byte Spill leaq (,%rax,4), %rbp movslq %r15d, %r14 movq %rbp, %rsi imulq %r14, %rsi leaq 32(%rsp), %rdi movq %rsi, 96(%rsp) # 8-byte Spill callq hipMalloc movl %r12d, 40(%rsp) # 4-byte Spill movslq %r12d, %r12 imulq %rbp, %r12 leaq 24(%rsp), %rdi movq %r12, %rsi callq hipMalloc movl %r13d, 44(%rsp) # 4-byte Spill movslq %r13d, %rbp imulq %r14, %rbp shlq $2, %rbp leaq 16(%rsp), %rdi movq %rbp, %rsi callq hipMalloc movq 24(%rsp), %rdi movq 72(%rsp), %rsi # 8-byte Reload movq %r12, %rdx movl $1, %ecx callq hipMemcpy movq 16(%rsp), %rdi movq 80(%rsp), %rsi # 8-byte Reload movq %rbp, %rdx movl $1, %ecx callq hipMemcpy movq 64(%rsp), %rdi xorl %esi, %esi callq hipEventRecord cvtsi2ss %r14d, %xmm0 mulss .LCPI1_0(%rip), %xmm0 callq ceilf@PLT cvttss2si %xmm0, %r14d xorps %xmm0, %xmm0 cvtsi2ssl 88(%rsp), %xmm0 # 4-byte Folded Reload mulss .LCPI1_0(%rip), %xmm0 callq ceilf@PLT cvttss2si %xmm0, %edi shlq $32, %rdi orq %r14, %rdi movabsq $274877907008, %rdx # imm = 0x4000000040 movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_2 # %bb.1: movq 32(%rsp), %rax movq 24(%rsp), %rcx movq 16(%rsp), %rdx movq %rax, 168(%rsp) movq %rcx, 160(%rsp) movl %ebx, 60(%rsp) movl 40(%rsp), %eax # 4-byte Reload movl %eax, 56(%rsp) movq %rdx, 152(%rsp) movl 44(%rsp), %eax # 4-byte Reload movl %eax, 52(%rsp) movl %r15d, 48(%rsp) leaq 168(%rsp), %rax movq %rax, 176(%rsp) leaq 160(%rsp), %rax movq %rax, 184(%rsp) leaq 60(%rsp), %rax movq %rax, 192(%rsp) leaq 56(%rsp), %rax movq %rax, 200(%rsp) leaq 152(%rsp), %rax movq %rax, 208(%rsp) leaq 52(%rsp), %rax movq %rax, 216(%rsp) leaq 48(%rsp), %rax movq %rax, 224(%rsp) leaq 136(%rsp), %rdi leaq 120(%rsp), %rsi leaq 112(%rsp), %rdx leaq 104(%rsp), %rcx callq __hipPopCallConfiguration movq 136(%rsp), %rsi movl 144(%rsp), %edx movq 120(%rsp), %rcx movl 128(%rsp), %r8d leaq 176(%rsp), %r9 movl $_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, %edi pushq 104(%rsp) .cfi_adjust_cfa_offset 8 pushq 120(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_2: imull %ebx, %r15d movslq %r15d, %rax leaq (,%rax,4), %rcx testl %eax, %eax movq $-1, %rdi cmovnsq %rcx, %rdi callq _Znam movq %rax, %rbx movq 32(%rsp), %rsi movq %rax, %rdi movq 96(%rsp), %rdx # 8-byte Reload movl $2, %ecx callq hipMemcpy movq 8(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq 8(%rsp), %rdi callq hipEventSynchronize movl $0, 176(%rsp) movq 64(%rsp), %rsi movq 8(%rsp), %rdx leaq 176(%rsp), %rdi callq hipEventElapsedTime movss 176(%rsp), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str, %edi movb $1, %al callq printf movq 32(%rsp), %rdi callq hipFree movq 24(%rsp), %rdi callq hipFree movq 16(%rsp), %rdi callq hipFree movq %rbx, %rax addq $232, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size _Z14MatrixMultGPU0PfiiS_ii, .Lfunc_end1-_Z14MatrixMultGPU0PfiiS_ii .cfi_endproc # -- End function .section .text._Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii,"axG",@progbits,_Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii,comdat .weak _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii # -- Begin function _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .p2align 4, 0x90 .type _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii,@function _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii: # @_Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .cfi_startproc # %bb.0: subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 88(%rsp) movq %rsi, 80(%rsp) movl %edx, 20(%rsp) movl %ecx, 16(%rsp) movq %r8, 72(%rsp) movl %r9d, 12(%rsp) leaq 88(%rsp), %rax movq %rax, 96(%rsp) leaq 80(%rsp), %rax movq %rax, 104(%rsp) leaq 20(%rsp), %rax movq %rax, 112(%rsp) leaq 16(%rsp), %rax movq %rax, 120(%rsp) leaq 72(%rsp), %rax movq %rax, 128(%rsp) leaq 12(%rsp), %rax movq %rax, 136(%rsp) leaq 160(%rsp), %rax movq %rax, 144(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 96(%rsp), %r9 movl $_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $168, %rsp .cfi_adjust_cfa_offset -168 retq .Lfunc_end2: .size _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii, .Lfunc_end2-_Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .cfi_endproc # -- End function .text .globl _Z12MatrixMulCPUPfiiS_ii # -- Begin function _Z12MatrixMulCPUPfiiS_ii .p2align 4, 0x90 .type _Z12MatrixMulCPUPfiiS_ii,@function _Z12MatrixMulCPUPfiiS_ii: # @_Z12MatrixMulCPUPfiiS_ii .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 pushq %rax .cfi_def_cfa_offset 64 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %r9d, %ebx movl %r8d, %r12d movq %rcx, (%rsp) # 8-byte Spill movl %edx, %ebp movl %esi, %r13d movq %rdi, %r15 movl %r9d, %eax imull %esi, %eax cltq leaq (,%rax,4), %rcx testl %eax, %eax movq $-1, %rdi cmovnsq %rcx, %rdi callq _Znam testl %r13d, %r13d jle .LBB3_10 # %bb.1: # %.preheader.lr.ph movslq %r12d, %rcx movl %r13d, %edx movl %ebx, %esi movl %ebp, %edi leaq (,%rdx,4), %r8 shlq $2, %rcx xorl %r9d, %r9d jmp .LBB3_2 .p2align 4, 0x90 .LBB3_9: # %._crit_edge33 # in Loop: Header=BB3_2 Depth=1 incq %r9 addq %r8, %r15 cmpq %rdx, %r9 je .LBB3_10 .LBB3_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB3_4 Depth 2 # Child Loop BB3_6 Depth 3 testl %ebx, %ebx jle .LBB3_9 # %bb.3: # %.lr.ph32 # in Loop: Header=BB3_2 Depth=1 movq %r9, %r10 imulq %rdx, %r10 leaq (%rax,%r10,4), %r10 movq (%rsp), %r11 # 8-byte Reload xorl %r12d, %r12d jmp .LBB3_4 .p2align 4, 0x90 .LBB3_8: # in Loop: Header=BB3_4 Depth=2 incq %r12 addq $4, %r11 cmpq %rsi, %r12 je .LBB3_9 .LBB3_4: # Parent Loop BB3_2 Depth=1 # => This Loop Header: Depth=2 # Child Loop BB3_6 Depth 3 movl $0, (%r10,%r12,4) testl %ebp, %ebp jle .LBB3_8 # %bb.5: # %.lr.ph # in Loop: Header=BB3_4 Depth=2 movss (%r10,%r12,4), %xmm0 # xmm0 = mem[0],zero,zero,zero movq %r11, %r13 xorl %r14d, %r14d .p2align 4, 0x90 .LBB3_6: # Parent Loop BB3_2 Depth=1 # Parent Loop BB3_4 Depth=2 # => This Inner Loop Header: Depth=3 movss (%r15,%r14,4), %xmm1 # xmm1 = mem[0],zero,zero,zero mulss (%r13), %xmm1 addss %xmm1, %xmm0 incq %r14 addq %rcx, %r13 cmpq %r14, %rdi jne .LBB3_6 # %bb.7: # %._crit_edge # in Loop: Header=BB3_4 Depth=2 movss %xmm0, (%r10,%r12,4) jmp .LBB3_8 .LBB3_10: # %._crit_edge35 addq $8, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end3: .size _Z12MatrixMulCPUPfiiS_ii, .Lfunc_end3-_Z12MatrixMulCPUPfiiS_ii .cfi_endproc # -- End function .globl _Z13CompareMatrixPfS_ii # -- Begin function _Z13CompareMatrixPfS_ii .p2align 4, 0x90 .type _Z13CompareMatrixPfS_ii,@function _Z13CompareMatrixPfS_ii: # @_Z13CompareMatrixPfS_ii .cfi_startproc # %bb.0: testl %edx, %edx jle .LBB4_1 # %bb.3: # %.preheader.lr.ph movslq %ecx, %rax movl %edx, %edx movl %eax, %r8d shlq $2, %rax xorps %xmm0, %xmm0 xorl %r9d, %r9d jmp .LBB4_4 .p2align 4, 0x90 .LBB4_7: # %._crit_edge # in Loop: Header=BB4_4 Depth=1 incq %r9 addq $4, %rsi addq $4, %rdi cmpq %rdx, %r9 je .LBB4_2 .LBB4_4: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB4_6 Depth 2 testl %ecx, %ecx jle .LBB4_7 # %bb.5: # %.lr.ph.preheader # in Loop: Header=BB4_4 Depth=1 movq %r8, %r10 xorl %r11d, %r11d .p2align 4, 0x90 .LBB4_6: # %.lr.ph # Parent Loop BB4_4 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rdi,%r11), %xmm1 # xmm1 = mem[0],zero,zero,zero subss (%rsi,%r11), %xmm1 addss %xmm1, %xmm0 addq %rax, %r11 decq %r10 jne .LBB4_6 jmp .LBB4_7 .LBB4_1: xorps %xmm0, %xmm0 .LBB4_2: # %._crit_edge22 retq .Lfunc_end4: .size _Z13CompareMatrixPfS_ii, .Lfunc_end4-_Z13CompareMatrixPfS_ii .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function _Z10FillMatrixPfiif .LCPI5_0: .long 0x30000000 # float 4.65661287E-10 .LCPI5_1: .long 0x41200000 # float 10 .text .globl _Z10FillMatrixPfiif .p2align 4, 0x90 .type _Z10FillMatrixPfiif,@function _Z10FillMatrixPfiif: # @_Z10FillMatrixPfiif .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $24, %rsp .cfi_def_cfa_offset 80 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %edx, (%rsp) # 4-byte Spill movq %rdi, 8(%rsp) # 8-byte Spill testl %esi, %esi jle .LBB5_8 # %bb.1: # %.preheader.lr.ph movl %esi, %ebp movl %esi, %eax movq %rax, 16(%rsp) # 8-byte Spill movl (%rsp), %r12d # 4-byte Reload xorl %r13d, %r13d xorps %xmm2, %xmm2 xorl %r14d, %r14d movss %xmm0, 4(%rsp) # 4-byte Spill jmp .LBB5_2 .p2align 4, 0x90 .LBB5_7: # %._crit_edge # in Loop: Header=BB5_2 Depth=1 incq %r14 addl %ebp, %r13d cmpq 16(%rsp), %r14 # 8-byte Folded Reload je .LBB5_8 .LBB5_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB5_4 Depth 2 cmpl $0, (%rsp) # 4-byte Folded Reload jle .LBB5_7 # %bb.3: # %.lr.ph # in Loop: Header=BB5_2 Depth=1 movl %r13d, %eax movq 8(%rsp), %rcx # 8-byte Reload leaq (%rcx,%rax,4), %rbx xorl %r15d, %r15d jmp .LBB5_4 .p2align 4, 0x90 .LBB5_6: # in Loop: Header=BB5_4 Depth=2 movss %xmm1, (%rbx,%r15,4) incq %r15 cmpq %r15, %r12 je .LBB5_7 .LBB5_4: # Parent Loop BB5_2 Depth=1 # => This Inner Loop Header: Depth=2 ucomiss %xmm0, %xmm2 movaps %xmm0, %xmm1 jbe .LBB5_6 # %bb.5: # in Loop: Header=BB5_4 Depth=2 callq rand xorps %xmm2, %xmm2 movss 4(%rsp), %xmm0 # 4-byte Reload # xmm0 = mem[0],zero,zero,zero xorps %xmm1, %xmm1 cvtsi2ss %eax, %xmm1 mulss .LCPI5_0(%rip), %xmm1 mulss .LCPI5_1(%rip), %xmm1 jmp .LBB5_6 .LBB5_8: # %._crit_edge19 addq $24, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end5: .size _Z10FillMatrixPfiif, .Lfunc_end5-_Z10FillMatrixPfiif .cfi_endproc # -- End function .globl _Z11PrintMatrixPfii # -- Begin function _Z11PrintMatrixPfii .p2align 4, 0x90 .type _Z11PrintMatrixPfii,@function _Z11PrintMatrixPfii: # @_Z11PrintMatrixPfii .cfi_startproc # %bb.0: pushq %rbp .cfi_def_cfa_offset 16 pushq %r15 .cfi_def_cfa_offset 24 pushq %r14 .cfi_def_cfa_offset 32 pushq %r13 .cfi_def_cfa_offset 40 pushq %r12 .cfi_def_cfa_offset 48 pushq %rbx .cfi_def_cfa_offset 56 subq $24, %rsp .cfi_def_cfa_offset 80 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %edx, 4(%rsp) # 4-byte Spill movq %rdi, 8(%rsp) # 8-byte Spill testl %esi, %esi jle .LBB6_6 # %bb.1: # %.preheader.lr.ph movl %esi, %ebp movl %esi, %eax movq %rax, 16(%rsp) # 8-byte Spill movl 4(%rsp), %r12d # 4-byte Reload xorl %r13d, %r13d xorl %r14d, %r14d jmp .LBB6_2 .p2align 4, 0x90 .LBB6_5: # %._crit_edge # in Loop: Header=BB6_2 Depth=1 movl $10, %edi callq putchar@PLT incq %r14 addl %ebp, %r13d cmpq 16(%rsp), %r14 # 8-byte Folded Reload je .LBB6_6 .LBB6_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB6_4 Depth 2 cmpl $0, 4(%rsp) # 4-byte Folded Reload jle .LBB6_5 # %bb.3: # %.lr.ph # in Loop: Header=BB6_2 Depth=1 movl %r13d, %eax movq 8(%rsp), %rcx # 8-byte Reload leaq (%rcx,%rax,4), %rbx xorl %r15d, %r15d .p2align 4, 0x90 .LBB6_4: # Parent Loop BB6_2 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rbx,%r15,4), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str.1, %edi movb $1, %al callq printf incq %r15 cmpq %r15, %r12 jne .LBB6_4 jmp .LBB6_5 .LBB6_6: # %._crit_edge13 addq $24, %rsp .cfi_def_cfa_offset 56 popq %rbx .cfi_def_cfa_offset 48 popq %r12 .cfi_def_cfa_offset 40 popq %r13 .cfi_def_cfa_offset 32 popq %r14 .cfi_def_cfa_offset 24 popq %r15 .cfi_def_cfa_offset 16 popq %rbp .cfi_def_cfa_offset 8 retq .Lfunc_end6: .size _Z11PrintMatrixPfii, .Lfunc_end6-_Z11PrintMatrixPfii .cfi_endproc # -- End function .globl main # -- Begin function main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %r15 .cfi_def_cfa_offset 16 pushq %r14 .cfi_def_cfa_offset 24 pushq %rbx .cfi_def_cfa_offset 32 .cfi_offset %rbx, -32 .cfi_offset %r14, -24 .cfi_offset %r15, -16 movl $4194304, %edi # imm = 0x400000 callq _Znam movq %rax, %rbx movl $4194304, %edi # imm = 0x400000 callq _Znam movq %rax, %r14 xorl %eax, %eax movq %rbx, %rcx .p2align 4, 0x90 .LBB7_1: # %.preheader.i # =>This Loop Header: Depth=1 # Child Loop BB7_2 Depth 2 xorl %edx, %edx .p2align 4, 0x90 .LBB7_2: # Parent Loop BB7_1 Depth=1 # => This Inner Loop Header: Depth=2 movl $1065353216, (%rcx,%rdx,4) # imm = 0x3F800000 incq %rdx cmpq $1024, %rdx # imm = 0x400 jne .LBB7_2 # %bb.3: # %._crit_edge.i # in Loop: Header=BB7_1 Depth=1 incq %rax addq $4096, %rcx # imm = 0x1000 cmpq $1024, %rax # imm = 0x400 jne .LBB7_1 # %bb.4: # %.preheader.i19.preheader xorl %eax, %eax movq %r14, %rcx .p2align 4, 0x90 .LBB7_5: # %.preheader.i19 # =>This Loop Header: Depth=1 # Child Loop BB7_6 Depth 2 xorl %edx, %edx .p2align 4, 0x90 .LBB7_6: # Parent Loop BB7_5 Depth=1 # => This Inner Loop Header: Depth=2 movl $1065353216, (%rcx,%rdx,4) # imm = 0x3F800000 incq %rdx cmpq $1024, %rdx # imm = 0x400 jne .LBB7_6 # %bb.7: # %._crit_edge.i24 # in Loop: Header=BB7_5 Depth=1 incq %rax addq $4096, %rcx # imm = 0x1000 cmpq $1024, %rax # imm = 0x400 jne .LBB7_5 # %bb.8: # %_Z10FillMatrixPfiif.exit27 movq %rbx, %rdi movl $1024, %esi # imm = 0x400 movl $1024, %edx # imm = 0x400 movq %r14, %rcx movl $1024, %r8d # imm = 0x400 movl $1024, %r9d # imm = 0x400 callq _Z14MatrixMultGPU0PfiiS_ii movq %rax, %r15 movq %rbx, %rdi callq _ZdaPv movq %r14, %rdi callq _ZdaPv movq %r15, %rdi callq _ZdlPv xorl %eax, %eax popq %rbx .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 retq .Lfunc_end7: .size main, .Lfunc_end7-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 subq $32, %rsp .cfi_def_cfa_offset 48 .cfi_offset %rbx, -16 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB8_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB8_2: movq __hip_gpubin_handle(%rip), %rbx xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z11kMatrixMul0PfS_iiS_ii, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, %esi movl $.L__unnamed_2, %edx movl $.L__unnamed_2, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $32, %rsp .cfi_def_cfa_offset 16 popq %rbx .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end8: .size __hip_module_ctor, .Lfunc_end8-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB9_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB9_2: retq .Lfunc_end9: .size __hip_module_dtor, .Lfunc_end9-__hip_module_dtor .cfi_endproc # -- End function .type _Z11kMatrixMul0PfS_iiS_ii,@object # @_Z11kMatrixMul0PfS_iiS_ii .section .rodata,"a",@progbits .globl _Z11kMatrixMul0PfS_iiS_ii .p2align 3, 0x0 _Z11kMatrixMul0PfS_iiS_ii: .quad _Z26__device_stub__kMatrixMul0PfS_iiS_ii .size _Z11kMatrixMul0PfS_iiS_ii, 8 .type _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,@object # @_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .section .rodata._Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,"aG",@progbits,_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii,comdat .weak _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .p2align 3, 0x0 _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii: .quad _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .size _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "calculation Time:%f ms\n" .size .L.str, 24 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz "%f," .size .L.str.1, 4 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z11kMatrixMul0PfS_iiS_ii" .size .L__unnamed_1, 26 .type .L__unnamed_2,@object # @1 .L__unnamed_2: .asciz "_Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii" .size .L__unnamed_2, 36 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z26__device_stub__kMatrixMul0PfS_iiS_ii .addrsig_sym _Z26__device_stub__kMatrixMul1ILi64EEvPfS0_iiS0_ii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z11kMatrixMul0PfS_iiS_ii .addrsig_sym _Z11kMatrixMul1ILi64EEvPfS0_iiS0_ii .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
#include <stdio.h> __global__ void square( int * d_in){ int totalSum; if (threadIdx.x == 0) totalSum = 0; __syncthreads(); int localVal = d_in[threadIdx.x]; atomicAdd(&totalSum, 1); __syncthreads(); } int main(int argc, char ** argv) { const int ARRAY_SIZE = 64; const int ARRAY_BYTES = ARRAY_SIZE * sizeof(float); // generate the input array on the host int h_in[ARRAY_SIZE]; for (int i = 0; i < ARRAY_SIZE; i++) { h_in[i] = i; } int * d_in; cudaMalloc((void**) &d_in, ARRAY_BYTES); // cudaMalloc((void*) &totalSum, sizeof(float)); cudaMemcpy(d_in, h_in, ARRAY_BYTES, cudaMemcpyHostToDevice); square<<<1, ARRAY_SIZE>>>(d_in); // cudaMemcpy(ans, totalSum, sizeof(float), cudaMemcpyDeviceToHost); // printf("%f\n",ans); cudaFree(d_in); return 0; }
code for sm_80 Function : _Z6squarePi .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fec0000010000 */ /*0020*/ BPT.TRAP 0x1 ; /* 0x000000040000795c */ /* 0x000fea0000300000 */ /*0030*/ BRA 0x30; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0040*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0050*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0060*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0070*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0080*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0090*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include <stdio.h> __global__ void square( int * d_in){ int totalSum; if (threadIdx.x == 0) totalSum = 0; __syncthreads(); int localVal = d_in[threadIdx.x]; atomicAdd(&totalSum, 1); __syncthreads(); } int main(int argc, char ** argv) { const int ARRAY_SIZE = 64; const int ARRAY_BYTES = ARRAY_SIZE * sizeof(float); // generate the input array on the host int h_in[ARRAY_SIZE]; for (int i = 0; i < ARRAY_SIZE; i++) { h_in[i] = i; } int * d_in; cudaMalloc((void**) &d_in, ARRAY_BYTES); // cudaMalloc((void*) &totalSum, sizeof(float)); cudaMemcpy(d_in, h_in, ARRAY_BYTES, cudaMemcpyHostToDevice); square<<<1, ARRAY_SIZE>>>(d_in); // cudaMemcpy(ans, totalSum, sizeof(float), cudaMemcpyDeviceToHost); // printf("%f\n",ans); cudaFree(d_in); return 0; }
.file "tmpxft_0011fdb2_00000000-6_atomics.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2060: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2060: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z25__device_stub__Z6squarePiPi .type _Z25__device_stub__Z6squarePiPi, @function _Z25__device_stub__Z6squarePiPi: .LFB2082: .cfi_startproc endbr64 subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 8(%rsp) movq %fs:40, %rax movq %rax, 88(%rsp) xorl %eax, %eax leaq 8(%rsp), %rax movq %rax, 80(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) leaq 24(%rsp), %rcx leaq 16(%rsp), %rdx leaq 44(%rsp), %rsi leaq 32(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 88(%rsp), %rax subq %fs:40, %rax jne .L8 addq $104, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 24(%rsp) .cfi_def_cfa_offset 120 pushq 24(%rsp) .cfi_def_cfa_offset 128 leaq 96(%rsp), %r9 movq 60(%rsp), %rcx movl 68(%rsp), %r8d movq 48(%rsp), %rsi movl 56(%rsp), %edx leaq _Z6squarePi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 112 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2082: .size _Z25__device_stub__Z6squarePiPi, .-_Z25__device_stub__Z6squarePiPi .globl _Z6squarePi .type _Z6squarePi, @function _Z6squarePi: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z25__device_stub__Z6squarePiPi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z6squarePi, .-_Z6squarePi .globl main .type main, @function main: .LFB2057: .cfi_startproc endbr64 subq $312, %rsp .cfi_def_cfa_offset 320 movq %fs:40, %rax movq %rax, 296(%rsp) xorl %eax, %eax .L12: movl %eax, 32(%rsp,%rax,4) addq $1, %rax cmpq $64, %rax jne .L12 movq %rsp, %rdi movl $256, %esi call cudaMalloc@PLT leaq 32(%rsp), %rsi movl $1, %ecx movl $256, %edx movq (%rsp), %rdi call cudaMemcpy@PLT movl $64, 20(%rsp) movl $1, 24(%rsp) movl $1, 8(%rsp) movl $1, 12(%rsp) movl $0, %r9d movl $0, %r8d movq 20(%rsp), %rdx movl $1, %ecx movq 8(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L17 .L13: movq (%rsp), %rdi call cudaFree@PLT movq 296(%rsp), %rax subq %fs:40, %rax jne .L18 movl $0, %eax addq $312, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L17: .cfi_restore_state movq (%rsp), %rdi call _Z25__device_stub__Z6squarePiPi jmp .L13 .L18: call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z6squarePi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2085: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z6squarePi(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2085: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include <stdio.h> __global__ void square( int * d_in){ int totalSum; if (threadIdx.x == 0) totalSum = 0; __syncthreads(); int localVal = d_in[threadIdx.x]; atomicAdd(&totalSum, 1); __syncthreads(); } int main(int argc, char ** argv) { const int ARRAY_SIZE = 64; const int ARRAY_BYTES = ARRAY_SIZE * sizeof(float); // generate the input array on the host int h_in[ARRAY_SIZE]; for (int i = 0; i < ARRAY_SIZE; i++) { h_in[i] = i; } int * d_in; cudaMalloc((void**) &d_in, ARRAY_BYTES); // cudaMalloc((void*) &totalSum, sizeof(float)); cudaMemcpy(d_in, h_in, ARRAY_BYTES, cudaMemcpyHostToDevice); square<<<1, ARRAY_SIZE>>>(d_in); // cudaMemcpy(ans, totalSum, sizeof(float), cudaMemcpyDeviceToHost); // printf("%f\n",ans); cudaFree(d_in); return 0; }
#include <hip/hip_runtime.h> #include <stdio.h> __global__ void square( int * d_in){ int totalSum; if (threadIdx.x == 0) totalSum = 0; __syncthreads(); int localVal = d_in[threadIdx.x]; atomicAdd(&totalSum, 1); __syncthreads(); } int main(int argc, char ** argv) { const int ARRAY_SIZE = 64; const int ARRAY_BYTES = ARRAY_SIZE * sizeof(float); // generate the input array on the host int h_in[ARRAY_SIZE]; for (int i = 0; i < ARRAY_SIZE; i++) { h_in[i] = i; } int * d_in; hipMalloc((void**) &d_in, ARRAY_BYTES); // cudaMalloc((void*) &totalSum, sizeof(float)); hipMemcpy(d_in, h_in, ARRAY_BYTES, hipMemcpyHostToDevice); square<<<1, ARRAY_SIZE>>>(d_in); // cudaMemcpy(ans, totalSum, sizeof(float), cudaMemcpyDeviceToHost); // printf("%f\n",ans); hipFree(d_in); return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> #include <stdio.h> __global__ void square( int * d_in){ int totalSum; if (threadIdx.x == 0) totalSum = 0; __syncthreads(); int localVal = d_in[threadIdx.x]; atomicAdd(&totalSum, 1); __syncthreads(); } int main(int argc, char ** argv) { const int ARRAY_SIZE = 64; const int ARRAY_BYTES = ARRAY_SIZE * sizeof(float); // generate the input array on the host int h_in[ARRAY_SIZE]; for (int i = 0; i < ARRAY_SIZE; i++) { h_in[i] = i; } int * d_in; hipMalloc((void**) &d_in, ARRAY_BYTES); // cudaMalloc((void*) &totalSum, sizeof(float)); hipMemcpy(d_in, h_in, ARRAY_BYTES, hipMemcpyHostToDevice); square<<<1, ARRAY_SIZE>>>(d_in); // cudaMemcpy(ans, totalSum, sizeof(float), cudaMemcpyDeviceToHost); // printf("%f\n",ans); hipFree(d_in); return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z6squarePi .globl _Z6squarePi .p2align 8 .type _Z6squarePi,@function _Z6squarePi: s_load_b64 s[0:1], s[0:1], 0x4 v_and_b32_e32 v1, 0x3ff, v0 v_bfe_u32 v3, v0, 10, 10 v_bfe_u32 v0, v0, 20, 10 s_waitcnt lgkmcnt(0) s_lshr_b32 s0, s0, 16 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_mul_i32 s0, s0, s1 v_mul_lo_u32 v2, s0, v1 s_mov_b32 s0, exec_lo s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u32_u24 v2, v3, s1, v2 v_add_lshl_u32 v0, v2, v0, 2 v_cmpx_eq_u32_e32 0, v1 s_cbranch_execz .LBB0_2 v_mov_b32_e32 v1, 0 ds_store_b32 v0, v1 .LBB0_2: s_or_b32 exec_lo, exec_lo, s0 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv ds_load_b32 v1, v0 s_waitcnt lgkmcnt(0) v_add_nc_u32_e32 v1, 1, v1 ds_store_b32 v0, v1 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z6squarePi .amdhsa_group_segment_fixed_size 4096 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 8 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 1 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 2 .amdhsa_next_free_vgpr 4 .amdhsa_next_free_sgpr 2 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z6squarePi, .Lfunc_end0-_Z6squarePi .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer .group_segment_fixed_size: 4096 .kernarg_segment_align: 8 .kernarg_segment_size: 8 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z6squarePi .private_segment_fixed_size: 0 .sgpr_count: 2 .sgpr_spill_count: 0 .symbol: _Z6squarePi.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 4 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <hip/hip_runtime.h> #include <stdio.h> __global__ void square( int * d_in){ int totalSum; if (threadIdx.x == 0) totalSum = 0; __syncthreads(); int localVal = d_in[threadIdx.x]; atomicAdd(&totalSum, 1); __syncthreads(); } int main(int argc, char ** argv) { const int ARRAY_SIZE = 64; const int ARRAY_BYTES = ARRAY_SIZE * sizeof(float); // generate the input array on the host int h_in[ARRAY_SIZE]; for (int i = 0; i < ARRAY_SIZE; i++) { h_in[i] = i; } int * d_in; hipMalloc((void**) &d_in, ARRAY_BYTES); // cudaMalloc((void*) &totalSum, sizeof(float)); hipMemcpy(d_in, h_in, ARRAY_BYTES, hipMemcpyHostToDevice); square<<<1, ARRAY_SIZE>>>(d_in); // cudaMemcpy(ans, totalSum, sizeof(float), cudaMemcpyDeviceToHost); // printf("%f\n",ans); hipFree(d_in); return 0; }
.text .file "atomics.hip" .globl _Z21__device_stub__squarePi # -- Begin function _Z21__device_stub__squarePi .p2align 4, 0x90 .type _Z21__device_stub__squarePi,@function _Z21__device_stub__squarePi: # @_Z21__device_stub__squarePi .cfi_startproc # %bb.0: subq $72, %rsp .cfi_def_cfa_offset 80 movq %rdi, 64(%rsp) leaq 64(%rsp), %rax movq %rax, (%rsp) leaq 48(%rsp), %rdi leaq 32(%rsp), %rsi leaq 24(%rsp), %rdx leaq 16(%rsp), %rcx callq __hipPopCallConfiguration movq 48(%rsp), %rsi movl 56(%rsp), %edx movq 32(%rsp), %rcx movl 40(%rsp), %r8d movq %rsp, %r9 movl $_Z6squarePi, %edi pushq 16(%rsp) .cfi_adjust_cfa_offset 8 pushq 32(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $88, %rsp .cfi_adjust_cfa_offset -88 retq .Lfunc_end0: .size _Z21__device_stub__squarePi, .Lfunc_end0-_Z21__device_stub__squarePi .cfi_endproc # -- End function .globl main # -- Begin function main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: subq $344, %rsp # imm = 0x158 .cfi_def_cfa_offset 352 xorl %eax, %eax .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 movl %eax, 80(%rsp,%rax,4) incq %rax cmpq $64, %rax jne .LBB1_1 # %bb.2: leaq 8(%rsp), %rdi movl $256, %esi # imm = 0x100 callq hipMalloc movq 8(%rsp), %rdi leaq 80(%rsp), %rsi movl $256, %edx # imm = 0x100 movl $1, %ecx callq hipMemcpy movabsq $4294967297, %rdi # imm = 0x100000001 leaq 63(%rdi), %rdx movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_4 # %bb.3: movq 8(%rsp), %rax movq %rax, 72(%rsp) leaq 72(%rsp), %rax movq %rax, 16(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 16(%rsp), %r9 movl $_Z6squarePi, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_4: movq 8(%rsp), %rdi callq hipFree xorl %eax, %eax addq $344, %rsp # imm = 0x158 .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z6squarePi, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z6squarePi,@object # @_Z6squarePi .section .rodata,"a",@progbits .globl _Z6squarePi .p2align 3, 0x0 _Z6squarePi: .quad _Z21__device_stub__squarePi .size _Z6squarePi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z6squarePi" .size .L__unnamed_1, 12 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z21__device_stub__squarePi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z6squarePi .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z6squarePi .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fec0000010000 */ /*0020*/ BPT.TRAP 0x1 ; /* 0x000000040000795c */ /* 0x000fea0000300000 */ /*0030*/ BRA 0x30; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0040*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0050*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0060*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0070*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0080*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0090*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z6squarePi .globl _Z6squarePi .p2align 8 .type _Z6squarePi,@function _Z6squarePi: s_load_b64 s[0:1], s[0:1], 0x4 v_and_b32_e32 v1, 0x3ff, v0 v_bfe_u32 v3, v0, 10, 10 v_bfe_u32 v0, v0, 20, 10 s_waitcnt lgkmcnt(0) s_lshr_b32 s0, s0, 16 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_mul_i32 s0, s0, s1 v_mul_lo_u32 v2, s0, v1 s_mov_b32 s0, exec_lo s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u32_u24 v2, v3, s1, v2 v_add_lshl_u32 v0, v2, v0, 2 v_cmpx_eq_u32_e32 0, v1 s_cbranch_execz .LBB0_2 v_mov_b32_e32 v1, 0 ds_store_b32 v0, v1 .LBB0_2: s_or_b32 exec_lo, exec_lo, s0 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv ds_load_b32 v1, v0 s_waitcnt lgkmcnt(0) v_add_nc_u32_e32 v1, 1, v1 ds_store_b32 v0, v1 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z6squarePi .amdhsa_group_segment_fixed_size 4096 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 8 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 1 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 2 .amdhsa_next_free_vgpr 4 .amdhsa_next_free_sgpr 2 .amdhsa_reserve_vcc 0 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z6squarePi, .Lfunc_end0-_Z6squarePi .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer .group_segment_fixed_size: 4096 .kernarg_segment_align: 8 .kernarg_segment_size: 8 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z6squarePi .private_segment_fixed_size: 0 .sgpr_count: 2 .sgpr_spill_count: 0 .symbol: _Z6squarePi.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 4 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_0011fdb2_00000000-6_atomics.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2060: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2060: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z25__device_stub__Z6squarePiPi .type _Z25__device_stub__Z6squarePiPi, @function _Z25__device_stub__Z6squarePiPi: .LFB2082: .cfi_startproc endbr64 subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 8(%rsp) movq %fs:40, %rax movq %rax, 88(%rsp) xorl %eax, %eax leaq 8(%rsp), %rax movq %rax, 80(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) leaq 24(%rsp), %rcx leaq 16(%rsp), %rdx leaq 44(%rsp), %rsi leaq 32(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 88(%rsp), %rax subq %fs:40, %rax jne .L8 addq $104, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 24(%rsp) .cfi_def_cfa_offset 120 pushq 24(%rsp) .cfi_def_cfa_offset 128 leaq 96(%rsp), %r9 movq 60(%rsp), %rcx movl 68(%rsp), %r8d movq 48(%rsp), %rsi movl 56(%rsp), %edx leaq _Z6squarePi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 112 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2082: .size _Z25__device_stub__Z6squarePiPi, .-_Z25__device_stub__Z6squarePiPi .globl _Z6squarePi .type _Z6squarePi, @function _Z6squarePi: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z25__device_stub__Z6squarePiPi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z6squarePi, .-_Z6squarePi .globl main .type main, @function main: .LFB2057: .cfi_startproc endbr64 subq $312, %rsp .cfi_def_cfa_offset 320 movq %fs:40, %rax movq %rax, 296(%rsp) xorl %eax, %eax .L12: movl %eax, 32(%rsp,%rax,4) addq $1, %rax cmpq $64, %rax jne .L12 movq %rsp, %rdi movl $256, %esi call cudaMalloc@PLT leaq 32(%rsp), %rsi movl $1, %ecx movl $256, %edx movq (%rsp), %rdi call cudaMemcpy@PLT movl $64, 20(%rsp) movl $1, 24(%rsp) movl $1, 8(%rsp) movl $1, 12(%rsp) movl $0, %r9d movl $0, %r8d movq 20(%rsp), %rdx movl $1, %ecx movq 8(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L17 .L13: movq (%rsp), %rdi call cudaFree@PLT movq 296(%rsp), %rax subq %fs:40, %rax jne .L18 movl $0, %eax addq $312, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L17: .cfi_restore_state movq (%rsp), %rdi call _Z25__device_stub__Z6squarePiPi jmp .L13 .L18: call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z6squarePi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2085: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z6squarePi(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2085: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "atomics.hip" .globl _Z21__device_stub__squarePi # -- Begin function _Z21__device_stub__squarePi .p2align 4, 0x90 .type _Z21__device_stub__squarePi,@function _Z21__device_stub__squarePi: # @_Z21__device_stub__squarePi .cfi_startproc # %bb.0: subq $72, %rsp .cfi_def_cfa_offset 80 movq %rdi, 64(%rsp) leaq 64(%rsp), %rax movq %rax, (%rsp) leaq 48(%rsp), %rdi leaq 32(%rsp), %rsi leaq 24(%rsp), %rdx leaq 16(%rsp), %rcx callq __hipPopCallConfiguration movq 48(%rsp), %rsi movl 56(%rsp), %edx movq 32(%rsp), %rcx movl 40(%rsp), %r8d movq %rsp, %r9 movl $_Z6squarePi, %edi pushq 16(%rsp) .cfi_adjust_cfa_offset 8 pushq 32(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $88, %rsp .cfi_adjust_cfa_offset -88 retq .Lfunc_end0: .size _Z21__device_stub__squarePi, .Lfunc_end0-_Z21__device_stub__squarePi .cfi_endproc # -- End function .globl main # -- Begin function main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: subq $344, %rsp # imm = 0x158 .cfi_def_cfa_offset 352 xorl %eax, %eax .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 movl %eax, 80(%rsp,%rax,4) incq %rax cmpq $64, %rax jne .LBB1_1 # %bb.2: leaq 8(%rsp), %rdi movl $256, %esi # imm = 0x100 callq hipMalloc movq 8(%rsp), %rdi leaq 80(%rsp), %rsi movl $256, %edx # imm = 0x100 movl $1, %ecx callq hipMemcpy movabsq $4294967297, %rdi # imm = 0x100000001 leaq 63(%rdi), %rdx movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_4 # %bb.3: movq 8(%rsp), %rax movq %rax, 72(%rsp) leaq 72(%rsp), %rax movq %rax, 16(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 16(%rsp), %r9 movl $_Z6squarePi, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_4: movq 8(%rsp), %rdi callq hipFree xorl %eax, %eax addq $344, %rsp # imm = 0x158 .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z6squarePi, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z6squarePi,@object # @_Z6squarePi .section .rodata,"a",@progbits .globl _Z6squarePi .p2align 3, 0x0 _Z6squarePi: .quad _Z21__device_stub__squarePi .size _Z6squarePi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z6squarePi" .size .L__unnamed_1, 12 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z21__device_stub__squarePi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z6squarePi .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
// A shared memory matrix multiplication program #include "stdio.h" #include "stdlib.h" // Keep the SIZE evenly divisible by TILE_WIDTH #define SIZE 512 #define TILE_WIDTH 16 // kernels that are called by another kernel use the __device__ identifier __device__ float * GetSubMatrix(float * large_matrix, int row, int col) { // this returns the address of the first element in a sub-matrix // when using this sub-matrix you have to jump ahead SIZE elements per row float * subMatrix = &large_matrix[ row * SIZE * TILE_WIDTH + col * TILE_WIDTH ]; return subMatrix; } // matrix multiplication kernel definition __global__ void MatrixMulKernel(float * A,float * B,float * C) { int blockRow = blockIdx.y; int blockCol = blockIdx.x; // Each thread block computes one sub-matrix Csub of C // Here we call another kernel from within a kernel. // Note that we do not need to pass any grid or block // information between the <<< >>> symbols. float * Csub = GetSubMatrix(C, blockRow, blockCol); // Each thread computes on element of Csub // by accumulating results into Csub float Cvalue = 0.0; // Thread row and column within Csub int row = threadIdx.y; int col = threadIdx.x; // Iterate through n sub_matrices, where n depends on the size of the tile and matrix int sub_matrices_per_row = gridDim.x; int i; for ( i = 0; i < sub_matrices_per_row; i++ ) { // Get sub-matrices // Block threads will work on these sub-matrices float * Asub = GetSubMatrix(A, blockRow, i); float * Bsub = GetSubMatrix(B, i, blockCol); // Shared memory used to store Asub and Bsub __shared__ float As[TILE_WIDTH][TILE_WIDTH]; __shared__ float Bs[TILE_WIDTH][TILE_WIDTH]; // Load Asub and Bsub from device memory into shared memory // Each thread loads one element of each sub-matrix As[row][col] = Asub[ row*SIZE + col ]; Bs[row][col] = Bsub[ row*SIZE + col ]; // Sync threads to make sure sub-matrices are completely loaded into shared memory // Remember, this only syncs threads within a block __syncthreads(); int j; // Multiply Asub and Bsub together, using fast shared memory for data access for ( j = 0; j < TILE_WIDTH ; j++ ) Cvalue += As[row][j] * Bs[j][col]; // Sync threads to ensure the preceding computation is done before loading // two new sub-matrices of A and B in the next iteration __syncthreads(); } Csub[ row*SIZE + col ] = Cvalue; } int main(int argc, char ** argv) { float h_A[SIZE*SIZE],h_B[SIZE*SIZE],h_C[SIZE*SIZE]; float * d_A, * d_B, * d_C; // initialize host matrices with arbitrary data int i; for (i=0;i<SIZE*SIZE;i++) { h_A[i] = (float)i; h_B[i] = (float)SIZE * (float)SIZE - (float)i - 1.00; h_C[i] = 0.0; } // allocate space on device size_t size = SIZE*SIZE*sizeof(float); cudaMalloc(&d_A,size); cudaMalloc(&d_B,size); cudaMalloc(&d_C,size); //copy data to device cudaMemcpy(d_A,h_A,size,cudaMemcpyHostToDevice); cudaMemcpy(d_B,h_B,size,cudaMemcpyHostToDevice); cudaMemcpy(d_C,h_C,size,cudaMemcpyHostToDevice); dim3 threadsPerBlock(TILE_WIDTH,TILE_WIDTH); // 2d block int blocks = ( SIZE + TILE_WIDTH - 1 ) / TILE_WIDTH; dim3 blocksPerGrid(blocks,blocks); // 2d grid // invoke the kernel here MatrixMulKernel<<< blocksPerGrid, threadsPerBlock >>>(d_A,d_B,d_C); // copy results back to host cudaMemcpy(h_C,d_C,size,cudaMemcpyDeviceToHost); // Free up device memory cudaFree(d_A); cudaFree(d_B); cudaFree(d_C); return 0; }
.file "tmpxft_0019743a_00000000-6_mat_multiply.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2061: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2061: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z12GetSubMatrixPfii .type _Z12GetSubMatrixPfii, @function _Z12GetSubMatrixPfii: .LFB2057: .cfi_startproc endbr64 pushq %rax .cfi_def_cfa_offset 16 popq %rax .cfi_def_cfa_offset 8 subq $24, %rsp .cfi_def_cfa_offset 32 movl $1, 12(%rsp) movl 12(%rsp), %edi call exit@PLT .cfi_endproc .LFE2057: .size _Z12GetSubMatrixPfii, .-_Z12GetSubMatrixPfii .globl _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_ .type _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_, @function _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_: .LFB2083: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movq %fs:40, %rax movq %rax, 120(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L9 .L5: movq 120(%rsp), %rax subq %fs:40, %rax jne .L10 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L9: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 152 pushq 40(%rsp) .cfi_def_cfa_offset 160 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z15MatrixMulKernelPfS_S_(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L5 .L10: call __stack_chk_fail@PLT .cfi_endproc .LFE2083: .size _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_, .-_Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_ .globl _Z15MatrixMulKernelPfS_S_ .type _Z15MatrixMulKernelPfS_S_, @function _Z15MatrixMulKernelPfS_S_: .LFB2084: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2084: .size _Z15MatrixMulKernelPfS_S_, .-_Z15MatrixMulKernelPfS_S_ .globl main .type main, @function main: .LFB2058: .cfi_startproc endbr64 leaq -3145728(%rsp), %r11 .cfi_def_cfa 11, 3145736 .LPSRL0: subq $4096, %rsp orq $0, (%rsp) cmpq %r11, %rsp jne .LPSRL0 .cfi_def_cfa_register 7 subq $72, %rsp .cfi_def_cfa_offset 3145808 movq %fs:40, %rax movq %rax, 3145784(%rsp) xorl %eax, %eax movss .LC0(%rip), %xmm3 movss .LC1(%rip), %xmm2 .L14: pxor %xmm1, %xmm1 cvtsi2ssl %eax, %xmm1 movss %xmm1, 48(%rsp,%rax,4) movaps %xmm3, %xmm0 subss %xmm1, %xmm0 subss %xmm2, %xmm0 movss %xmm0, 1048624(%rsp,%rax,4) movl $0x00000000, 2097200(%rsp,%rax,4) addq $1, %rax cmpq $262144, %rax jne .L14 movq %rsp, %rdi movl $1048576, %esi call cudaMalloc@PLT leaq 8(%rsp), %rdi movl $1048576, %esi call cudaMalloc@PLT leaq 16(%rsp), %rdi movl $1048576, %esi call cudaMalloc@PLT leaq 48(%rsp), %rsi movl $1, %ecx movl $1048576, %edx movq (%rsp), %rdi call cudaMemcpy@PLT leaq 1048624(%rsp), %rsi movl $1, %ecx movl $1048576, %edx movq 8(%rsp), %rdi call cudaMemcpy@PLT leaq 2097200(%rsp), %rsi movl $1, %ecx movl $1048576, %edx movq 16(%rsp), %rdi call cudaMemcpy@PLT movl $16, 24(%rsp) movl $16, 28(%rsp) movl $1, 32(%rsp) movl $32, 36(%rsp) movl $32, 40(%rsp) movl $1, 44(%rsp) movl $0, %r9d movl $0, %r8d movq 24(%rsp), %rdx movl $1, %ecx movq 36(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L19 .L15: leaq 2097200(%rsp), %rdi movl $2, %ecx movl $1048576, %edx movq 16(%rsp), %rsi call cudaMemcpy@PLT movq (%rsp), %rdi call cudaFree@PLT movq 8(%rsp), %rdi call cudaFree@PLT movq 16(%rsp), %rdi call cudaFree@PLT movq 3145784(%rsp), %rax subq %fs:40, %rax jne .L20 movl $0, %eax addq $3145800, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L19: .cfi_restore_state movq 16(%rsp), %rdx movq 8(%rsp), %rsi movq (%rsp), %rdi call _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_ jmp .L15 .L20: call __stack_chk_fail@PLT .cfi_endproc .LFE2058: .size main, .-main .section .rodata.str1.1,"aMS",@progbits,1 .LC3: .string "_Z15MatrixMulKernelPfS_S_" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2086: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC3(%rip), %rdx movq %rdx, %rcx leaq _Z15MatrixMulKernelPfS_S_(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2086: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC0: .long 1216348160 .align 4 .LC1: .long 1065353216 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
// A shared memory matrix multiplication program #include "stdio.h" #include "stdlib.h" // Keep the SIZE evenly divisible by TILE_WIDTH #define SIZE 512 #define TILE_WIDTH 16 // kernels that are called by another kernel use the __device__ identifier __device__ float * GetSubMatrix(float * large_matrix, int row, int col) { // this returns the address of the first element in a sub-matrix // when using this sub-matrix you have to jump ahead SIZE elements per row float * subMatrix = &large_matrix[ row * SIZE * TILE_WIDTH + col * TILE_WIDTH ]; return subMatrix; } // matrix multiplication kernel definition __global__ void MatrixMulKernel(float * A,float * B,float * C) { int blockRow = blockIdx.y; int blockCol = blockIdx.x; // Each thread block computes one sub-matrix Csub of C // Here we call another kernel from within a kernel. // Note that we do not need to pass any grid or block // information between the <<< >>> symbols. float * Csub = GetSubMatrix(C, blockRow, blockCol); // Each thread computes on element of Csub // by accumulating results into Csub float Cvalue = 0.0; // Thread row and column within Csub int row = threadIdx.y; int col = threadIdx.x; // Iterate through n sub_matrices, where n depends on the size of the tile and matrix int sub_matrices_per_row = gridDim.x; int i; for ( i = 0; i < sub_matrices_per_row; i++ ) { // Get sub-matrices // Block threads will work on these sub-matrices float * Asub = GetSubMatrix(A, blockRow, i); float * Bsub = GetSubMatrix(B, i, blockCol); // Shared memory used to store Asub and Bsub __shared__ float As[TILE_WIDTH][TILE_WIDTH]; __shared__ float Bs[TILE_WIDTH][TILE_WIDTH]; // Load Asub and Bsub from device memory into shared memory // Each thread loads one element of each sub-matrix As[row][col] = Asub[ row*SIZE + col ]; Bs[row][col] = Bsub[ row*SIZE + col ]; // Sync threads to make sure sub-matrices are completely loaded into shared memory // Remember, this only syncs threads within a block __syncthreads(); int j; // Multiply Asub and Bsub together, using fast shared memory for data access for ( j = 0; j < TILE_WIDTH ; j++ ) Cvalue += As[row][j] * Bs[j][col]; // Sync threads to ensure the preceding computation is done before loading // two new sub-matrices of A and B in the next iteration __syncthreads(); } Csub[ row*SIZE + col ] = Cvalue; } int main(int argc, char ** argv) { float h_A[SIZE*SIZE],h_B[SIZE*SIZE],h_C[SIZE*SIZE]; float * d_A, * d_B, * d_C; // initialize host matrices with arbitrary data int i; for (i=0;i<SIZE*SIZE;i++) { h_A[i] = (float)i; h_B[i] = (float)SIZE * (float)SIZE - (float)i - 1.00; h_C[i] = 0.0; } // allocate space on device size_t size = SIZE*SIZE*sizeof(float); cudaMalloc(&d_A,size); cudaMalloc(&d_B,size); cudaMalloc(&d_C,size); //copy data to device cudaMemcpy(d_A,h_A,size,cudaMemcpyHostToDevice); cudaMemcpy(d_B,h_B,size,cudaMemcpyHostToDevice); cudaMemcpy(d_C,h_C,size,cudaMemcpyHostToDevice); dim3 threadsPerBlock(TILE_WIDTH,TILE_WIDTH); // 2d block int blocks = ( SIZE + TILE_WIDTH - 1 ) / TILE_WIDTH; dim3 blocksPerGrid(blocks,blocks); // 2d grid // invoke the kernel here MatrixMulKernel<<< blocksPerGrid, threadsPerBlock >>>(d_A,d_B,d_C); // copy results back to host cudaMemcpy(h_C,d_C,size,cudaMemcpyDeviceToHost); // Free up device memory cudaFree(d_A); cudaFree(d_B); cudaFree(d_C); return 0; }
// A shared memory matrix multiplication program #include <hip/hip_runtime.h> #include "stdio.h" #include "stdlib.h" // Keep the SIZE evenly divisible by TILE_WIDTH #define SIZE 512 #define TILE_WIDTH 16 // kernels that are called by another kernel use the __device__ identifier __device__ float * GetSubMatrix(float * large_matrix, int row, int col) { // this returns the address of the first element in a sub-matrix // when using this sub-matrix you have to jump ahead SIZE elements per row float * subMatrix = &large_matrix[ row * SIZE * TILE_WIDTH + col * TILE_WIDTH ]; return subMatrix; } // matrix multiplication kernel definition __global__ void MatrixMulKernel(float * A,float * B,float * C) { int blockRow = blockIdx.y; int blockCol = blockIdx.x; // Each thread block computes one sub-matrix Csub of C // Here we call another kernel from within a kernel. // Note that we do not need to pass any grid or block // information between the <<< >>> symbols. float * Csub = GetSubMatrix(C, blockRow, blockCol); // Each thread computes on element of Csub // by accumulating results into Csub float Cvalue = 0.0; // Thread row and column within Csub int row = threadIdx.y; int col = threadIdx.x; // Iterate through n sub_matrices, where n depends on the size of the tile and matrix int sub_matrices_per_row = gridDim.x; int i; for ( i = 0; i < sub_matrices_per_row; i++ ) { // Get sub-matrices // Block threads will work on these sub-matrices float * Asub = GetSubMatrix(A, blockRow, i); float * Bsub = GetSubMatrix(B, i, blockCol); // Shared memory used to store Asub and Bsub __shared__ float As[TILE_WIDTH][TILE_WIDTH]; __shared__ float Bs[TILE_WIDTH][TILE_WIDTH]; // Load Asub and Bsub from device memory into shared memory // Each thread loads one element of each sub-matrix As[row][col] = Asub[ row*SIZE + col ]; Bs[row][col] = Bsub[ row*SIZE + col ]; // Sync threads to make sure sub-matrices are completely loaded into shared memory // Remember, this only syncs threads within a block __syncthreads(); int j; // Multiply Asub and Bsub together, using fast shared memory for data access for ( j = 0; j < TILE_WIDTH ; j++ ) Cvalue += As[row][j] * Bs[j][col]; // Sync threads to ensure the preceding computation is done before loading // two new sub-matrices of A and B in the next iteration __syncthreads(); } Csub[ row*SIZE + col ] = Cvalue; } int main(int argc, char ** argv) { float h_A[SIZE*SIZE],h_B[SIZE*SIZE],h_C[SIZE*SIZE]; float * d_A, * d_B, * d_C; // initialize host matrices with arbitrary data int i; for (i=0;i<SIZE*SIZE;i++) { h_A[i] = (float)i; h_B[i] = (float)SIZE * (float)SIZE - (float)i - 1.00; h_C[i] = 0.0; } // allocate space on device size_t size = SIZE*SIZE*sizeof(float); hipMalloc(&d_A,size); hipMalloc(&d_B,size); hipMalloc(&d_C,size); //copy data to device hipMemcpy(d_A,h_A,size,hipMemcpyHostToDevice); hipMemcpy(d_B,h_B,size,hipMemcpyHostToDevice); hipMemcpy(d_C,h_C,size,hipMemcpyHostToDevice); dim3 threadsPerBlock(TILE_WIDTH,TILE_WIDTH); // 2d block int blocks = ( SIZE + TILE_WIDTH - 1 ) / TILE_WIDTH; dim3 blocksPerGrid(blocks,blocks); // 2d grid // invoke the kernel here MatrixMulKernel<<< blocksPerGrid, threadsPerBlock >>>(d_A,d_B,d_C); // copy results back to host hipMemcpy(h_C,d_C,size,hipMemcpyDeviceToHost); // Free up device memory hipFree(d_A); hipFree(d_B); hipFree(d_C); return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
// A shared memory matrix multiplication program #include <hip/hip_runtime.h> #include "stdio.h" #include "stdlib.h" // Keep the SIZE evenly divisible by TILE_WIDTH #define SIZE 512 #define TILE_WIDTH 16 // kernels that are called by another kernel use the __device__ identifier __device__ float * GetSubMatrix(float * large_matrix, int row, int col) { // this returns the address of the first element in a sub-matrix // when using this sub-matrix you have to jump ahead SIZE elements per row float * subMatrix = &large_matrix[ row * SIZE * TILE_WIDTH + col * TILE_WIDTH ]; return subMatrix; } // matrix multiplication kernel definition __global__ void MatrixMulKernel(float * A,float * B,float * C) { int blockRow = blockIdx.y; int blockCol = blockIdx.x; // Each thread block computes one sub-matrix Csub of C // Here we call another kernel from within a kernel. // Note that we do not need to pass any grid or block // information between the <<< >>> symbols. float * Csub = GetSubMatrix(C, blockRow, blockCol); // Each thread computes on element of Csub // by accumulating results into Csub float Cvalue = 0.0; // Thread row and column within Csub int row = threadIdx.y; int col = threadIdx.x; // Iterate through n sub_matrices, where n depends on the size of the tile and matrix int sub_matrices_per_row = gridDim.x; int i; for ( i = 0; i < sub_matrices_per_row; i++ ) { // Get sub-matrices // Block threads will work on these sub-matrices float * Asub = GetSubMatrix(A, blockRow, i); float * Bsub = GetSubMatrix(B, i, blockCol); // Shared memory used to store Asub and Bsub __shared__ float As[TILE_WIDTH][TILE_WIDTH]; __shared__ float Bs[TILE_WIDTH][TILE_WIDTH]; // Load Asub and Bsub from device memory into shared memory // Each thread loads one element of each sub-matrix As[row][col] = Asub[ row*SIZE + col ]; Bs[row][col] = Bsub[ row*SIZE + col ]; // Sync threads to make sure sub-matrices are completely loaded into shared memory // Remember, this only syncs threads within a block __syncthreads(); int j; // Multiply Asub and Bsub together, using fast shared memory for data access for ( j = 0; j < TILE_WIDTH ; j++ ) Cvalue += As[row][j] * Bs[j][col]; // Sync threads to ensure the preceding computation is done before loading // two new sub-matrices of A and B in the next iteration __syncthreads(); } Csub[ row*SIZE + col ] = Cvalue; } int main(int argc, char ** argv) { float h_A[SIZE*SIZE],h_B[SIZE*SIZE],h_C[SIZE*SIZE]; float * d_A, * d_B, * d_C; // initialize host matrices with arbitrary data int i; for (i=0;i<SIZE*SIZE;i++) { h_A[i] = (float)i; h_B[i] = (float)SIZE * (float)SIZE - (float)i - 1.00; h_C[i] = 0.0; } // allocate space on device size_t size = SIZE*SIZE*sizeof(float); hipMalloc(&d_A,size); hipMalloc(&d_B,size); hipMalloc(&d_C,size); //copy data to device hipMemcpy(d_A,h_A,size,hipMemcpyHostToDevice); hipMemcpy(d_B,h_B,size,hipMemcpyHostToDevice); hipMemcpy(d_C,h_C,size,hipMemcpyHostToDevice); dim3 threadsPerBlock(TILE_WIDTH,TILE_WIDTH); // 2d block int blocks = ( SIZE + TILE_WIDTH - 1 ) / TILE_WIDTH; dim3 blocksPerGrid(blocks,blocks); // 2d grid // invoke the kernel here MatrixMulKernel<<< blocksPerGrid, threadsPerBlock >>>(d_A,d_B,d_C); // copy results back to host hipMemcpy(h_C,d_C,size,hipMemcpyDeviceToHost); // Free up device memory hipFree(d_A); hipFree(d_B); hipFree(d_C); return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z15MatrixMulKernelPfS_S_ .globl _Z15MatrixMulKernelPfS_S_ .p2align 8 .type _Z15MatrixMulKernelPfS_S_,@function _Z15MatrixMulKernelPfS_S_: s_load_b32 s4, s[0:1], 0x18 v_bfe_u32 v1, v0, 10, 10 v_and_b32_e32 v0, 0x3ff, v0 s_lshl_b32 s2, s15, 13 s_lshl_b32 s3, s14, 4 s_waitcnt lgkmcnt(0) s_cmp_lt_i32 s4, 1 s_cbranch_scc1 .LBB0_5 s_load_b128 s[8:11], s[0:1], 0x0 v_lshlrev_b32_e32 v2, 9, v1 v_lshlrev_b32_e32 v5, 2, v0 v_lshlrev_b32_e32 v3, 6, v1 s_delay_alu instid0(VALU_DEP_3) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_add_lshl_u32 v8, v2, v0, 2 v_mov_b32_e32 v2, 0 v_add_nc_u32_e32 v4, v3, v5 v_add_nc_u32_e32 v5, 0x400, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_add_nc_u32_e32 v10, v5, v3 s_waitcnt lgkmcnt(0) v_add_co_u32 v6, s5, s8, v8 v_add_co_ci_u32_e64 v7, null, s9, 0, s5 v_add_co_u32 v8, s5, s10, v8 s_delay_alu instid0(VALU_DEP_1) v_add_co_ci_u32_e64 v9, null, s11, 0, s5 s_mov_b32 s5, 0 s_set_inst_prefetch_distance 0x1 .p2align 6 .LBB0_2: s_lshl_b32 s6, s5, 4 s_lshl_b32 s7, s5, 13 s_add_i32 s6, s6, s2 s_add_i32 s8, s7, s3 s_ashr_i32 s7, s6, 31 s_ashr_i32 s9, s8, 31 s_lshl_b64 s[6:7], s[6:7], 2 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) v_add_co_u32 v11, vcc_lo, v6, s6 v_add_co_ci_u32_e32 v12, vcc_lo, s7, v7, vcc_lo s_lshl_b64 s[6:7], s[8:9], 2 v_add_co_u32 v13, vcc_lo, v8, s6 v_add_co_ci_u32_e32 v14, vcc_lo, s7, v9, vcc_lo global_load_b32 v12, v[11:12], off global_load_b32 v13, v[13:14], off v_mov_b32_e32 v11, v5 s_mov_b32 s6, 0 s_waitcnt vmcnt(1) ds_store_b32 v4, v12 s_waitcnt vmcnt(0) ds_store_b32 v10, v13 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv .LBB0_3: v_add_nc_u32_e32 v12, s6, v3 s_add_i32 s6, s6, 4 ds_load_b32 v13, v11 ds_load_b32 v12, v12 v_add_nc_u32_e32 v11, 64, v11 s_cmp_lg_u32 s6, 64 s_waitcnt lgkmcnt(0) v_fmac_f32_e32 v2, v12, v13 s_cbranch_scc1 .LBB0_3 s_add_i32 s5, s5, 1 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_lg_u32 s5, s4 s_barrier buffer_gl0_inv s_cbranch_scc1 .LBB0_2 s_branch .LBB0_6 .LBB0_5: v_mov_b32_e32 v2, 0 .LBB0_6: s_set_inst_prefetch_distance 0x2 s_load_b64 s[0:1], s[0:1], 0x10 v_lshlrev_b32_e32 v1, 9, v1 s_add_i32 s2, s3, s2 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_ashr_i32 s3, s2, 31 s_lshl_b64 s[2:3], s[2:3], 2 s_delay_alu instid0(VALU_DEP_1) v_add_lshl_u32 v0, v1, v0, 2 s_waitcnt lgkmcnt(0) s_add_u32 s0, s0, s2 s_addc_u32 s1, s1, s3 global_store_b32 v0, v2, s[0:1] s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z15MatrixMulKernelPfS_S_ .amdhsa_group_segment_fixed_size 2048 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 280 .amdhsa_user_sgpr_count 14 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 1 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 1 .amdhsa_next_free_vgpr 15 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z15MatrixMulKernelPfS_S_, .Lfunc_end0-_Z15MatrixMulKernelPfS_S_ .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: hidden_block_count_x - .offset: 28 .size: 4 .value_kind: hidden_block_count_y - .offset: 32 .size: 4 .value_kind: hidden_block_count_z - .offset: 36 .size: 2 .value_kind: hidden_group_size_x - .offset: 38 .size: 2 .value_kind: hidden_group_size_y - .offset: 40 .size: 2 .value_kind: hidden_group_size_z - .offset: 42 .size: 2 .value_kind: hidden_remainder_x - .offset: 44 .size: 2 .value_kind: hidden_remainder_y - .offset: 46 .size: 2 .value_kind: hidden_remainder_z - .offset: 64 .size: 8 .value_kind: hidden_global_offset_x - .offset: 72 .size: 8 .value_kind: hidden_global_offset_y - .offset: 80 .size: 8 .value_kind: hidden_global_offset_z - .offset: 88 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 2048 .kernarg_segment_align: 8 .kernarg_segment_size: 280 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z15MatrixMulKernelPfS_S_ .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z15MatrixMulKernelPfS_S_.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 15 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
// A shared memory matrix multiplication program #include <hip/hip_runtime.h> #include "stdio.h" #include "stdlib.h" // Keep the SIZE evenly divisible by TILE_WIDTH #define SIZE 512 #define TILE_WIDTH 16 // kernels that are called by another kernel use the __device__ identifier __device__ float * GetSubMatrix(float * large_matrix, int row, int col) { // this returns the address of the first element in a sub-matrix // when using this sub-matrix you have to jump ahead SIZE elements per row float * subMatrix = &large_matrix[ row * SIZE * TILE_WIDTH + col * TILE_WIDTH ]; return subMatrix; } // matrix multiplication kernel definition __global__ void MatrixMulKernel(float * A,float * B,float * C) { int blockRow = blockIdx.y; int blockCol = blockIdx.x; // Each thread block computes one sub-matrix Csub of C // Here we call another kernel from within a kernel. // Note that we do not need to pass any grid or block // information between the <<< >>> symbols. float * Csub = GetSubMatrix(C, blockRow, blockCol); // Each thread computes on element of Csub // by accumulating results into Csub float Cvalue = 0.0; // Thread row and column within Csub int row = threadIdx.y; int col = threadIdx.x; // Iterate through n sub_matrices, where n depends on the size of the tile and matrix int sub_matrices_per_row = gridDim.x; int i; for ( i = 0; i < sub_matrices_per_row; i++ ) { // Get sub-matrices // Block threads will work on these sub-matrices float * Asub = GetSubMatrix(A, blockRow, i); float * Bsub = GetSubMatrix(B, i, blockCol); // Shared memory used to store Asub and Bsub __shared__ float As[TILE_WIDTH][TILE_WIDTH]; __shared__ float Bs[TILE_WIDTH][TILE_WIDTH]; // Load Asub and Bsub from device memory into shared memory // Each thread loads one element of each sub-matrix As[row][col] = Asub[ row*SIZE + col ]; Bs[row][col] = Bsub[ row*SIZE + col ]; // Sync threads to make sure sub-matrices are completely loaded into shared memory // Remember, this only syncs threads within a block __syncthreads(); int j; // Multiply Asub and Bsub together, using fast shared memory for data access for ( j = 0; j < TILE_WIDTH ; j++ ) Cvalue += As[row][j] * Bs[j][col]; // Sync threads to ensure the preceding computation is done before loading // two new sub-matrices of A and B in the next iteration __syncthreads(); } Csub[ row*SIZE + col ] = Cvalue; } int main(int argc, char ** argv) { float h_A[SIZE*SIZE],h_B[SIZE*SIZE],h_C[SIZE*SIZE]; float * d_A, * d_B, * d_C; // initialize host matrices with arbitrary data int i; for (i=0;i<SIZE*SIZE;i++) { h_A[i] = (float)i; h_B[i] = (float)SIZE * (float)SIZE - (float)i - 1.00; h_C[i] = 0.0; } // allocate space on device size_t size = SIZE*SIZE*sizeof(float); hipMalloc(&d_A,size); hipMalloc(&d_B,size); hipMalloc(&d_C,size); //copy data to device hipMemcpy(d_A,h_A,size,hipMemcpyHostToDevice); hipMemcpy(d_B,h_B,size,hipMemcpyHostToDevice); hipMemcpy(d_C,h_C,size,hipMemcpyHostToDevice); dim3 threadsPerBlock(TILE_WIDTH,TILE_WIDTH); // 2d block int blocks = ( SIZE + TILE_WIDTH - 1 ) / TILE_WIDTH; dim3 blocksPerGrid(blocks,blocks); // 2d grid // invoke the kernel here MatrixMulKernel<<< blocksPerGrid, threadsPerBlock >>>(d_A,d_B,d_C); // copy results back to host hipMemcpy(h_C,d_C,size,hipMemcpyDeviceToHost); // Free up device memory hipFree(d_A); hipFree(d_B); hipFree(d_C); return 0; }
.text .file "mat_multiply.hip" .globl _Z30__device_stub__MatrixMulKernelPfS_S_ # -- Begin function _Z30__device_stub__MatrixMulKernelPfS_S_ .p2align 4, 0x90 .type _Z30__device_stub__MatrixMulKernelPfS_S_,@function _Z30__device_stub__MatrixMulKernelPfS_S_: # @_Z30__device_stub__MatrixMulKernelPfS_S_ .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movq %rdx, 56(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z15MatrixMulKernelPfS_S_, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $120, %rsp .cfi_adjust_cfa_offset -120 retq .Lfunc_end0: .size _Z30__device_stub__MatrixMulKernelPfS_S_, .Lfunc_end0-_Z30__device_stub__MatrixMulKernelPfS_S_ .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function main .LCPI1_0: .long 0x48800000 # float 262144 .LCPI1_1: .long 0xbf800000 # float -1 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 subq $3145856, %rsp # imm = 0x300080 .cfi_def_cfa_offset 3145872 .cfi_offset %rbx, -16 leaq 128(%rsp), %rdi xorl %ebx, %ebx movl $1048576, %edx # imm = 0x100000 xorl %esi, %esi callq memset@PLT movss .LCPI1_0(%rip), %xmm0 # xmm0 = mem[0],zero,zero,zero movss .LCPI1_1(%rip), %xmm1 # xmm1 = mem[0],zero,zero,zero .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 xorps %xmm2, %xmm2 cvtsi2ss %ebx, %xmm2 movss %xmm2, 2097280(%rsp,%rbx,4) movaps %xmm0, %xmm3 subss %xmm2, %xmm3 addss %xmm1, %xmm3 movss %xmm3, 1048704(%rsp,%rbx,4) incq %rbx cmpq $262144, %rbx # imm = 0x40000 jne .LBB1_1 # %bb.2: leaq 16(%rsp), %rdi movl $1048576, %esi # imm = 0x100000 callq hipMalloc leaq 8(%rsp), %rdi movl $1048576, %esi # imm = 0x100000 callq hipMalloc movq %rsp, %rdi movl $1048576, %esi # imm = 0x100000 callq hipMalloc movq 16(%rsp), %rdi leaq 2097280(%rsp), %rsi movl $1048576, %edx # imm = 0x100000 movl $1, %ecx callq hipMemcpy movq 8(%rsp), %rdi leaq 1048704(%rsp), %rsi movl $1048576, %edx # imm = 0x100000 movl $1, %ecx callq hipMemcpy movq (%rsp), %rdi leaq 128(%rsp), %rsi movl $1048576, %edx # imm = 0x100000 movl $1, %ecx callq hipMemcpy movabsq $137438953504, %rdi # imm = 0x2000000020 movabsq $68719476752, %rdx # imm = 0x1000000010 movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_4 # %bb.3: movq 16(%rsp), %rax movq 8(%rsp), %rcx movq (%rsp), %rdx movq %rax, 88(%rsp) movq %rcx, 80(%rsp) movq %rdx, 72(%rsp) leaq 88(%rsp), %rax movq %rax, 96(%rsp) leaq 80(%rsp), %rax movq %rax, 104(%rsp) leaq 72(%rsp), %rax movq %rax, 112(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 96(%rsp), %r9 movl $_Z15MatrixMulKernelPfS_S_, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_4: movq (%rsp), %rsi leaq 128(%rsp), %rdi movl $1048576, %edx # imm = 0x100000 movl $2, %ecx callq hipMemcpy movq 16(%rsp), %rdi callq hipFree movq 8(%rsp), %rdi callq hipFree movq (%rsp), %rdi callq hipFree xorl %eax, %eax addq $3145856, %rsp # imm = 0x300080 .cfi_def_cfa_offset 16 popq %rbx .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z15MatrixMulKernelPfS_S_, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z15MatrixMulKernelPfS_S_,@object # @_Z15MatrixMulKernelPfS_S_ .section .rodata,"a",@progbits .globl _Z15MatrixMulKernelPfS_S_ .p2align 3, 0x0 _Z15MatrixMulKernelPfS_S_: .quad _Z30__device_stub__MatrixMulKernelPfS_S_ .size _Z15MatrixMulKernelPfS_S_, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z15MatrixMulKernelPfS_S_" .size .L__unnamed_1, 26 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z30__device_stub__MatrixMulKernelPfS_S_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z15MatrixMulKernelPfS_S_ .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_0019743a_00000000-6_mat_multiply.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2061: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2061: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z12GetSubMatrixPfii .type _Z12GetSubMatrixPfii, @function _Z12GetSubMatrixPfii: .LFB2057: .cfi_startproc endbr64 pushq %rax .cfi_def_cfa_offset 16 popq %rax .cfi_def_cfa_offset 8 subq $24, %rsp .cfi_def_cfa_offset 32 movl $1, 12(%rsp) movl 12(%rsp), %edi call exit@PLT .cfi_endproc .LFE2057: .size _Z12GetSubMatrixPfii, .-_Z12GetSubMatrixPfii .globl _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_ .type _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_, @function _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_: .LFB2083: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movq %fs:40, %rax movq %rax, 120(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L9 .L5: movq 120(%rsp), %rax subq %fs:40, %rax jne .L10 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L9: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 152 pushq 40(%rsp) .cfi_def_cfa_offset 160 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z15MatrixMulKernelPfS_S_(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L5 .L10: call __stack_chk_fail@PLT .cfi_endproc .LFE2083: .size _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_, .-_Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_ .globl _Z15MatrixMulKernelPfS_S_ .type _Z15MatrixMulKernelPfS_S_, @function _Z15MatrixMulKernelPfS_S_: .LFB2084: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2084: .size _Z15MatrixMulKernelPfS_S_, .-_Z15MatrixMulKernelPfS_S_ .globl main .type main, @function main: .LFB2058: .cfi_startproc endbr64 leaq -3145728(%rsp), %r11 .cfi_def_cfa 11, 3145736 .LPSRL0: subq $4096, %rsp orq $0, (%rsp) cmpq %r11, %rsp jne .LPSRL0 .cfi_def_cfa_register 7 subq $72, %rsp .cfi_def_cfa_offset 3145808 movq %fs:40, %rax movq %rax, 3145784(%rsp) xorl %eax, %eax movss .LC0(%rip), %xmm3 movss .LC1(%rip), %xmm2 .L14: pxor %xmm1, %xmm1 cvtsi2ssl %eax, %xmm1 movss %xmm1, 48(%rsp,%rax,4) movaps %xmm3, %xmm0 subss %xmm1, %xmm0 subss %xmm2, %xmm0 movss %xmm0, 1048624(%rsp,%rax,4) movl $0x00000000, 2097200(%rsp,%rax,4) addq $1, %rax cmpq $262144, %rax jne .L14 movq %rsp, %rdi movl $1048576, %esi call cudaMalloc@PLT leaq 8(%rsp), %rdi movl $1048576, %esi call cudaMalloc@PLT leaq 16(%rsp), %rdi movl $1048576, %esi call cudaMalloc@PLT leaq 48(%rsp), %rsi movl $1, %ecx movl $1048576, %edx movq (%rsp), %rdi call cudaMemcpy@PLT leaq 1048624(%rsp), %rsi movl $1, %ecx movl $1048576, %edx movq 8(%rsp), %rdi call cudaMemcpy@PLT leaq 2097200(%rsp), %rsi movl $1, %ecx movl $1048576, %edx movq 16(%rsp), %rdi call cudaMemcpy@PLT movl $16, 24(%rsp) movl $16, 28(%rsp) movl $1, 32(%rsp) movl $32, 36(%rsp) movl $32, 40(%rsp) movl $1, 44(%rsp) movl $0, %r9d movl $0, %r8d movq 24(%rsp), %rdx movl $1, %ecx movq 36(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L19 .L15: leaq 2097200(%rsp), %rdi movl $2, %ecx movl $1048576, %edx movq 16(%rsp), %rsi call cudaMemcpy@PLT movq (%rsp), %rdi call cudaFree@PLT movq 8(%rsp), %rdi call cudaFree@PLT movq 16(%rsp), %rdi call cudaFree@PLT movq 3145784(%rsp), %rax subq %fs:40, %rax jne .L20 movl $0, %eax addq $3145800, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L19: .cfi_restore_state movq 16(%rsp), %rdx movq 8(%rsp), %rsi movq (%rsp), %rdi call _Z39__device_stub__Z15MatrixMulKernelPfS_S_PfS_S_ jmp .L15 .L20: call __stack_chk_fail@PLT .cfi_endproc .LFE2058: .size main, .-main .section .rodata.str1.1,"aMS",@progbits,1 .LC3: .string "_Z15MatrixMulKernelPfS_S_" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2086: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC3(%rip), %rdx movq %rdx, %rcx leaq _Z15MatrixMulKernelPfS_S_(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2086: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC0: .long 1216348160 .align 4 .LC1: .long 1065353216 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "mat_multiply.hip" .globl _Z30__device_stub__MatrixMulKernelPfS_S_ # -- Begin function _Z30__device_stub__MatrixMulKernelPfS_S_ .p2align 4, 0x90 .type _Z30__device_stub__MatrixMulKernelPfS_S_,@function _Z30__device_stub__MatrixMulKernelPfS_S_: # @_Z30__device_stub__MatrixMulKernelPfS_S_ .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movq %rdx, 56(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z15MatrixMulKernelPfS_S_, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $120, %rsp .cfi_adjust_cfa_offset -120 retq .Lfunc_end0: .size _Z30__device_stub__MatrixMulKernelPfS_S_, .Lfunc_end0-_Z30__device_stub__MatrixMulKernelPfS_S_ .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function main .LCPI1_0: .long 0x48800000 # float 262144 .LCPI1_1: .long 0xbf800000 # float -1 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 subq $3145856, %rsp # imm = 0x300080 .cfi_def_cfa_offset 3145872 .cfi_offset %rbx, -16 leaq 128(%rsp), %rdi xorl %ebx, %ebx movl $1048576, %edx # imm = 0x100000 xorl %esi, %esi callq memset@PLT movss .LCPI1_0(%rip), %xmm0 # xmm0 = mem[0],zero,zero,zero movss .LCPI1_1(%rip), %xmm1 # xmm1 = mem[0],zero,zero,zero .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 xorps %xmm2, %xmm2 cvtsi2ss %ebx, %xmm2 movss %xmm2, 2097280(%rsp,%rbx,4) movaps %xmm0, %xmm3 subss %xmm2, %xmm3 addss %xmm1, %xmm3 movss %xmm3, 1048704(%rsp,%rbx,4) incq %rbx cmpq $262144, %rbx # imm = 0x40000 jne .LBB1_1 # %bb.2: leaq 16(%rsp), %rdi movl $1048576, %esi # imm = 0x100000 callq hipMalloc leaq 8(%rsp), %rdi movl $1048576, %esi # imm = 0x100000 callq hipMalloc movq %rsp, %rdi movl $1048576, %esi # imm = 0x100000 callq hipMalloc movq 16(%rsp), %rdi leaq 2097280(%rsp), %rsi movl $1048576, %edx # imm = 0x100000 movl $1, %ecx callq hipMemcpy movq 8(%rsp), %rdi leaq 1048704(%rsp), %rsi movl $1048576, %edx # imm = 0x100000 movl $1, %ecx callq hipMemcpy movq (%rsp), %rdi leaq 128(%rsp), %rsi movl $1048576, %edx # imm = 0x100000 movl $1, %ecx callq hipMemcpy movabsq $137438953504, %rdi # imm = 0x2000000020 movabsq $68719476752, %rdx # imm = 0x1000000010 movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_4 # %bb.3: movq 16(%rsp), %rax movq 8(%rsp), %rcx movq (%rsp), %rdx movq %rax, 88(%rsp) movq %rcx, 80(%rsp) movq %rdx, 72(%rsp) leaq 88(%rsp), %rax movq %rax, 96(%rsp) leaq 80(%rsp), %rax movq %rax, 104(%rsp) leaq 72(%rsp), %rax movq %rax, 112(%rsp) leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 96(%rsp), %r9 movl $_Z15MatrixMulKernelPfS_S_, %edi pushq 24(%rsp) .cfi_adjust_cfa_offset 8 pushq 40(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_4: movq (%rsp), %rsi leaq 128(%rsp), %rdi movl $1048576, %edx # imm = 0x100000 movl $2, %ecx callq hipMemcpy movq 16(%rsp), %rdi callq hipFree movq 8(%rsp), %rdi callq hipFree movq (%rsp), %rdi callq hipFree xorl %eax, %eax addq $3145856, %rsp # imm = 0x300080 .cfi_def_cfa_offset 16 popq %rbx .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z15MatrixMulKernelPfS_S_, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z15MatrixMulKernelPfS_S_,@object # @_Z15MatrixMulKernelPfS_S_ .section .rodata,"a",@progbits .globl _Z15MatrixMulKernelPfS_S_ .p2align 3, 0x0 _Z15MatrixMulKernelPfS_S_: .quad _Z30__device_stub__MatrixMulKernelPfS_S_ .size _Z15MatrixMulKernelPfS_S_, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z15MatrixMulKernelPfS_S_" .size .L__unnamed_1, 26 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z30__device_stub__MatrixMulKernelPfS_S_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z15MatrixMulKernelPfS_S_ .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
#include "includes.h" __global__ void _setPrecisionKernel(float* data, size_t size, int* precision) { unsigned int idx = blockDim.x * blockIdx.x + threadIdx.x; if (idx >= size) return; int prec = precision[idx]; int mul = 1; while(prec--) mul *= 10; data[idx] = (float)(int)(data[idx]*mul); data[idx] /= mul; }
code for sm_80 Function : _Z19_setPrecisionKernelPfmPi .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ IMAD.MOV.U32 R1, RZ, RZ, c[0x0][0x28] ; /* 0x00000a00ff017624 */ /* 0x000fe400078e00ff */ /*0010*/ S2R R0, SR_CTAID.X ; /* 0x0000000000007919 */ /* 0x000e280000002500 */ /*0020*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e240000002100 */ /*0030*/ IMAD R0, R0, c[0x0][0x0], R3 ; /* 0x0000000000007a24 */ /* 0x001fca00078e0203 */ /*0040*/ ISETP.GE.U32.AND P0, PT, R0, c[0x0][0x168], PT ; /* 0x00005a0000007a0c */ /* 0x000fc80003f06070 */ /*0050*/ ISETP.GE.U32.AND.EX P0, PT, RZ, c[0x0][0x16c], PT, P0 ; /* 0x00005b00ff007a0c */ /* 0x000fda0003f06100 */ /*0060*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0070*/ IMAD.SHL.U32 R4, R0, 0x4, RZ ; /* 0x0000000400047824 */ /* 0x000fe200078e00ff */ /*0080*/ SHF.R.U32.HI R5, RZ, 0x1e, R0 ; /* 0x0000001eff057819 */ /* 0x000fe20000011600 */ /*0090*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fc60000000a00 */ /*00a0*/ IADD3 R2, P0, R4, c[0x0][0x170], RZ ; /* 0x00005c0004027a10 */ /* 0x000fc80007f1e0ff */ /*00b0*/ IADD3.X R3, R5, c[0x0][0x174], RZ, P0, !PT ; /* 0x00005d0005037a10 */ /* 0x000fca00007fe4ff */ /*00c0*/ LDG.E R2, [R2.64] ; /* 0x0000000402027981 */ /* 0x000ea2000c1e1900 */ /*00d0*/ BSSY B0, 0x3a0 ; /* 0x000002c000007945 */ /* 0x000fe20003800000 */ /*00e0*/ IMAD.MOV.U32 R0, RZ, RZ, 0x3f800000 ; /* 0x3f800000ff007424 */ /* 0x000fe200078e00ff */ /*00f0*/ ISETP.NE.AND P0, PT, R2, RZ, PT ; /* 0x000000ff0200720c */ /* 0x004fda0003f05270 */ /*0100*/ @!P0 BRA 0x390 ; /* 0x0000028000008947 */ /* 0x000fea0003800000 */ /*0110*/ IADD3 R0, R2.reuse, -0x1, RZ ; /* 0xffffffff02007810 */ /* 0x040fe20007ffe0ff */ /*0120*/ BSSY B1, 0x300 ; /* 0x000001d000017945 */ /* 0x000fe20003800000 */ /*0130*/ LOP3.LUT R3, R2, 0x3, RZ, 0xc0, !PT ; /* 0x0000000302037812 */ /* 0x000fe400078ec0ff */ /*0140*/ ISETP.GE.U32.AND P0, PT, R0, 0x3, PT ; /* 0x000000030000780c */ /* 0x000fe20003f06070 */ /*0150*/ IMAD.MOV.U32 R0, RZ, RZ, 0x1 ; /* 0x00000001ff007424 */ /* 0x000fd800078e00ff */ /*0160*/ @!P0 BRA 0x2f0 ; /* 0x0000018000008947 */ /* 0x000fea0003800000 */ /*0170*/ IMAD.IADD R2, R2, 0x1, -R3 ; /* 0x0000000102027824 */ /* 0x000fe400078e0a03 */ /*0180*/ IMAD.MOV.U32 R0, RZ, RZ, 0x1 ; /* 0x00000001ff007424 */ /* 0x000fc600078e00ff */ /*0190*/ ISETP.GT.AND P0, PT, R2, RZ, PT ; /* 0x000000ff0200720c */ /* 0x000fda0003f04270 */ /*01a0*/ @!P0 BRA 0x2b0 ; /* 0x0000010000008947 */ /* 0x000fea0003800000 */ /*01b0*/ ISETP.GT.AND P1, PT, R2, 0xc, PT ; /* 0x0000000c0200780c */ /* 0x000fe20003f24270 */ /*01c0*/ BSSY B2, 0x250 ; /* 0x0000008000027945 */ /* 0x000fe20003800000 */ /*01d0*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x80, 0x0 ; /* 0x000000000000781c */ /* 0x000fd60003f0f070 */ /*01e0*/ @!P1 BRA 0x240 ; /* 0x0000005000009947 */ /* 0x000fea0003800000 */ /*01f0*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe40003f0e170 */ /*0200*/ IADD3 R2, R2, -0x10, RZ ; /* 0xfffffff002027810 */ /* 0x000fe20007ffe0ff */ /*0210*/ IMAD R0, R0, 0x6fc10000, RZ ; /* 0x6fc1000000007824 */ /* 0x000fc600078e02ff */ /*0220*/ ISETP.GT.AND P1, PT, R2, 0xc, PT ; /* 0x0000000c0200780c */ /* 0x000fda0003f24270 */ /*0230*/ @P1 BRA 0x200 ; /* 0xffffffc000001947 */ /* 0x000fea000383ffff */ /*0240*/ BSYNC B2 ; /* 0x0000000000027941 */ /* 0x000fea0003800000 */ /*0250*/ ISETP.GT.AND P1, PT, R2, 0x4, PT ; /* 0x000000040200780c */ /* 0x000fda0003f24270 */ /*0260*/ @P1 PLOP3.LUT P0, PT, PT, PT, PT, 0x8, 0x0 ; /* 0x000000000000181c */ /* 0x000fe20003f0e170 */ /*0270*/ @P1 IMAD R0, R0, 0x5f5e100, RZ ; /* 0x05f5e10000001824 */ /* 0x000fe200078e02ff */ /*0280*/ @P1 IADD3 R2, R2, -0x8, RZ ; /* 0xfffffff802021810 */ /* 0x000fd60007ffe0ff */ /*0290*/ ISETP.NE.OR P0, PT, R2, RZ, P0 ; /* 0x000000ff0200720c */ /* 0x000fda0000705670 */ /*02a0*/ @!P0 BRA 0x2f0 ; /* 0x0000004000008947 */ /* 0x000fea0003800000 */ /*02b0*/ IADD3 R2, R2, -0x4, RZ ; /* 0xfffffffc02027810 */ /* 0x000fe20007ffe0ff */ /*02c0*/ IMAD R0, R0, 0x2710, RZ ; /* 0x0000271000007824 */ /* 0x000fc600078e02ff */ /*02d0*/ ISETP.NE.AND P0, PT, R2, RZ, PT ; /* 0x000000ff0200720c */ /* 0x000fda0003f05270 */ /*02e0*/ @P0 BRA 0x2b0 ; /* 0xffffffc000000947 */ /* 0x000fea000383ffff */ /*02f0*/ BSYNC B1 ; /* 0x0000000000017941 */ /* 0x000fea0003800000 */ /*0300*/ ISETP.NE.AND P0, PT, R3, RZ, PT ; /* 0x000000ff0300720c */ /* 0x000fe20003f05270 */ /*0310*/ BSSY B1, 0x380 ; /* 0x0000006000017945 */ /* 0x000fd80003800000 */ /*0320*/ @!P0 BRA 0x370 ; /* 0x0000004000008947 */ /* 0x000fea0003800000 */ /*0330*/ IADD3 R3, R3, -0x1, RZ ; /* 0xffffffff03037810 */ /* 0x000fe20007ffe0ff */ /*0340*/ IMAD R0, R0, 0xa, RZ ; /* 0x0000000a00007824 */ /* 0x000fc600078e02ff */ /*0350*/ ISETP.NE.AND P0, PT, R3, RZ, PT ; /* 0x000000ff0300720c */ /* 0x000fda0003f05270 */ /*0360*/ @P0 BRA 0x330 ; /* 0xffffffc000000947 */ /* 0x000fea000383ffff */ /*0370*/ BSYNC B1 ; /* 0x0000000000017941 */ /* 0x000fea0003800000 */ /*0380*/ I2F R0, R0 ; /* 0x0000000000007306 */ /* 0x000e240000201400 */ /*0390*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*03a0*/ IADD3 R2, P0, R4, c[0x0][0x160], RZ ; /* 0x0000580004027a10 */ /* 0x000fc80007f1e0ff */ /*03b0*/ IADD3.X R3, R5, c[0x0][0x164], RZ, P0, !PT ; /* 0x0000590005037a10 */ /* 0x000fca00007fe4ff */ /*03c0*/ LDG.E R5, [R2.64] ; /* 0x0000000402057981 */ /* 0x000ea2000c1e1900 */ /*03d0*/ MUFU.RCP R7, R0 ; /* 0x0000000000077308 */ /* 0x001e220000001000 */ /*03e0*/ BSSY B0, 0x4d0 ; /* 0x000000e000007945 */ /* 0x000fe20003800000 */ /*03f0*/ FFMA R4, R7, -R0, 1 ; /* 0x3f80000007047423 */ /* 0x001fc80000000800 */ /*0400*/ FFMA R4, R7, R4, R7 ; /* 0x0000000407047223 */ /* 0x000fe40000000007 */ /*0410*/ FMUL R5, R5, R0 ; /* 0x0000000005057220 */ /* 0x004fcc0000400000 */ /*0420*/ F2I.TRUNC.NTZ R5, R5 ; /* 0x0000000500057305 */ /* 0x000e30000020f100 */ /*0430*/ I2F R9, R5 ; /* 0x0000000500097306 */ /* 0x001e300000201400 */ /*0440*/ FCHK P0, R9, R0 ; /* 0x0000000009007302 */ /* 0x001e220000000000 */ /*0450*/ FFMA R7, R9, R4, RZ ; /* 0x0000000409077223 */ /* 0x000fc800000000ff */ /*0460*/ FFMA R6, R7, -R0, R9 ; /* 0x8000000007067223 */ /* 0x000fc80000000009 */ /*0470*/ FFMA R7, R4, R6, R7 ; /* 0x0000000604077223 */ /* 0x000fe20000000007 */ /*0480*/ @!P0 BRA 0x4c0 ; /* 0x0000003000008947 */ /* 0x001fea0003800000 */ /*0490*/ IMAD.MOV.U32 R6, RZ, RZ, R0 ; /* 0x000000ffff067224 */ /* 0x000fe200078e0000 */ /*04a0*/ MOV R0, 0x4c0 ; /* 0x000004c000007802 */ /* 0x000fe40000000f00 */ /*04b0*/ CALL.REL.NOINC 0x4f0 ; /* 0x0000003000007944 */ /* 0x000fea0003c00000 */ /*04c0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*04d0*/ STG.E [R2.64], R7 ; /* 0x0000000702007986 */ /* 0x000fe2000c101904 */ /*04e0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*04f0*/ SHF.R.U32.HI R5, RZ, 0x17, R6 ; /* 0x00000017ff057819 */ /* 0x000fe20000011606 */ /*0500*/ BSSY B1, 0xb50 ; /* 0x0000064000017945 */ /* 0x000fe20003800000 */ /*0510*/ SHF.R.U32.HI R4, RZ, 0x17, R9.reuse ; /* 0x00000017ff047819 */ /* 0x100fe40000011609 */ /*0520*/ LOP3.LUT R13, R5, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff050d7812 */ /* 0x000fe200078ec0ff */ /*0530*/ IMAD.MOV.U32 R5, RZ, RZ, R9 ; /* 0x000000ffff057224 */ /* 0x000fe200078e0009 */ /*0540*/ LOP3.LUT R10, R4, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff040a7812 */ /* 0x000fe400078ec0ff */ /*0550*/ IADD3 R11, R13, -0x1, RZ ; /* 0xffffffff0d0b7810 */ /* 0x000fc40007ffe0ff */ /*0560*/ IADD3 R8, R10, -0x1, RZ ; /* 0xffffffff0a087810 */ /* 0x000fe40007ffe0ff */ /*0570*/ ISETP.GT.U32.AND P0, PT, R11, 0xfd, PT ; /* 0x000000fd0b00780c */ /* 0x000fc80003f04070 */ /*0580*/ ISETP.GT.U32.OR P0, PT, R8, 0xfd, P0 ; /* 0x000000fd0800780c */ /* 0x000fda0000704470 */ /*0590*/ @!P0 IMAD.MOV.U32 R7, RZ, RZ, RZ ; /* 0x000000ffff078224 */ /* 0x000fe200078e00ff */ /*05a0*/ @!P0 BRA 0x730 ; /* 0x0000018000008947 */ /* 0x000fea0003800000 */ /*05b0*/ FSETP.GTU.FTZ.AND P1, PT, |R6|, +INF , PT ; /* 0x7f8000000600780b */ /* 0x000fe20003f3c200 */ /*05c0*/ IMAD.MOV.U32 R4, RZ, RZ, R9 ; /* 0x000000ffff047224 */ /* 0x000fe200078e0009 */ /*05d0*/ FSETP.GTU.FTZ.AND P0, PT, |R9|, +INF , PT ; /* 0x7f8000000900780b */ /* 0x000fc80003f1c200 */ /*05e0*/ PLOP3.LUT P0, PT, P0, P1, PT, 0xa8, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000703570 */ /*05f0*/ @P0 BRA 0xb30 ; /* 0x0000053000000947 */ /* 0x000fea0003800000 */ /*0600*/ LOP3.LUT P0, RZ, R6, 0x7fffffff, R5, 0xc8, !PT ; /* 0x7fffffff06ff7812 */ /* 0x000fda000780c805 */ /*0610*/ @!P0 BRA 0xb10 ; /* 0x000004f000008947 */ /* 0x000fea0003800000 */ /*0620*/ FSETP.NEU.FTZ.AND P2, PT, |R4|, +INF , PT ; /* 0x7f8000000400780b */ /* 0x000fe40003f5d200 */ /*0630*/ FSETP.NEU.FTZ.AND P1, PT, |R6|, +INF , PT ; /* 0x7f8000000600780b */ /* 0x000fe40003f3d200 */ /*0640*/ FSETP.NEU.FTZ.AND P0, PT, |R4|, +INF , PT ; /* 0x7f8000000400780b */ /* 0x000fd60003f1d200 */ /*0650*/ @!P1 BRA !P2, 0xb10 ; /* 0x000004b000009947 */ /* 0x000fea0005000000 */ /*0660*/ LOP3.LUT P2, RZ, R5, 0x7fffffff, RZ, 0xc0, !PT ; /* 0x7fffffff05ff7812 */ /* 0x000fc8000784c0ff */ /*0670*/ PLOP3.LUT P1, PT, P1, P2, PT, 0x2a, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000f24572 */ /*0680*/ @P1 BRA 0xaf0 ; /* 0x0000046000001947 */ /* 0x000fea0003800000 */ /*0690*/ LOP3.LUT P1, RZ, R6, 0x7fffffff, RZ, 0xc0, !PT ; /* 0x7fffffff06ff7812 */ /* 0x000fc8000782c0ff */ /*06a0*/ PLOP3.LUT P0, PT, P0, P1, PT, 0x2a, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000702572 */ /*06b0*/ @P0 BRA 0xac0 ; /* 0x0000040000000947 */ /* 0x000fea0003800000 */ /*06c0*/ ISETP.GE.AND P0, PT, R8, RZ, PT ; /* 0x000000ff0800720c */ /* 0x000fe40003f06270 */ /*06d0*/ ISETP.GE.AND P1, PT, R11, RZ, PT ; /* 0x000000ff0b00720c */ /* 0x000fd60003f26270 */ /*06e0*/ @P0 IMAD.MOV.U32 R7, RZ, RZ, RZ ; /* 0x000000ffff070224 */ /* 0x000fe400078e00ff */ /*06f0*/ @!P0 IMAD.MOV.U32 R7, RZ, RZ, -0x40 ; /* 0xffffffc0ff078424 */ /* 0x000fe400078e00ff */ /*0700*/ @!P0 FFMA R5, R4, 1.84467440737095516160e+19, RZ ; /* 0x5f80000004058823 */ /* 0x000fe400000000ff */ /*0710*/ @!P1 FFMA R6, R6, 1.84467440737095516160e+19, RZ ; /* 0x5f80000006069823 */ /* 0x000fe200000000ff */ /*0720*/ @!P1 IADD3 R7, R7, 0x40, RZ ; /* 0x0000004007079810 */ /* 0x000fe40007ffe0ff */ /*0730*/ LEA R9, R13, 0xc0800000, 0x17 ; /* 0xc08000000d097811 */ /* 0x000fe200078eb8ff */ /*0740*/ BSSY B2, 0xab0 ; /* 0x0000036000027945 */ /* 0x000fe80003800000 */ /*0750*/ IMAD.IADD R9, R6, 0x1, -R9 ; /* 0x0000000106097824 */ /* 0x000fe200078e0a09 */ /*0760*/ IADD3 R6, R10, -0x7f, RZ ; /* 0xffffff810a067810 */ /* 0x000fc60007ffe0ff */ /*0770*/ MUFU.RCP R4, R9 ; /* 0x0000000900047308 */ /* 0x000e220000001000 */ /*0780*/ FADD.FTZ R8, -R9, -RZ ; /* 0x800000ff09087221 */ /* 0x000fe40000010100 */ /*0790*/ IMAD R5, R6.reuse, -0x800000, R5 ; /* 0xff80000006057824 */ /* 0x040fe200078e0205 */ /*07a0*/ IADD3 R6, R6, 0x7f, -R13 ; /* 0x0000007f06067810 */ /* 0x000fca0007ffe80d */ /*07b0*/ IMAD.IADD R6, R6, 0x1, R7 ; /* 0x0000000106067824 */ /* 0x000fe400078e0207 */ /*07c0*/ FFMA R11, R4, R8, 1 ; /* 0x3f800000040b7423 */ /* 0x001fc80000000008 */ /*07d0*/ FFMA R10, R4, R11, R4 ; /* 0x0000000b040a7223 */ /* 0x000fc80000000004 */ /*07e0*/ FFMA R4, R5, R10, RZ ; /* 0x0000000a05047223 */ /* 0x000fc800000000ff */ /*07f0*/ FFMA R11, R8, R4, R5 ; /* 0x00000004080b7223 */ /* 0x000fc80000000005 */ /*0800*/ FFMA R11, R10, R11, R4 ; /* 0x0000000b0a0b7223 */ /* 0x000fc80000000004 */ /*0810*/ FFMA R8, R8, R11, R5 ; /* 0x0000000b08087223 */ /* 0x000fc80000000005 */ /*0820*/ FFMA R4, R10, R8, R11 ; /* 0x000000080a047223 */ /* 0x000fca000000000b */ /*0830*/ SHF.R.U32.HI R5, RZ, 0x17, R4 ; /* 0x00000017ff057819 */ /* 0x000fc80000011604 */ /*0840*/ LOP3.LUT R5, R5, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff05057812 */ /* 0x000fca00078ec0ff */ /*0850*/ IMAD.IADD R9, R5, 0x1, R6 ; /* 0x0000000105097824 */ /* 0x000fca00078e0206 */ /*0860*/ IADD3 R5, R9, -0x1, RZ ; /* 0xffffffff09057810 */ /* 0x000fc80007ffe0ff */ /*0870*/ ISETP.GE.U32.AND P0, PT, R5, 0xfe, PT ; /* 0x000000fe0500780c */ /* 0x000fda0003f06070 */ /*0880*/ @!P0 BRA 0xa90 ; /* 0x0000020000008947 */ /* 0x000fea0003800000 */ /*0890*/ ISETP.GT.AND P0, PT, R9, 0xfe, PT ; /* 0x000000fe0900780c */ /* 0x000fda0003f04270 */ /*08a0*/ @P0 BRA 0xa60 ; /* 0x000001b000000947 */ /* 0x000fea0003800000 */ /*08b0*/ ISETP.GE.AND P0, PT, R9, 0x1, PT ; /* 0x000000010900780c */ /* 0x000fda0003f06270 */ /*08c0*/ @P0 BRA 0xaa0 ; /* 0x000001d000000947 */ /* 0x000fea0003800000 */ /*08d0*/ ISETP.GE.AND P0, PT, R9, -0x18, PT ; /* 0xffffffe80900780c */ /* 0x000fe40003f06270 */ /*08e0*/ LOP3.LUT R4, R4, 0x80000000, RZ, 0xc0, !PT ; /* 0x8000000004047812 */ /* 0x000fd600078ec0ff */ /*08f0*/ @!P0 BRA 0xaa0 ; /* 0x000001a000008947 */ /* 0x000fea0003800000 */ /*0900*/ FFMA.RZ R5, R10.reuse, R8.reuse, R11.reuse ; /* 0x000000080a057223 */ /* 0x1c0fe2000000c00b */ /*0910*/ ISETP.NE.AND P2, PT, R9.reuse, RZ, PT ; /* 0x000000ff0900720c */ /* 0x040fe20003f45270 */ /*0920*/ FFMA.RM R6, R10.reuse, R8.reuse, R11.reuse ; /* 0x000000080a067223 */ /* 0x1c0fe2000000400b */ /*0930*/ ISETP.NE.AND P1, PT, R9, RZ, PT ; /* 0x000000ff0900720c */ /* 0x000fe40003f25270 */ /*0940*/ LOP3.LUT R7, R5, 0x7fffff, RZ, 0xc0, !PT ; /* 0x007fffff05077812 */ /* 0x000fe200078ec0ff */ /*0950*/ FFMA.RP R5, R10, R8, R11 ; /* 0x000000080a057223 */ /* 0x000fe2000000800b */ /*0960*/ IADD3 R8, R9, 0x20, RZ ; /* 0x0000002009087810 */ /* 0x000fe20007ffe0ff */ /*0970*/ IMAD.MOV R9, RZ, RZ, -R9 ; /* 0x000000ffff097224 */ /* 0x000fe200078e0a09 */ /*0980*/ LOP3.LUT R7, R7, 0x800000, RZ, 0xfc, !PT ; /* 0x0080000007077812 */ /* 0x000fe400078efcff */ /*0990*/ FSETP.NEU.FTZ.AND P0, PT, R5, R6, PT ; /* 0x000000060500720b */ /* 0x000fc40003f1d000 */ /*09a0*/ SHF.L.U32 R8, R7, R8, RZ ; /* 0x0000000807087219 */ /* 0x000fe400000006ff */ /*09b0*/ SEL R6, R9, RZ, P2 ; /* 0x000000ff09067207 */ /* 0x000fe40001000000 */ /*09c0*/ ISETP.NE.AND P1, PT, R8, RZ, P1 ; /* 0x000000ff0800720c */ /* 0x000fe40000f25270 */ /*09d0*/ SHF.R.U32.HI R6, RZ, R6, R7 ; /* 0x00000006ff067219 */ /* 0x000fe40000011607 */ /*09e0*/ PLOP3.LUT P0, PT, P0, P1, PT, 0xa8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe40000703570 */ /*09f0*/ SHF.R.U32.HI R8, RZ, 0x1, R6 ; /* 0x00000001ff087819 */ /* 0x000fc40000011606 */ /*0a00*/ SEL R5, RZ, 0x1, !P0 ; /* 0x00000001ff057807 */ /* 0x000fc80004000000 */ /*0a10*/ LOP3.LUT R5, R5, 0x1, R8, 0xf8, !PT ; /* 0x0000000105057812 */ /* 0x000fc800078ef808 */ /*0a20*/ LOP3.LUT R5, R5, R6, RZ, 0xc0, !PT ; /* 0x0000000605057212 */ /* 0x000fca00078ec0ff */ /*0a30*/ IMAD.IADD R5, R8, 0x1, R5 ; /* 0x0000000108057824 */ /* 0x000fca00078e0205 */ /*0a40*/ LOP3.LUT R4, R5, R4, RZ, 0xfc, !PT ; /* 0x0000000405047212 */ /* 0x000fe200078efcff */ /*0a50*/ BRA 0xaa0 ; /* 0x0000004000007947 */ /* 0x000fea0003800000 */ /*0a60*/ LOP3.LUT R4, R4, 0x80000000, RZ, 0xc0, !PT ; /* 0x8000000004047812 */ /* 0x000fc800078ec0ff */ /*0a70*/ LOP3.LUT R4, R4, 0x7f800000, RZ, 0xfc, !PT ; /* 0x7f80000004047812 */ /* 0x000fe200078efcff */ /*0a80*/ BRA 0xaa0 ; /* 0x0000001000007947 */ /* 0x000fea0003800000 */ /*0a90*/ IMAD R4, R6, 0x800000, R4 ; /* 0x0080000006047824 */ /* 0x000fe400078e0204 */ /*0aa0*/ BSYNC B2 ; /* 0x0000000000027941 */ /* 0x000fea0003800000 */ /*0ab0*/ BRA 0xb40 ; /* 0x0000008000007947 */ /* 0x000fea0003800000 */ /*0ac0*/ LOP3.LUT R4, R6, 0x80000000, R5, 0x48, !PT ; /* 0x8000000006047812 */ /* 0x000fc800078e4805 */ /*0ad0*/ LOP3.LUT R4, R4, 0x7f800000, RZ, 0xfc, !PT ; /* 0x7f80000004047812 */ /* 0x000fe200078efcff */ /*0ae0*/ BRA 0xb40 ; /* 0x0000005000007947 */ /* 0x000fea0003800000 */ /*0af0*/ LOP3.LUT R4, R6, 0x80000000, R5, 0x48, !PT ; /* 0x8000000006047812 */ /* 0x000fe200078e4805 */ /*0b00*/ BRA 0xb40 ; /* 0x0000003000007947 */ /* 0x000fea0003800000 */ /*0b10*/ MUFU.RSQ R4, -QNAN ; /* 0xffc0000000047908 */ /* 0x000e220000001400 */ /*0b20*/ BRA 0xb40 ; /* 0x0000001000007947 */ /* 0x000fea0003800000 */ /*0b30*/ FADD.FTZ R4, R4, R6 ; /* 0x0000000604047221 */ /* 0x000fe40000010000 */ /*0b40*/ BSYNC B1 ; /* 0x0000000000017941 */ /* 0x000fea0003800000 */ /*0b50*/ IMAD.MOV.U32 R7, RZ, RZ, R4 ; /* 0x000000ffff077224 */ /* 0x001fe400078e0004 */ /*0b60*/ IMAD.MOV.U32 R4, RZ, RZ, R0 ; /* 0x000000ffff047224 */ /* 0x000fe400078e0000 */ /*0b70*/ IMAD.MOV.U32 R5, RZ, RZ, 0x0 ; /* 0x00000000ff057424 */ /* 0x000fc800078e00ff */ /*0b80*/ RET.REL.NODEC R4 0x0 ; /* 0xfffff47004007950 */ /* 0x000fea0003c3ffff */ /*0b90*/ BRA 0xb90; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0ba0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0bb0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0bc0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0bd0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0be0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0bf0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c00*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c10*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c20*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c30*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c40*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c50*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c60*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c70*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include "includes.h" __global__ void _setPrecisionKernel(float* data, size_t size, int* precision) { unsigned int idx = blockDim.x * blockIdx.x + threadIdx.x; if (idx >= size) return; int prec = precision[idx]; int mul = 1; while(prec--) mul *= 10; data[idx] = (float)(int)(data[idx]*mul); data[idx] /= mul; }
.file "tmpxft_0016fd16_00000000-6__setPrecisionKernel.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2029: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2029: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi .type _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi, @function _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi: .LFB2051: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movq %fs:40, %rax movq %rax, 120(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 120(%rsp), %rax subq %fs:40, %rax jne .L8 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 152 pushq 40(%rsp) .cfi_def_cfa_offset 160 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z19_setPrecisionKernelPfmPi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2051: .size _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi, .-_Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi .globl _Z19_setPrecisionKernelPfmPi .type _Z19_setPrecisionKernelPfmPi, @function _Z19_setPrecisionKernelPfmPi: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z19_setPrecisionKernelPfmPi, .-_Z19_setPrecisionKernelPfmPi .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z19_setPrecisionKernelPfmPi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2054: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z19_setPrecisionKernelPfmPi(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2054: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include "includes.h" __global__ void _setPrecisionKernel(float* data, size_t size, int* precision) { unsigned int idx = blockDim.x * blockIdx.x + threadIdx.x; if (idx >= size) return; int prec = precision[idx]; int mul = 1; while(prec--) mul *= 10; data[idx] = (float)(int)(data[idx]*mul); data[idx] /= mul; }
#include <hip/hip_runtime.h> #include "includes.h" __global__ void _setPrecisionKernel(float* data, size_t size, int* precision) { unsigned int idx = blockDim.x * blockIdx.x + threadIdx.x; if (idx >= size) return; int prec = precision[idx]; int mul = 1; while(prec--) mul *= 10; data[idx] = (float)(int)(data[idx]*mul); data[idx] /= mul; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> #include "includes.h" __global__ void _setPrecisionKernel(float* data, size_t size, int* precision) { unsigned int idx = blockDim.x * blockIdx.x + threadIdx.x; if (idx >= size) return; int prec = precision[idx]; int mul = 1; while(prec--) mul *= 10; data[idx] = (float)(int)(data[idx]*mul); data[idx] /= mul; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z19_setPrecisionKernelPfmPi .globl _Z19_setPrecisionKernelPfmPi .p2align 8 .type _Z19_setPrecisionKernelPfmPi,@function _Z19_setPrecisionKernelPfmPi: s_clause 0x1 s_load_b32 s4, s[0:1], 0x24 s_load_b64 s[2:3], s[0:1], 0x8 s_waitcnt lgkmcnt(0) s_and_b32 s4, s4, 0xffff s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_mad_u64_u32 v[1:2], null, s15, s4, v[0:1] v_mov_b32_e32 v2, 0 v_cmp_gt_u64_e32 vcc_lo, s[2:3], v[1:2] s_and_saveexec_b32 s2, vcc_lo s_cbranch_execz .LBB0_6 s_load_b64 s[2:3], s[0:1], 0x10 v_lshlrev_b64 v[3:4], 2, v[1:2] v_mov_b32_e32 v0, 1.0 s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v3, vcc_lo, s2, v3 v_add_co_ci_u32_e32 v4, vcc_lo, s3, v4, vcc_lo s_mov_b32 s3, 0 s_mov_b32 s2, exec_lo global_load_b32 v3, v[3:4], off s_waitcnt vmcnt(0) v_cmpx_ne_u32_e32 0, v3 s_cbranch_execz .LBB0_5 s_mov_b32 s4, 1 .LBB0_3: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_mul_i32 s4, s4, 10 v_dual_mov_b32 v0, s4 :: v_dual_add_nc_u32 v3, -1, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_cmp_eq_u32_e32 vcc_lo, 0, v3 s_or_b32 s3, vcc_lo, s3 s_and_not1_b32 exec_lo, exec_lo, s3 s_cbranch_execnz .LBB0_3 s_or_b32 exec_lo, exec_lo, s3 v_cvt_f32_i32_e32 v0, v0 .LBB0_5: s_or_b32 exec_lo, exec_lo, s2 s_load_b64 s[0:1], s[0:1], 0x0 v_lshlrev_b64 v[1:2], 2, v[1:2] s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v1, vcc_lo, s0, v1 v_add_co_ci_u32_e32 v2, vcc_lo, s1, v2, vcc_lo global_load_b32 v3, v[1:2], off s_waitcnt vmcnt(0) v_mul_f32_e32 v3, v3, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cvt_i32_f32_e32 v3, v3 v_cvt_f32_i32_e32 v3, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_div_scale_f32 v4, null, v0, v0, v3 v_rcp_f32_e32 v5, v4 s_waitcnt_depctr 0xfff v_fma_f32 v6, -v4, v5, 1.0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_fmac_f32_e32 v5, v6, v5 v_div_scale_f32 v6, vcc_lo, v3, v0, v3 v_mul_f32_e32 v7, v6, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fma_f32 v8, -v4, v7, v6 v_fmac_f32_e32 v7, v8, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fma_f32 v4, -v4, v7, v6 v_div_fmas_f32 v4, v4, v5, v7 s_delay_alu instid0(VALU_DEP_1) v_div_fixup_f32 v0, v4, v0, v3 global_store_b32 v[1:2], v0, off .LBB0_6: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z19_setPrecisionKernelPfmPi .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 280 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 9 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z19_setPrecisionKernelPfmPi, .Lfunc_end0-_Z19_setPrecisionKernelPfmPi .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .offset: 8 .size: 8 .value_kind: by_value - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: hidden_block_count_x - .offset: 28 .size: 4 .value_kind: hidden_block_count_y - .offset: 32 .size: 4 .value_kind: hidden_block_count_z - .offset: 36 .size: 2 .value_kind: hidden_group_size_x - .offset: 38 .size: 2 .value_kind: hidden_group_size_y - .offset: 40 .size: 2 .value_kind: hidden_group_size_z - .offset: 42 .size: 2 .value_kind: hidden_remainder_x - .offset: 44 .size: 2 .value_kind: hidden_remainder_y - .offset: 46 .size: 2 .value_kind: hidden_remainder_z - .offset: 64 .size: 8 .value_kind: hidden_global_offset_x - .offset: 72 .size: 8 .value_kind: hidden_global_offset_y - .offset: 80 .size: 8 .value_kind: hidden_global_offset_z - .offset: 88 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 280 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z19_setPrecisionKernelPfmPi .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z19_setPrecisionKernelPfmPi.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 9 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <hip/hip_runtime.h> #include "includes.h" __global__ void _setPrecisionKernel(float* data, size_t size, int* precision) { unsigned int idx = blockDim.x * blockIdx.x + threadIdx.x; if (idx >= size) return; int prec = precision[idx]; int mul = 1; while(prec--) mul *= 10; data[idx] = (float)(int)(data[idx]*mul); data[idx] /= mul; }
.text .file "_setPrecisionKernel.hip" .globl _Z34__device_stub___setPrecisionKernelPfmPi # -- Begin function _Z34__device_stub___setPrecisionKernelPfmPi .p2align 4, 0x90 .type _Z34__device_stub___setPrecisionKernelPfmPi,@function _Z34__device_stub___setPrecisionKernelPfmPi: # @_Z34__device_stub___setPrecisionKernelPfmPi .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movq %rdx, 56(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z19_setPrecisionKernelPfmPi, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $120, %rsp .cfi_adjust_cfa_offset -120 retq .Lfunc_end0: .size _Z34__device_stub___setPrecisionKernelPfmPi, .Lfunc_end0-_Z34__device_stub___setPrecisionKernelPfmPi .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z19_setPrecisionKernelPfmPi, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type _Z19_setPrecisionKernelPfmPi,@object # @_Z19_setPrecisionKernelPfmPi .section .rodata,"a",@progbits .globl _Z19_setPrecisionKernelPfmPi .p2align 3, 0x0 _Z19_setPrecisionKernelPfmPi: .quad _Z34__device_stub___setPrecisionKernelPfmPi .size _Z19_setPrecisionKernelPfmPi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z19_setPrecisionKernelPfmPi" .size .L__unnamed_1, 29 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z34__device_stub___setPrecisionKernelPfmPi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z19_setPrecisionKernelPfmPi .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z19_setPrecisionKernelPfmPi .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ IMAD.MOV.U32 R1, RZ, RZ, c[0x0][0x28] ; /* 0x00000a00ff017624 */ /* 0x000fe400078e00ff */ /*0010*/ S2R R0, SR_CTAID.X ; /* 0x0000000000007919 */ /* 0x000e280000002500 */ /*0020*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e240000002100 */ /*0030*/ IMAD R0, R0, c[0x0][0x0], R3 ; /* 0x0000000000007a24 */ /* 0x001fca00078e0203 */ /*0040*/ ISETP.GE.U32.AND P0, PT, R0, c[0x0][0x168], PT ; /* 0x00005a0000007a0c */ /* 0x000fc80003f06070 */ /*0050*/ ISETP.GE.U32.AND.EX P0, PT, RZ, c[0x0][0x16c], PT, P0 ; /* 0x00005b00ff007a0c */ /* 0x000fda0003f06100 */ /*0060*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0070*/ IMAD.SHL.U32 R4, R0, 0x4, RZ ; /* 0x0000000400047824 */ /* 0x000fe200078e00ff */ /*0080*/ SHF.R.U32.HI R5, RZ, 0x1e, R0 ; /* 0x0000001eff057819 */ /* 0x000fe20000011600 */ /*0090*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fc60000000a00 */ /*00a0*/ IADD3 R2, P0, R4, c[0x0][0x170], RZ ; /* 0x00005c0004027a10 */ /* 0x000fc80007f1e0ff */ /*00b0*/ IADD3.X R3, R5, c[0x0][0x174], RZ, P0, !PT ; /* 0x00005d0005037a10 */ /* 0x000fca00007fe4ff */ /*00c0*/ LDG.E R2, [R2.64] ; /* 0x0000000402027981 */ /* 0x000ea2000c1e1900 */ /*00d0*/ BSSY B0, 0x3a0 ; /* 0x000002c000007945 */ /* 0x000fe20003800000 */ /*00e0*/ IMAD.MOV.U32 R0, RZ, RZ, 0x3f800000 ; /* 0x3f800000ff007424 */ /* 0x000fe200078e00ff */ /*00f0*/ ISETP.NE.AND P0, PT, R2, RZ, PT ; /* 0x000000ff0200720c */ /* 0x004fda0003f05270 */ /*0100*/ @!P0 BRA 0x390 ; /* 0x0000028000008947 */ /* 0x000fea0003800000 */ /*0110*/ IADD3 R0, R2.reuse, -0x1, RZ ; /* 0xffffffff02007810 */ /* 0x040fe20007ffe0ff */ /*0120*/ BSSY B1, 0x300 ; /* 0x000001d000017945 */ /* 0x000fe20003800000 */ /*0130*/ LOP3.LUT R3, R2, 0x3, RZ, 0xc0, !PT ; /* 0x0000000302037812 */ /* 0x000fe400078ec0ff */ /*0140*/ ISETP.GE.U32.AND P0, PT, R0, 0x3, PT ; /* 0x000000030000780c */ /* 0x000fe20003f06070 */ /*0150*/ IMAD.MOV.U32 R0, RZ, RZ, 0x1 ; /* 0x00000001ff007424 */ /* 0x000fd800078e00ff */ /*0160*/ @!P0 BRA 0x2f0 ; /* 0x0000018000008947 */ /* 0x000fea0003800000 */ /*0170*/ IMAD.IADD R2, R2, 0x1, -R3 ; /* 0x0000000102027824 */ /* 0x000fe400078e0a03 */ /*0180*/ IMAD.MOV.U32 R0, RZ, RZ, 0x1 ; /* 0x00000001ff007424 */ /* 0x000fc600078e00ff */ /*0190*/ ISETP.GT.AND P0, PT, R2, RZ, PT ; /* 0x000000ff0200720c */ /* 0x000fda0003f04270 */ /*01a0*/ @!P0 BRA 0x2b0 ; /* 0x0000010000008947 */ /* 0x000fea0003800000 */ /*01b0*/ ISETP.GT.AND P1, PT, R2, 0xc, PT ; /* 0x0000000c0200780c */ /* 0x000fe20003f24270 */ /*01c0*/ BSSY B2, 0x250 ; /* 0x0000008000027945 */ /* 0x000fe20003800000 */ /*01d0*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x80, 0x0 ; /* 0x000000000000781c */ /* 0x000fd60003f0f070 */ /*01e0*/ @!P1 BRA 0x240 ; /* 0x0000005000009947 */ /* 0x000fea0003800000 */ /*01f0*/ PLOP3.LUT P0, PT, PT, PT, PT, 0x8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe40003f0e170 */ /*0200*/ IADD3 R2, R2, -0x10, RZ ; /* 0xfffffff002027810 */ /* 0x000fe20007ffe0ff */ /*0210*/ IMAD R0, R0, 0x6fc10000, RZ ; /* 0x6fc1000000007824 */ /* 0x000fc600078e02ff */ /*0220*/ ISETP.GT.AND P1, PT, R2, 0xc, PT ; /* 0x0000000c0200780c */ /* 0x000fda0003f24270 */ /*0230*/ @P1 BRA 0x200 ; /* 0xffffffc000001947 */ /* 0x000fea000383ffff */ /*0240*/ BSYNC B2 ; /* 0x0000000000027941 */ /* 0x000fea0003800000 */ /*0250*/ ISETP.GT.AND P1, PT, R2, 0x4, PT ; /* 0x000000040200780c */ /* 0x000fda0003f24270 */ /*0260*/ @P1 PLOP3.LUT P0, PT, PT, PT, PT, 0x8, 0x0 ; /* 0x000000000000181c */ /* 0x000fe20003f0e170 */ /*0270*/ @P1 IMAD R0, R0, 0x5f5e100, RZ ; /* 0x05f5e10000001824 */ /* 0x000fe200078e02ff */ /*0280*/ @P1 IADD3 R2, R2, -0x8, RZ ; /* 0xfffffff802021810 */ /* 0x000fd60007ffe0ff */ /*0290*/ ISETP.NE.OR P0, PT, R2, RZ, P0 ; /* 0x000000ff0200720c */ /* 0x000fda0000705670 */ /*02a0*/ @!P0 BRA 0x2f0 ; /* 0x0000004000008947 */ /* 0x000fea0003800000 */ /*02b0*/ IADD3 R2, R2, -0x4, RZ ; /* 0xfffffffc02027810 */ /* 0x000fe20007ffe0ff */ /*02c0*/ IMAD R0, R0, 0x2710, RZ ; /* 0x0000271000007824 */ /* 0x000fc600078e02ff */ /*02d0*/ ISETP.NE.AND P0, PT, R2, RZ, PT ; /* 0x000000ff0200720c */ /* 0x000fda0003f05270 */ /*02e0*/ @P0 BRA 0x2b0 ; /* 0xffffffc000000947 */ /* 0x000fea000383ffff */ /*02f0*/ BSYNC B1 ; /* 0x0000000000017941 */ /* 0x000fea0003800000 */ /*0300*/ ISETP.NE.AND P0, PT, R3, RZ, PT ; /* 0x000000ff0300720c */ /* 0x000fe20003f05270 */ /*0310*/ BSSY B1, 0x380 ; /* 0x0000006000017945 */ /* 0x000fd80003800000 */ /*0320*/ @!P0 BRA 0x370 ; /* 0x0000004000008947 */ /* 0x000fea0003800000 */ /*0330*/ IADD3 R3, R3, -0x1, RZ ; /* 0xffffffff03037810 */ /* 0x000fe20007ffe0ff */ /*0340*/ IMAD R0, R0, 0xa, RZ ; /* 0x0000000a00007824 */ /* 0x000fc600078e02ff */ /*0350*/ ISETP.NE.AND P0, PT, R3, RZ, PT ; /* 0x000000ff0300720c */ /* 0x000fda0003f05270 */ /*0360*/ @P0 BRA 0x330 ; /* 0xffffffc000000947 */ /* 0x000fea000383ffff */ /*0370*/ BSYNC B1 ; /* 0x0000000000017941 */ /* 0x000fea0003800000 */ /*0380*/ I2F R0, R0 ; /* 0x0000000000007306 */ /* 0x000e240000201400 */ /*0390*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*03a0*/ IADD3 R2, P0, R4, c[0x0][0x160], RZ ; /* 0x0000580004027a10 */ /* 0x000fc80007f1e0ff */ /*03b0*/ IADD3.X R3, R5, c[0x0][0x164], RZ, P0, !PT ; /* 0x0000590005037a10 */ /* 0x000fca00007fe4ff */ /*03c0*/ LDG.E R5, [R2.64] ; /* 0x0000000402057981 */ /* 0x000ea2000c1e1900 */ /*03d0*/ MUFU.RCP R7, R0 ; /* 0x0000000000077308 */ /* 0x001e220000001000 */ /*03e0*/ BSSY B0, 0x4d0 ; /* 0x000000e000007945 */ /* 0x000fe20003800000 */ /*03f0*/ FFMA R4, R7, -R0, 1 ; /* 0x3f80000007047423 */ /* 0x001fc80000000800 */ /*0400*/ FFMA R4, R7, R4, R7 ; /* 0x0000000407047223 */ /* 0x000fe40000000007 */ /*0410*/ FMUL R5, R5, R0 ; /* 0x0000000005057220 */ /* 0x004fcc0000400000 */ /*0420*/ F2I.TRUNC.NTZ R5, R5 ; /* 0x0000000500057305 */ /* 0x000e30000020f100 */ /*0430*/ I2F R9, R5 ; /* 0x0000000500097306 */ /* 0x001e300000201400 */ /*0440*/ FCHK P0, R9, R0 ; /* 0x0000000009007302 */ /* 0x001e220000000000 */ /*0450*/ FFMA R7, R9, R4, RZ ; /* 0x0000000409077223 */ /* 0x000fc800000000ff */ /*0460*/ FFMA R6, R7, -R0, R9 ; /* 0x8000000007067223 */ /* 0x000fc80000000009 */ /*0470*/ FFMA R7, R4, R6, R7 ; /* 0x0000000604077223 */ /* 0x000fe20000000007 */ /*0480*/ @!P0 BRA 0x4c0 ; /* 0x0000003000008947 */ /* 0x001fea0003800000 */ /*0490*/ IMAD.MOV.U32 R6, RZ, RZ, R0 ; /* 0x000000ffff067224 */ /* 0x000fe200078e0000 */ /*04a0*/ MOV R0, 0x4c0 ; /* 0x000004c000007802 */ /* 0x000fe40000000f00 */ /*04b0*/ CALL.REL.NOINC 0x4f0 ; /* 0x0000003000007944 */ /* 0x000fea0003c00000 */ /*04c0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*04d0*/ STG.E [R2.64], R7 ; /* 0x0000000702007986 */ /* 0x000fe2000c101904 */ /*04e0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*04f0*/ SHF.R.U32.HI R5, RZ, 0x17, R6 ; /* 0x00000017ff057819 */ /* 0x000fe20000011606 */ /*0500*/ BSSY B1, 0xb50 ; /* 0x0000064000017945 */ /* 0x000fe20003800000 */ /*0510*/ SHF.R.U32.HI R4, RZ, 0x17, R9.reuse ; /* 0x00000017ff047819 */ /* 0x100fe40000011609 */ /*0520*/ LOP3.LUT R13, R5, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff050d7812 */ /* 0x000fe200078ec0ff */ /*0530*/ IMAD.MOV.U32 R5, RZ, RZ, R9 ; /* 0x000000ffff057224 */ /* 0x000fe200078e0009 */ /*0540*/ LOP3.LUT R10, R4, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff040a7812 */ /* 0x000fe400078ec0ff */ /*0550*/ IADD3 R11, R13, -0x1, RZ ; /* 0xffffffff0d0b7810 */ /* 0x000fc40007ffe0ff */ /*0560*/ IADD3 R8, R10, -0x1, RZ ; /* 0xffffffff0a087810 */ /* 0x000fe40007ffe0ff */ /*0570*/ ISETP.GT.U32.AND P0, PT, R11, 0xfd, PT ; /* 0x000000fd0b00780c */ /* 0x000fc80003f04070 */ /*0580*/ ISETP.GT.U32.OR P0, PT, R8, 0xfd, P0 ; /* 0x000000fd0800780c */ /* 0x000fda0000704470 */ /*0590*/ @!P0 IMAD.MOV.U32 R7, RZ, RZ, RZ ; /* 0x000000ffff078224 */ /* 0x000fe200078e00ff */ /*05a0*/ @!P0 BRA 0x730 ; /* 0x0000018000008947 */ /* 0x000fea0003800000 */ /*05b0*/ FSETP.GTU.FTZ.AND P1, PT, |R6|, +INF , PT ; /* 0x7f8000000600780b */ /* 0x000fe20003f3c200 */ /*05c0*/ IMAD.MOV.U32 R4, RZ, RZ, R9 ; /* 0x000000ffff047224 */ /* 0x000fe200078e0009 */ /*05d0*/ FSETP.GTU.FTZ.AND P0, PT, |R9|, +INF , PT ; /* 0x7f8000000900780b */ /* 0x000fc80003f1c200 */ /*05e0*/ PLOP3.LUT P0, PT, P0, P1, PT, 0xa8, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000703570 */ /*05f0*/ @P0 BRA 0xb30 ; /* 0x0000053000000947 */ /* 0x000fea0003800000 */ /*0600*/ LOP3.LUT P0, RZ, R6, 0x7fffffff, R5, 0xc8, !PT ; /* 0x7fffffff06ff7812 */ /* 0x000fda000780c805 */ /*0610*/ @!P0 BRA 0xb10 ; /* 0x000004f000008947 */ /* 0x000fea0003800000 */ /*0620*/ FSETP.NEU.FTZ.AND P2, PT, |R4|, +INF , PT ; /* 0x7f8000000400780b */ /* 0x000fe40003f5d200 */ /*0630*/ FSETP.NEU.FTZ.AND P1, PT, |R6|, +INF , PT ; /* 0x7f8000000600780b */ /* 0x000fe40003f3d200 */ /*0640*/ FSETP.NEU.FTZ.AND P0, PT, |R4|, +INF , PT ; /* 0x7f8000000400780b */ /* 0x000fd60003f1d200 */ /*0650*/ @!P1 BRA !P2, 0xb10 ; /* 0x000004b000009947 */ /* 0x000fea0005000000 */ /*0660*/ LOP3.LUT P2, RZ, R5, 0x7fffffff, RZ, 0xc0, !PT ; /* 0x7fffffff05ff7812 */ /* 0x000fc8000784c0ff */ /*0670*/ PLOP3.LUT P1, PT, P1, P2, PT, 0x2a, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000f24572 */ /*0680*/ @P1 BRA 0xaf0 ; /* 0x0000046000001947 */ /* 0x000fea0003800000 */ /*0690*/ LOP3.LUT P1, RZ, R6, 0x7fffffff, RZ, 0xc0, !PT ; /* 0x7fffffff06ff7812 */ /* 0x000fc8000782c0ff */ /*06a0*/ PLOP3.LUT P0, PT, P0, P1, PT, 0x2a, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000702572 */ /*06b0*/ @P0 BRA 0xac0 ; /* 0x0000040000000947 */ /* 0x000fea0003800000 */ /*06c0*/ ISETP.GE.AND P0, PT, R8, RZ, PT ; /* 0x000000ff0800720c */ /* 0x000fe40003f06270 */ /*06d0*/ ISETP.GE.AND P1, PT, R11, RZ, PT ; /* 0x000000ff0b00720c */ /* 0x000fd60003f26270 */ /*06e0*/ @P0 IMAD.MOV.U32 R7, RZ, RZ, RZ ; /* 0x000000ffff070224 */ /* 0x000fe400078e00ff */ /*06f0*/ @!P0 IMAD.MOV.U32 R7, RZ, RZ, -0x40 ; /* 0xffffffc0ff078424 */ /* 0x000fe400078e00ff */ /*0700*/ @!P0 FFMA R5, R4, 1.84467440737095516160e+19, RZ ; /* 0x5f80000004058823 */ /* 0x000fe400000000ff */ /*0710*/ @!P1 FFMA R6, R6, 1.84467440737095516160e+19, RZ ; /* 0x5f80000006069823 */ /* 0x000fe200000000ff */ /*0720*/ @!P1 IADD3 R7, R7, 0x40, RZ ; /* 0x0000004007079810 */ /* 0x000fe40007ffe0ff */ /*0730*/ LEA R9, R13, 0xc0800000, 0x17 ; /* 0xc08000000d097811 */ /* 0x000fe200078eb8ff */ /*0740*/ BSSY B2, 0xab0 ; /* 0x0000036000027945 */ /* 0x000fe80003800000 */ /*0750*/ IMAD.IADD R9, R6, 0x1, -R9 ; /* 0x0000000106097824 */ /* 0x000fe200078e0a09 */ /*0760*/ IADD3 R6, R10, -0x7f, RZ ; /* 0xffffff810a067810 */ /* 0x000fc60007ffe0ff */ /*0770*/ MUFU.RCP R4, R9 ; /* 0x0000000900047308 */ /* 0x000e220000001000 */ /*0780*/ FADD.FTZ R8, -R9, -RZ ; /* 0x800000ff09087221 */ /* 0x000fe40000010100 */ /*0790*/ IMAD R5, R6.reuse, -0x800000, R5 ; /* 0xff80000006057824 */ /* 0x040fe200078e0205 */ /*07a0*/ IADD3 R6, R6, 0x7f, -R13 ; /* 0x0000007f06067810 */ /* 0x000fca0007ffe80d */ /*07b0*/ IMAD.IADD R6, R6, 0x1, R7 ; /* 0x0000000106067824 */ /* 0x000fe400078e0207 */ /*07c0*/ FFMA R11, R4, R8, 1 ; /* 0x3f800000040b7423 */ /* 0x001fc80000000008 */ /*07d0*/ FFMA R10, R4, R11, R4 ; /* 0x0000000b040a7223 */ /* 0x000fc80000000004 */ /*07e0*/ FFMA R4, R5, R10, RZ ; /* 0x0000000a05047223 */ /* 0x000fc800000000ff */ /*07f0*/ FFMA R11, R8, R4, R5 ; /* 0x00000004080b7223 */ /* 0x000fc80000000005 */ /*0800*/ FFMA R11, R10, R11, R4 ; /* 0x0000000b0a0b7223 */ /* 0x000fc80000000004 */ /*0810*/ FFMA R8, R8, R11, R5 ; /* 0x0000000b08087223 */ /* 0x000fc80000000005 */ /*0820*/ FFMA R4, R10, R8, R11 ; /* 0x000000080a047223 */ /* 0x000fca000000000b */ /*0830*/ SHF.R.U32.HI R5, RZ, 0x17, R4 ; /* 0x00000017ff057819 */ /* 0x000fc80000011604 */ /*0840*/ LOP3.LUT R5, R5, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff05057812 */ /* 0x000fca00078ec0ff */ /*0850*/ IMAD.IADD R9, R5, 0x1, R6 ; /* 0x0000000105097824 */ /* 0x000fca00078e0206 */ /*0860*/ IADD3 R5, R9, -0x1, RZ ; /* 0xffffffff09057810 */ /* 0x000fc80007ffe0ff */ /*0870*/ ISETP.GE.U32.AND P0, PT, R5, 0xfe, PT ; /* 0x000000fe0500780c */ /* 0x000fda0003f06070 */ /*0880*/ @!P0 BRA 0xa90 ; /* 0x0000020000008947 */ /* 0x000fea0003800000 */ /*0890*/ ISETP.GT.AND P0, PT, R9, 0xfe, PT ; /* 0x000000fe0900780c */ /* 0x000fda0003f04270 */ /*08a0*/ @P0 BRA 0xa60 ; /* 0x000001b000000947 */ /* 0x000fea0003800000 */ /*08b0*/ ISETP.GE.AND P0, PT, R9, 0x1, PT ; /* 0x000000010900780c */ /* 0x000fda0003f06270 */ /*08c0*/ @P0 BRA 0xaa0 ; /* 0x000001d000000947 */ /* 0x000fea0003800000 */ /*08d0*/ ISETP.GE.AND P0, PT, R9, -0x18, PT ; /* 0xffffffe80900780c */ /* 0x000fe40003f06270 */ /*08e0*/ LOP3.LUT R4, R4, 0x80000000, RZ, 0xc0, !PT ; /* 0x8000000004047812 */ /* 0x000fd600078ec0ff */ /*08f0*/ @!P0 BRA 0xaa0 ; /* 0x000001a000008947 */ /* 0x000fea0003800000 */ /*0900*/ FFMA.RZ R5, R10.reuse, R8.reuse, R11.reuse ; /* 0x000000080a057223 */ /* 0x1c0fe2000000c00b */ /*0910*/ ISETP.NE.AND P2, PT, R9.reuse, RZ, PT ; /* 0x000000ff0900720c */ /* 0x040fe20003f45270 */ /*0920*/ FFMA.RM R6, R10.reuse, R8.reuse, R11.reuse ; /* 0x000000080a067223 */ /* 0x1c0fe2000000400b */ /*0930*/ ISETP.NE.AND P1, PT, R9, RZ, PT ; /* 0x000000ff0900720c */ /* 0x000fe40003f25270 */ /*0940*/ LOP3.LUT R7, R5, 0x7fffff, RZ, 0xc0, !PT ; /* 0x007fffff05077812 */ /* 0x000fe200078ec0ff */ /*0950*/ FFMA.RP R5, R10, R8, R11 ; /* 0x000000080a057223 */ /* 0x000fe2000000800b */ /*0960*/ IADD3 R8, R9, 0x20, RZ ; /* 0x0000002009087810 */ /* 0x000fe20007ffe0ff */ /*0970*/ IMAD.MOV R9, RZ, RZ, -R9 ; /* 0x000000ffff097224 */ /* 0x000fe200078e0a09 */ /*0980*/ LOP3.LUT R7, R7, 0x800000, RZ, 0xfc, !PT ; /* 0x0080000007077812 */ /* 0x000fe400078efcff */ /*0990*/ FSETP.NEU.FTZ.AND P0, PT, R5, R6, PT ; /* 0x000000060500720b */ /* 0x000fc40003f1d000 */ /*09a0*/ SHF.L.U32 R8, R7, R8, RZ ; /* 0x0000000807087219 */ /* 0x000fe400000006ff */ /*09b0*/ SEL R6, R9, RZ, P2 ; /* 0x000000ff09067207 */ /* 0x000fe40001000000 */ /*09c0*/ ISETP.NE.AND P1, PT, R8, RZ, P1 ; /* 0x000000ff0800720c */ /* 0x000fe40000f25270 */ /*09d0*/ SHF.R.U32.HI R6, RZ, R6, R7 ; /* 0x00000006ff067219 */ /* 0x000fe40000011607 */ /*09e0*/ PLOP3.LUT P0, PT, P0, P1, PT, 0xa8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe40000703570 */ /*09f0*/ SHF.R.U32.HI R8, RZ, 0x1, R6 ; /* 0x00000001ff087819 */ /* 0x000fc40000011606 */ /*0a00*/ SEL R5, RZ, 0x1, !P0 ; /* 0x00000001ff057807 */ /* 0x000fc80004000000 */ /*0a10*/ LOP3.LUT R5, R5, 0x1, R8, 0xf8, !PT ; /* 0x0000000105057812 */ /* 0x000fc800078ef808 */ /*0a20*/ LOP3.LUT R5, R5, R6, RZ, 0xc0, !PT ; /* 0x0000000605057212 */ /* 0x000fca00078ec0ff */ /*0a30*/ IMAD.IADD R5, R8, 0x1, R5 ; /* 0x0000000108057824 */ /* 0x000fca00078e0205 */ /*0a40*/ LOP3.LUT R4, R5, R4, RZ, 0xfc, !PT ; /* 0x0000000405047212 */ /* 0x000fe200078efcff */ /*0a50*/ BRA 0xaa0 ; /* 0x0000004000007947 */ /* 0x000fea0003800000 */ /*0a60*/ LOP3.LUT R4, R4, 0x80000000, RZ, 0xc0, !PT ; /* 0x8000000004047812 */ /* 0x000fc800078ec0ff */ /*0a70*/ LOP3.LUT R4, R4, 0x7f800000, RZ, 0xfc, !PT ; /* 0x7f80000004047812 */ /* 0x000fe200078efcff */ /*0a80*/ BRA 0xaa0 ; /* 0x0000001000007947 */ /* 0x000fea0003800000 */ /*0a90*/ IMAD R4, R6, 0x800000, R4 ; /* 0x0080000006047824 */ /* 0x000fe400078e0204 */ /*0aa0*/ BSYNC B2 ; /* 0x0000000000027941 */ /* 0x000fea0003800000 */ /*0ab0*/ BRA 0xb40 ; /* 0x0000008000007947 */ /* 0x000fea0003800000 */ /*0ac0*/ LOP3.LUT R4, R6, 0x80000000, R5, 0x48, !PT ; /* 0x8000000006047812 */ /* 0x000fc800078e4805 */ /*0ad0*/ LOP3.LUT R4, R4, 0x7f800000, RZ, 0xfc, !PT ; /* 0x7f80000004047812 */ /* 0x000fe200078efcff */ /*0ae0*/ BRA 0xb40 ; /* 0x0000005000007947 */ /* 0x000fea0003800000 */ /*0af0*/ LOP3.LUT R4, R6, 0x80000000, R5, 0x48, !PT ; /* 0x8000000006047812 */ /* 0x000fe200078e4805 */ /*0b00*/ BRA 0xb40 ; /* 0x0000003000007947 */ /* 0x000fea0003800000 */ /*0b10*/ MUFU.RSQ R4, -QNAN ; /* 0xffc0000000047908 */ /* 0x000e220000001400 */ /*0b20*/ BRA 0xb40 ; /* 0x0000001000007947 */ /* 0x000fea0003800000 */ /*0b30*/ FADD.FTZ R4, R4, R6 ; /* 0x0000000604047221 */ /* 0x000fe40000010000 */ /*0b40*/ BSYNC B1 ; /* 0x0000000000017941 */ /* 0x000fea0003800000 */ /*0b50*/ IMAD.MOV.U32 R7, RZ, RZ, R4 ; /* 0x000000ffff077224 */ /* 0x001fe400078e0004 */ /*0b60*/ IMAD.MOV.U32 R4, RZ, RZ, R0 ; /* 0x000000ffff047224 */ /* 0x000fe400078e0000 */ /*0b70*/ IMAD.MOV.U32 R5, RZ, RZ, 0x0 ; /* 0x00000000ff057424 */ /* 0x000fc800078e00ff */ /*0b80*/ RET.REL.NODEC R4 0x0 ; /* 0xfffff47004007950 */ /* 0x000fea0003c3ffff */ /*0b90*/ BRA 0xb90; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0ba0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0bb0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0bc0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0bd0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0be0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0bf0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c00*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c10*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c20*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c30*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c40*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c50*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c60*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0c70*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z19_setPrecisionKernelPfmPi .globl _Z19_setPrecisionKernelPfmPi .p2align 8 .type _Z19_setPrecisionKernelPfmPi,@function _Z19_setPrecisionKernelPfmPi: s_clause 0x1 s_load_b32 s4, s[0:1], 0x24 s_load_b64 s[2:3], s[0:1], 0x8 s_waitcnt lgkmcnt(0) s_and_b32 s4, s4, 0xffff s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_mad_u64_u32 v[1:2], null, s15, s4, v[0:1] v_mov_b32_e32 v2, 0 v_cmp_gt_u64_e32 vcc_lo, s[2:3], v[1:2] s_and_saveexec_b32 s2, vcc_lo s_cbranch_execz .LBB0_6 s_load_b64 s[2:3], s[0:1], 0x10 v_lshlrev_b64 v[3:4], 2, v[1:2] v_mov_b32_e32 v0, 1.0 s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v3, vcc_lo, s2, v3 v_add_co_ci_u32_e32 v4, vcc_lo, s3, v4, vcc_lo s_mov_b32 s3, 0 s_mov_b32 s2, exec_lo global_load_b32 v3, v[3:4], off s_waitcnt vmcnt(0) v_cmpx_ne_u32_e32 0, v3 s_cbranch_execz .LBB0_5 s_mov_b32 s4, 1 .LBB0_3: s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_mul_i32 s4, s4, 10 v_dual_mov_b32 v0, s4 :: v_dual_add_nc_u32 v3, -1, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_cmp_eq_u32_e32 vcc_lo, 0, v3 s_or_b32 s3, vcc_lo, s3 s_and_not1_b32 exec_lo, exec_lo, s3 s_cbranch_execnz .LBB0_3 s_or_b32 exec_lo, exec_lo, s3 v_cvt_f32_i32_e32 v0, v0 .LBB0_5: s_or_b32 exec_lo, exec_lo, s2 s_load_b64 s[0:1], s[0:1], 0x0 v_lshlrev_b64 v[1:2], 2, v[1:2] s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v1, vcc_lo, s0, v1 v_add_co_ci_u32_e32 v2, vcc_lo, s1, v2, vcc_lo global_load_b32 v3, v[1:2], off s_waitcnt vmcnt(0) v_mul_f32_e32 v3, v3, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cvt_i32_f32_e32 v3, v3 v_cvt_f32_i32_e32 v3, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_div_scale_f32 v4, null, v0, v0, v3 v_rcp_f32_e32 v5, v4 s_waitcnt_depctr 0xfff v_fma_f32 v6, -v4, v5, 1.0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_fmac_f32_e32 v5, v6, v5 v_div_scale_f32 v6, vcc_lo, v3, v0, v3 v_mul_f32_e32 v7, v6, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fma_f32 v8, -v4, v7, v6 v_fmac_f32_e32 v7, v8, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fma_f32 v4, -v4, v7, v6 v_div_fmas_f32 v4, v4, v5, v7 s_delay_alu instid0(VALU_DEP_1) v_div_fixup_f32 v0, v4, v0, v3 global_store_b32 v[1:2], v0, off .LBB0_6: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z19_setPrecisionKernelPfmPi .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 280 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 9 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z19_setPrecisionKernelPfmPi, .Lfunc_end0-_Z19_setPrecisionKernelPfmPi .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .offset: 8 .size: 8 .value_kind: by_value - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: hidden_block_count_x - .offset: 28 .size: 4 .value_kind: hidden_block_count_y - .offset: 32 .size: 4 .value_kind: hidden_block_count_z - .offset: 36 .size: 2 .value_kind: hidden_group_size_x - .offset: 38 .size: 2 .value_kind: hidden_group_size_y - .offset: 40 .size: 2 .value_kind: hidden_group_size_z - .offset: 42 .size: 2 .value_kind: hidden_remainder_x - .offset: 44 .size: 2 .value_kind: hidden_remainder_y - .offset: 46 .size: 2 .value_kind: hidden_remainder_z - .offset: 64 .size: 8 .value_kind: hidden_global_offset_x - .offset: 72 .size: 8 .value_kind: hidden_global_offset_y - .offset: 80 .size: 8 .value_kind: hidden_global_offset_z - .offset: 88 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 280 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z19_setPrecisionKernelPfmPi .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z19_setPrecisionKernelPfmPi.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 9 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_0016fd16_00000000-6__setPrecisionKernel.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2029: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2029: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi .type _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi, @function _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi: .LFB2051: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movq %fs:40, %rax movq %rax, 120(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 120(%rsp), %rax subq %fs:40, %rax jne .L8 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 152 pushq 40(%rsp) .cfi_def_cfa_offset 160 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z19_setPrecisionKernelPfmPi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2051: .size _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi, .-_Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi .globl _Z19_setPrecisionKernelPfmPi .type _Z19_setPrecisionKernelPfmPi, @function _Z19_setPrecisionKernelPfmPi: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z42__device_stub__Z19_setPrecisionKernelPfmPiPfmPi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z19_setPrecisionKernelPfmPi, .-_Z19_setPrecisionKernelPfmPi .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z19_setPrecisionKernelPfmPi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2054: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z19_setPrecisionKernelPfmPi(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2054: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "_setPrecisionKernel.hip" .globl _Z34__device_stub___setPrecisionKernelPfmPi # -- Begin function _Z34__device_stub___setPrecisionKernelPfmPi .p2align 4, 0x90 .type _Z34__device_stub___setPrecisionKernelPfmPi,@function _Z34__device_stub___setPrecisionKernelPfmPi: # @_Z34__device_stub___setPrecisionKernelPfmPi .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movq %rdx, 56(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z19_setPrecisionKernelPfmPi, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $120, %rsp .cfi_adjust_cfa_offset -120 retq .Lfunc_end0: .size _Z34__device_stub___setPrecisionKernelPfmPi, .Lfunc_end0-_Z34__device_stub___setPrecisionKernelPfmPi .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z19_setPrecisionKernelPfmPi, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type _Z19_setPrecisionKernelPfmPi,@object # @_Z19_setPrecisionKernelPfmPi .section .rodata,"a",@progbits .globl _Z19_setPrecisionKernelPfmPi .p2align 3, 0x0 _Z19_setPrecisionKernelPfmPi: .quad _Z34__device_stub___setPrecisionKernelPfmPi .size _Z19_setPrecisionKernelPfmPi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z19_setPrecisionKernelPfmPi" .size .L__unnamed_1, 29 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z34__device_stub___setPrecisionKernelPfmPi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z19_setPrecisionKernelPfmPi .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
#include <stdlib.h> #include <stdio.h> #include <string.h> #include <math.h> #include <cuda_runtime.h>
code for sm_80
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include <stdlib.h> #include <stdio.h> #include <string.h> #include <math.h> #include <cuda_runtime.h>
.file "tmpxft_001bbb56_00000000-6_cubemap_texture.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2059: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2059: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2082: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2082: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include <stdlib.h> #include <stdio.h> #include <string.h> #include <math.h> #include <cuda_runtime.h>
#include <stdlib.h> #include <stdio.h> #include <string.h> #include <math.h> #include <hip/hip_runtime.h>
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <stdlib.h> #include <stdio.h> #include <string.h> #include <math.h> #include <hip/hip_runtime.h>
.text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .amdgpu_metadata --- amdhsa.kernels: [] amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <stdlib.h> #include <stdio.h> #include <string.h> #include <math.h> #include <hip/hip_runtime.h>
.text .file "cubemap_texture.hip" .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80
.text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .amdgpu_metadata --- amdhsa.kernels: [] amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_001bbb56_00000000-6_cubemap_texture.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2059: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2059: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2082: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2082: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "cubemap_texture.hip" .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
//#define REARRANGED_DOMAIN __global__ void _interpolate_from_vertices_to_edges( int N, double* vertex_values, double* edge_values) { const int k = threadIdx.x+threadIdx.y*blockDim.x+ (blockIdx.x+blockIdx.y*gridDim.x)*blockDim.x*blockDim.y; if ( k >= N ) return; #ifndef REARRANGED_DOMAIN int k3= k*3; #endif double q0, q1, q2; //for (k=0; k<N; k++) { #ifndef REARRANGED_DOMAIN q0 = vertex_values[k3 + 0]; q1 = vertex_values[k3 + 1]; q2 = vertex_values[k3 + 2]; edge_values[k3 + 0] = 0.5*(q1+q2); edge_values[k3 + 1] = 0.5*(q0+q2); edge_values[k3 + 2] = 0.5*(q0+q1); #else q0 = vertex_values[k]; q1 = vertex_values[k + N]; q2 = vertex_values[k + 2*N]; edge_values[k] = 0.5*(q1+q2); edge_values[k + N] = 0.5*(q0+q2); edge_values[k + 2*N] = 0.5*(q0+q1); #endif //} }
code for sm_80 Function : _Z35_interpolate_from_vertices_to_edgesiPdS_ .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R0, SR_CTAID.Y ; /* 0x0000000000007919 */ /* 0x000e280000002600 */ /*0020*/ S2R R3, SR_CTAID.X ; /* 0x0000000000037919 */ /* 0x000e280000002500 */ /*0030*/ S2R R5, SR_TID.Y ; /* 0x0000000000057919 */ /* 0x000e680000002200 */ /*0040*/ S2R R7, SR_TID.X ; /* 0x0000000000077919 */ /* 0x000ea20000002100 */ /*0050*/ IMAD R0, R0, c[0x0][0xc], R3 ; /* 0x0000030000007a24 */ /* 0x001fc800078e0203 */ /*0060*/ IMAD R0, R0, c[0x0][0x4], R5 ; /* 0x0000010000007a24 */ /* 0x002fc800078e0205 */ /*0070*/ IMAD R0, R0, c[0x0][0x0], R7 ; /* 0x0000000000007a24 */ /* 0x004fca00078e0207 */ /*0080*/ ISETP.GE.AND P0, PT, R0, c[0x0][0x160], PT ; /* 0x0000580000007a0c */ /* 0x000fda0003f06270 */ /*0090*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*00a0*/ HFMA2.MMA R13, -RZ, RZ, 0, 4.76837158203125e-07 ; /* 0x00000008ff0d7435 */ /* 0x000fe200000001ff */ /*00b0*/ LEA R0, R0, R0, 0x1 ; /* 0x0000000000007211 */ /* 0x000fe200078e08ff */ /*00c0*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd00000000a00 */ /*00d0*/ IMAD.WIDE R2, R0, R13, c[0x0][0x168] ; /* 0x00005a0000027625 */ /* 0x000fca00078e020d */ /*00e0*/ LDG.E.64 R4, [R2.64] ; /* 0x0000000402047981 */ /* 0x000ea8000c1e1b00 */ /*00f0*/ LDG.E.64 R6, [R2.64+0x10] ; /* 0x0000100402067981 */ /* 0x000ea8000c1e1b00 */ /*0100*/ LDG.E.64 R8, [R2.64+0x8] ; /* 0x0000080402087981 */ /* 0x000ee2000c1e1b00 */ /*0110*/ DADD R10, R4, R6 ; /* 0x00000000040a7229 */ /* 0x004e080000000006 */ /*0120*/ DADD R6, R6, R8 ; /* 0x0000000006067229 */ /* 0x008e480000000008 */ /*0130*/ DADD R8, R4, R8 ; /* 0x0000000004087229 */ /* 0x0005e40000000008 */ /*0140*/ IMAD.WIDE R4, R0, R13, c[0x0][0x170] ; /* 0x00005c0000047625 */ /* 0x004fe400078e020d */ /*0150*/ DMUL R10, R10, 0.5 ; /* 0x3fe000000a0a7828 */ /* 0x001e080000000000 */ /*0160*/ DMUL R6, R6, 0.5 ; /* 0x3fe0000006067828 */ /* 0x002e460000000000 */ /*0170*/ STG.E.64 [R4.64+0x8], R10 ; /* 0x0000080a04007986 */ /* 0x001fe2000c101b04 */ /*0180*/ DMUL R8, R8, 0.5 ; /* 0x3fe0000008087828 */ /* 0x000e060000000000 */ /*0190*/ STG.E.64 [R4.64], R6 ; /* 0x0000000604007986 */ /* 0x002fe8000c101b04 */ /*01a0*/ STG.E.64 [R4.64+0x10], R8 ; /* 0x0000100804007986 */ /* 0x001fe2000c101b04 */ /*01b0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*01c0*/ BRA 0x1c0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*01d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0200*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0210*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0220*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0230*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0240*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0250*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0260*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0270*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
//#define REARRANGED_DOMAIN __global__ void _interpolate_from_vertices_to_edges( int N, double* vertex_values, double* edge_values) { const int k = threadIdx.x+threadIdx.y*blockDim.x+ (blockIdx.x+blockIdx.y*gridDim.x)*blockDim.x*blockDim.y; if ( k >= N ) return; #ifndef REARRANGED_DOMAIN int k3= k*3; #endif double q0, q1, q2; //for (k=0; k<N; k++) { #ifndef REARRANGED_DOMAIN q0 = vertex_values[k3 + 0]; q1 = vertex_values[k3 + 1]; q2 = vertex_values[k3 + 2]; edge_values[k3 + 0] = 0.5*(q1+q2); edge_values[k3 + 1] = 0.5*(q0+q2); edge_values[k3 + 2] = 0.5*(q0+q1); #else q0 = vertex_values[k]; q1 = vertex_values[k + N]; q2 = vertex_values[k + 2*N]; edge_values[k] = 0.5*(q1+q2); edge_values[k + N] = 0.5*(q0+q2); edge_values[k + 2*N] = 0.5*(q0+q1); #endif //} }
.file "tmpxft_000df930_00000000-6_interpolate_from_vertices_to_edges.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2029: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2029: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_ .type _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_, @function _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_: .LFB2051: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movl %edi, 28(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movq %fs:40, %rax movq %rax, 120(%rsp) xorl %eax, %eax leaq 28(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 120(%rsp), %rax subq %fs:40, %rax jne .L8 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 152 pushq 40(%rsp) .cfi_def_cfa_offset 160 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z35_interpolate_from_vertices_to_edgesiPdS_(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2051: .size _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_, .-_Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_ .globl _Z35_interpolate_from_vertices_to_edgesiPdS_ .type _Z35_interpolate_from_vertices_to_edgesiPdS_, @function _Z35_interpolate_from_vertices_to_edgesiPdS_: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z35_interpolate_from_vertices_to_edgesiPdS_, .-_Z35_interpolate_from_vertices_to_edgesiPdS_ .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC0: .string "_Z35_interpolate_from_vertices_to_edgesiPdS_" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2054: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z35_interpolate_from_vertices_to_edgesiPdS_(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2054: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
//#define REARRANGED_DOMAIN __global__ void _interpolate_from_vertices_to_edges( int N, double* vertex_values, double* edge_values) { const int k = threadIdx.x+threadIdx.y*blockDim.x+ (blockIdx.x+blockIdx.y*gridDim.x)*blockDim.x*blockDim.y; if ( k >= N ) return; #ifndef REARRANGED_DOMAIN int k3= k*3; #endif double q0, q1, q2; //for (k=0; k<N; k++) { #ifndef REARRANGED_DOMAIN q0 = vertex_values[k3 + 0]; q1 = vertex_values[k3 + 1]; q2 = vertex_values[k3 + 2]; edge_values[k3 + 0] = 0.5*(q1+q2); edge_values[k3 + 1] = 0.5*(q0+q2); edge_values[k3 + 2] = 0.5*(q0+q1); #else q0 = vertex_values[k]; q1 = vertex_values[k + N]; q2 = vertex_values[k + 2*N]; edge_values[k] = 0.5*(q1+q2); edge_values[k + N] = 0.5*(q0+q2); edge_values[k + 2*N] = 0.5*(q0+q1); #endif //} }
#include <hip/hip_runtime.h> //#define REARRANGED_DOMAIN __global__ void _interpolate_from_vertices_to_edges( int N, double* vertex_values, double* edge_values) { const int k = threadIdx.x+threadIdx.y*blockDim.x+ (blockIdx.x+blockIdx.y*gridDim.x)*blockDim.x*blockDim.y; if ( k >= N ) return; #ifndef REARRANGED_DOMAIN int k3= k*3; #endif double q0, q1, q2; //for (k=0; k<N; k++) { #ifndef REARRANGED_DOMAIN q0 = vertex_values[k3 + 0]; q1 = vertex_values[k3 + 1]; q2 = vertex_values[k3 + 2]; edge_values[k3 + 0] = 0.5*(q1+q2); edge_values[k3 + 1] = 0.5*(q0+q2); edge_values[k3 + 2] = 0.5*(q0+q1); #else q0 = vertex_values[k]; q1 = vertex_values[k + N]; q2 = vertex_values[k + 2*N]; edge_values[k] = 0.5*(q1+q2); edge_values[k + N] = 0.5*(q0+q2); edge_values[k + 2*N] = 0.5*(q0+q1); #endif //} }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> //#define REARRANGED_DOMAIN __global__ void _interpolate_from_vertices_to_edges( int N, double* vertex_values, double* edge_values) { const int k = threadIdx.x+threadIdx.y*blockDim.x+ (blockIdx.x+blockIdx.y*gridDim.x)*blockDim.x*blockDim.y; if ( k >= N ) return; #ifndef REARRANGED_DOMAIN int k3= k*3; #endif double q0, q1, q2; //for (k=0; k<N; k++) { #ifndef REARRANGED_DOMAIN q0 = vertex_values[k3 + 0]; q1 = vertex_values[k3 + 1]; q2 = vertex_values[k3 + 2]; edge_values[k3 + 0] = 0.5*(q1+q2); edge_values[k3 + 1] = 0.5*(q0+q2); edge_values[k3 + 2] = 0.5*(q0+q1); #else q0 = vertex_values[k]; q1 = vertex_values[k + N]; q2 = vertex_values[k + 2*N]; edge_values[k] = 0.5*(q1+q2); edge_values[k + N] = 0.5*(q0+q2); edge_values[k + 2*N] = 0.5*(q0+q1); #endif //} }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z35_interpolate_from_vertices_to_edgesiPdS_ .globl _Z35_interpolate_from_vertices_to_edgesiPdS_ .p2align 8 .type _Z35_interpolate_from_vertices_to_edgesiPdS_,@function _Z35_interpolate_from_vertices_to_edgesiPdS_: s_clause 0x1 s_load_b32 s2, s[0:1], 0x18 s_load_b32 s3, s[0:1], 0x24 v_bfe_u32 v1, v0, 10, 10 s_load_b32 s4, s[0:1], 0x0 s_waitcnt lgkmcnt(0) s_mul_i32 s2, s2, s15 s_lshr_b32 s5, s3, 16 s_add_i32 s2, s2, s14 s_delay_alu instid0(SALU_CYCLE_1) v_mad_u64_u32 v[2:3], null, s2, s5, v[1:2] v_and_b32_e32 v3, 0x3ff, v0 s_and_b32 s2, s3, 0xffff s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) v_mad_u64_u32 v[0:1], null, v2, s2, v[3:4] s_mov_b32 s2, exec_lo s_delay_alu instid0(VALU_DEP_1) v_cmpx_gt_i32_e64 s4, v0 s_cbranch_execz .LBB0_2 v_lshl_add_u32 v0, v0, 1, v0 s_load_b128 s[0:3], s[0:1], 0x8 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_ashrrev_i32_e32 v1, 31, v0 v_lshlrev_b64 v[0:1], 3, v[0:1] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v10, vcc_lo, v0, 8 v_add_co_ci_u32_e32 v11, vcc_lo, 0, v1, vcc_lo v_add_co_u32 v12, vcc_lo, v0, 16 v_add_co_ci_u32_e32 v13, vcc_lo, 0, v1, vcc_lo s_waitcnt lgkmcnt(0) v_add_co_u32 v2, vcc_lo, s0, v10 v_add_co_ci_u32_e32 v3, vcc_lo, s1, v11, vcc_lo v_add_co_u32 v4, vcc_lo, s0, v12 v_add_co_ci_u32_e32 v5, vcc_lo, s1, v13, vcc_lo v_add_co_u32 v6, vcc_lo, s0, v0 v_add_co_ci_u32_e32 v7, vcc_lo, s1, v1, vcc_lo s_clause 0x2 global_load_b64 v[2:3], v[2:3], off global_load_b64 v[4:5], v[4:5], off global_load_b64 v[6:7], v[6:7], off v_add_co_u32 v0, vcc_lo, s2, v0 v_add_co_ci_u32_e32 v1, vcc_lo, s3, v1, vcc_lo s_waitcnt vmcnt(1) v_add_f64 v[8:9], v[2:3], v[4:5] s_waitcnt vmcnt(0) v_add_f64 v[4:5], v[6:7], v[4:5] v_add_f64 v[2:3], v[6:7], v[2:3] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_mul_f64 v[6:7], v[8:9], 0.5 v_mul_f64 v[4:5], v[4:5], 0.5 s_delay_alu instid0(VALU_DEP_3) v_mul_f64 v[2:3], v[2:3], 0.5 v_add_co_u32 v8, vcc_lo, s2, v10 v_add_co_ci_u32_e32 v9, vcc_lo, s3, v11, vcc_lo v_add_co_u32 v10, vcc_lo, s2, v12 v_add_co_ci_u32_e32 v11, vcc_lo, s3, v13, vcc_lo s_clause 0x2 global_store_b64 v[0:1], v[6:7], off global_store_b64 v[8:9], v[4:5], off global_store_b64 v[10:11], v[2:3], off .LBB0_2: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z35_interpolate_from_vertices_to_edgesiPdS_ .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 280 .amdhsa_user_sgpr_count 14 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 1 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 1 .amdhsa_next_free_vgpr 14 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z35_interpolate_from_vertices_to_edgesiPdS_, .Lfunc_end0-_Z35_interpolate_from_vertices_to_edgesiPdS_ .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .offset: 0 .size: 4 .value_kind: by_value - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: hidden_block_count_x - .offset: 28 .size: 4 .value_kind: hidden_block_count_y - .offset: 32 .size: 4 .value_kind: hidden_block_count_z - .offset: 36 .size: 2 .value_kind: hidden_group_size_x - .offset: 38 .size: 2 .value_kind: hidden_group_size_y - .offset: 40 .size: 2 .value_kind: hidden_group_size_z - .offset: 42 .size: 2 .value_kind: hidden_remainder_x - .offset: 44 .size: 2 .value_kind: hidden_remainder_y - .offset: 46 .size: 2 .value_kind: hidden_remainder_z - .offset: 64 .size: 8 .value_kind: hidden_global_offset_x - .offset: 72 .size: 8 .value_kind: hidden_global_offset_y - .offset: 80 .size: 8 .value_kind: hidden_global_offset_z - .offset: 88 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 280 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z35_interpolate_from_vertices_to_edgesiPdS_ .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z35_interpolate_from_vertices_to_edgesiPdS_.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 14 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <hip/hip_runtime.h> //#define REARRANGED_DOMAIN __global__ void _interpolate_from_vertices_to_edges( int N, double* vertex_values, double* edge_values) { const int k = threadIdx.x+threadIdx.y*blockDim.x+ (blockIdx.x+blockIdx.y*gridDim.x)*blockDim.x*blockDim.y; if ( k >= N ) return; #ifndef REARRANGED_DOMAIN int k3= k*3; #endif double q0, q1, q2; //for (k=0; k<N; k++) { #ifndef REARRANGED_DOMAIN q0 = vertex_values[k3 + 0]; q1 = vertex_values[k3 + 1]; q2 = vertex_values[k3 + 2]; edge_values[k3 + 0] = 0.5*(q1+q2); edge_values[k3 + 1] = 0.5*(q0+q2); edge_values[k3 + 2] = 0.5*(q0+q1); #else q0 = vertex_values[k]; q1 = vertex_values[k + N]; q2 = vertex_values[k + 2*N]; edge_values[k] = 0.5*(q1+q2); edge_values[k + N] = 0.5*(q0+q2); edge_values[k + 2*N] = 0.5*(q0+q1); #endif //} }
.text .file "interpolate_from_vertices_to_edges.hip" .globl _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ # -- Begin function _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .p2align 4, 0x90 .type _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_,@function _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_: # @_Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movl %edi, 12(%rsp) movq %rsi, 72(%rsp) movq %rdx, 64(%rsp) leaq 12(%rsp), %rax movq %rax, 80(%rsp) leaq 72(%rsp), %rax movq %rax, 88(%rsp) leaq 64(%rsp), %rax movq %rax, 96(%rsp) leaq 48(%rsp), %rdi leaq 32(%rsp), %rsi leaq 24(%rsp), %rdx leaq 16(%rsp), %rcx callq __hipPopCallConfiguration movq 48(%rsp), %rsi movl 56(%rsp), %edx movq 32(%rsp), %rcx movl 40(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z35_interpolate_from_vertices_to_edgesiPdS_, %edi pushq 16(%rsp) .cfi_adjust_cfa_offset 8 pushq 32(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $120, %rsp .cfi_adjust_cfa_offset -120 retq .Lfunc_end0: .size _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_, .Lfunc_end0-_Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z35_interpolate_from_vertices_to_edgesiPdS_, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type _Z35_interpolate_from_vertices_to_edgesiPdS_,@object # @_Z35_interpolate_from_vertices_to_edgesiPdS_ .section .rodata,"a",@progbits .globl _Z35_interpolate_from_vertices_to_edgesiPdS_ .p2align 3, 0x0 _Z35_interpolate_from_vertices_to_edgesiPdS_: .quad _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .size _Z35_interpolate_from_vertices_to_edgesiPdS_, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z35_interpolate_from_vertices_to_edgesiPdS_" .size .L__unnamed_1, 45 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z35_interpolate_from_vertices_to_edgesiPdS_ .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z35_interpolate_from_vertices_to_edgesiPdS_ .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R0, SR_CTAID.Y ; /* 0x0000000000007919 */ /* 0x000e280000002600 */ /*0020*/ S2R R3, SR_CTAID.X ; /* 0x0000000000037919 */ /* 0x000e280000002500 */ /*0030*/ S2R R5, SR_TID.Y ; /* 0x0000000000057919 */ /* 0x000e680000002200 */ /*0040*/ S2R R7, SR_TID.X ; /* 0x0000000000077919 */ /* 0x000ea20000002100 */ /*0050*/ IMAD R0, R0, c[0x0][0xc], R3 ; /* 0x0000030000007a24 */ /* 0x001fc800078e0203 */ /*0060*/ IMAD R0, R0, c[0x0][0x4], R5 ; /* 0x0000010000007a24 */ /* 0x002fc800078e0205 */ /*0070*/ IMAD R0, R0, c[0x0][0x0], R7 ; /* 0x0000000000007a24 */ /* 0x004fca00078e0207 */ /*0080*/ ISETP.GE.AND P0, PT, R0, c[0x0][0x160], PT ; /* 0x0000580000007a0c */ /* 0x000fda0003f06270 */ /*0090*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*00a0*/ HFMA2.MMA R13, -RZ, RZ, 0, 4.76837158203125e-07 ; /* 0x00000008ff0d7435 */ /* 0x000fe200000001ff */ /*00b0*/ LEA R0, R0, R0, 0x1 ; /* 0x0000000000007211 */ /* 0x000fe200078e08ff */ /*00c0*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd00000000a00 */ /*00d0*/ IMAD.WIDE R2, R0, R13, c[0x0][0x168] ; /* 0x00005a0000027625 */ /* 0x000fca00078e020d */ /*00e0*/ LDG.E.64 R4, [R2.64] ; /* 0x0000000402047981 */ /* 0x000ea8000c1e1b00 */ /*00f0*/ LDG.E.64 R6, [R2.64+0x10] ; /* 0x0000100402067981 */ /* 0x000ea8000c1e1b00 */ /*0100*/ LDG.E.64 R8, [R2.64+0x8] ; /* 0x0000080402087981 */ /* 0x000ee2000c1e1b00 */ /*0110*/ DADD R10, R4, R6 ; /* 0x00000000040a7229 */ /* 0x004e080000000006 */ /*0120*/ DADD R6, R6, R8 ; /* 0x0000000006067229 */ /* 0x008e480000000008 */ /*0130*/ DADD R8, R4, R8 ; /* 0x0000000004087229 */ /* 0x0005e40000000008 */ /*0140*/ IMAD.WIDE R4, R0, R13, c[0x0][0x170] ; /* 0x00005c0000047625 */ /* 0x004fe400078e020d */ /*0150*/ DMUL R10, R10, 0.5 ; /* 0x3fe000000a0a7828 */ /* 0x001e080000000000 */ /*0160*/ DMUL R6, R6, 0.5 ; /* 0x3fe0000006067828 */ /* 0x002e460000000000 */ /*0170*/ STG.E.64 [R4.64+0x8], R10 ; /* 0x0000080a04007986 */ /* 0x001fe2000c101b04 */ /*0180*/ DMUL R8, R8, 0.5 ; /* 0x3fe0000008087828 */ /* 0x000e060000000000 */ /*0190*/ STG.E.64 [R4.64], R6 ; /* 0x0000000604007986 */ /* 0x002fe8000c101b04 */ /*01a0*/ STG.E.64 [R4.64+0x10], R8 ; /* 0x0000100804007986 */ /* 0x001fe2000c101b04 */ /*01b0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*01c0*/ BRA 0x1c0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*01d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0200*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0210*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0220*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0230*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0240*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0250*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0260*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0270*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z35_interpolate_from_vertices_to_edgesiPdS_ .globl _Z35_interpolate_from_vertices_to_edgesiPdS_ .p2align 8 .type _Z35_interpolate_from_vertices_to_edgesiPdS_,@function _Z35_interpolate_from_vertices_to_edgesiPdS_: s_clause 0x1 s_load_b32 s2, s[0:1], 0x18 s_load_b32 s3, s[0:1], 0x24 v_bfe_u32 v1, v0, 10, 10 s_load_b32 s4, s[0:1], 0x0 s_waitcnt lgkmcnt(0) s_mul_i32 s2, s2, s15 s_lshr_b32 s5, s3, 16 s_add_i32 s2, s2, s14 s_delay_alu instid0(SALU_CYCLE_1) v_mad_u64_u32 v[2:3], null, s2, s5, v[1:2] v_and_b32_e32 v3, 0x3ff, v0 s_and_b32 s2, s3, 0xffff s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) v_mad_u64_u32 v[0:1], null, v2, s2, v[3:4] s_mov_b32 s2, exec_lo s_delay_alu instid0(VALU_DEP_1) v_cmpx_gt_i32_e64 s4, v0 s_cbranch_execz .LBB0_2 v_lshl_add_u32 v0, v0, 1, v0 s_load_b128 s[0:3], s[0:1], 0x8 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_ashrrev_i32_e32 v1, 31, v0 v_lshlrev_b64 v[0:1], 3, v[0:1] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v10, vcc_lo, v0, 8 v_add_co_ci_u32_e32 v11, vcc_lo, 0, v1, vcc_lo v_add_co_u32 v12, vcc_lo, v0, 16 v_add_co_ci_u32_e32 v13, vcc_lo, 0, v1, vcc_lo s_waitcnt lgkmcnt(0) v_add_co_u32 v2, vcc_lo, s0, v10 v_add_co_ci_u32_e32 v3, vcc_lo, s1, v11, vcc_lo v_add_co_u32 v4, vcc_lo, s0, v12 v_add_co_ci_u32_e32 v5, vcc_lo, s1, v13, vcc_lo v_add_co_u32 v6, vcc_lo, s0, v0 v_add_co_ci_u32_e32 v7, vcc_lo, s1, v1, vcc_lo s_clause 0x2 global_load_b64 v[2:3], v[2:3], off global_load_b64 v[4:5], v[4:5], off global_load_b64 v[6:7], v[6:7], off v_add_co_u32 v0, vcc_lo, s2, v0 v_add_co_ci_u32_e32 v1, vcc_lo, s3, v1, vcc_lo s_waitcnt vmcnt(1) v_add_f64 v[8:9], v[2:3], v[4:5] s_waitcnt vmcnt(0) v_add_f64 v[4:5], v[6:7], v[4:5] v_add_f64 v[2:3], v[6:7], v[2:3] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_mul_f64 v[6:7], v[8:9], 0.5 v_mul_f64 v[4:5], v[4:5], 0.5 s_delay_alu instid0(VALU_DEP_3) v_mul_f64 v[2:3], v[2:3], 0.5 v_add_co_u32 v8, vcc_lo, s2, v10 v_add_co_ci_u32_e32 v9, vcc_lo, s3, v11, vcc_lo v_add_co_u32 v10, vcc_lo, s2, v12 v_add_co_ci_u32_e32 v11, vcc_lo, s3, v13, vcc_lo s_clause 0x2 global_store_b64 v[0:1], v[6:7], off global_store_b64 v[8:9], v[4:5], off global_store_b64 v[10:11], v[2:3], off .LBB0_2: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z35_interpolate_from_vertices_to_edgesiPdS_ .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 280 .amdhsa_user_sgpr_count 14 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 1 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 1 .amdhsa_next_free_vgpr 14 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z35_interpolate_from_vertices_to_edgesiPdS_, .Lfunc_end0-_Z35_interpolate_from_vertices_to_edgesiPdS_ .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .offset: 0 .size: 4 .value_kind: by_value - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: hidden_block_count_x - .offset: 28 .size: 4 .value_kind: hidden_block_count_y - .offset: 32 .size: 4 .value_kind: hidden_block_count_z - .offset: 36 .size: 2 .value_kind: hidden_group_size_x - .offset: 38 .size: 2 .value_kind: hidden_group_size_y - .offset: 40 .size: 2 .value_kind: hidden_group_size_z - .offset: 42 .size: 2 .value_kind: hidden_remainder_x - .offset: 44 .size: 2 .value_kind: hidden_remainder_y - .offset: 46 .size: 2 .value_kind: hidden_remainder_z - .offset: 64 .size: 8 .value_kind: hidden_global_offset_x - .offset: 72 .size: 8 .value_kind: hidden_global_offset_y - .offset: 80 .size: 8 .value_kind: hidden_global_offset_z - .offset: 88 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 280 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z35_interpolate_from_vertices_to_edgesiPdS_ .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z35_interpolate_from_vertices_to_edgesiPdS_.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 14 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_000df930_00000000-6_interpolate_from_vertices_to_edges.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2029: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2029: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_ .type _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_, @function _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_: .LFB2051: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movl %edi, 28(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movq %fs:40, %rax movq %rax, 120(%rsp) xorl %eax, %eax leaq 28(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 120(%rsp), %rax subq %fs:40, %rax jne .L8 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 152 pushq 40(%rsp) .cfi_def_cfa_offset 160 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z35_interpolate_from_vertices_to_edgesiPdS_(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2051: .size _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_, .-_Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_ .globl _Z35_interpolate_from_vertices_to_edgesiPdS_ .type _Z35_interpolate_from_vertices_to_edgesiPdS_, @function _Z35_interpolate_from_vertices_to_edgesiPdS_: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z58__device_stub__Z35_interpolate_from_vertices_to_edgesiPdS_iPdS_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z35_interpolate_from_vertices_to_edgesiPdS_, .-_Z35_interpolate_from_vertices_to_edgesiPdS_ .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC0: .string "_Z35_interpolate_from_vertices_to_edgesiPdS_" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2054: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z35_interpolate_from_vertices_to_edgesiPdS_(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2054: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "interpolate_from_vertices_to_edges.hip" .globl _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ # -- Begin function _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .p2align 4, 0x90 .type _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_,@function _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_: # @_Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movl %edi, 12(%rsp) movq %rsi, 72(%rsp) movq %rdx, 64(%rsp) leaq 12(%rsp), %rax movq %rax, 80(%rsp) leaq 72(%rsp), %rax movq %rax, 88(%rsp) leaq 64(%rsp), %rax movq %rax, 96(%rsp) leaq 48(%rsp), %rdi leaq 32(%rsp), %rsi leaq 24(%rsp), %rdx leaq 16(%rsp), %rcx callq __hipPopCallConfiguration movq 48(%rsp), %rsi movl 56(%rsp), %edx movq 32(%rsp), %rcx movl 40(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z35_interpolate_from_vertices_to_edgesiPdS_, %edi pushq 16(%rsp) .cfi_adjust_cfa_offset 8 pushq 32(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $120, %rsp .cfi_adjust_cfa_offset -120 retq .Lfunc_end0: .size _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_, .Lfunc_end0-_Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z35_interpolate_from_vertices_to_edgesiPdS_, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type _Z35_interpolate_from_vertices_to_edgesiPdS_,@object # @_Z35_interpolate_from_vertices_to_edgesiPdS_ .section .rodata,"a",@progbits .globl _Z35_interpolate_from_vertices_to_edgesiPdS_ .p2align 3, 0x0 _Z35_interpolate_from_vertices_to_edgesiPdS_: .quad _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .size _Z35_interpolate_from_vertices_to_edgesiPdS_, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z35_interpolate_from_vertices_to_edgesiPdS_" .size .L__unnamed_1, 45 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z50__device_stub___interpolate_from_vertices_to_edgesiPdS_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z35_interpolate_from_vertices_to_edgesiPdS_ .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
#include "stdio.h" #include "time.h" #include <iostream> #include <cuda.h> #include <cuda_runtime.h> #include "device_launch_parameters.h" // Defining number of elements in Array #define N 10000000 // Defining Kernel function for vector addition __global__ void gpuAdd(int *d_a, int *d_b, int *d_c) { // Getting block index of current kernel int tid = blockIdx.x * 1000 + threadIdx.x; if (tid < N) d_c[tid] = d_a[tid] + d_b[tid]; } int main(void) { // Defining host arrays int *h_a, *h_b, *h_c; h_a = (int *)malloc(sizeof(int) * N); h_b = (int *)malloc(sizeof(int) * N); h_c = (int *)malloc(sizeof(int) * N); // Defining device pointers int *d_a, *d_b, *d_c; // Initializing two arrays for addition for (int i = 0; i < N; i++) { h_a[i] = i + 1; h_b[i] = i - 1; } // Allocate the memory cudaMalloc((void **)&d_a, sizeof(int) * N); cudaMalloc((void **)&d_b, sizeof(int) * N); cudaMalloc((void **)&d_c, sizeof(int) * N); // Copy input arrays from host to device memory cudaMemcpy(d_a, h_a, N * sizeof(int), cudaMemcpyHostToDevice); cudaMemcpy(d_b, h_b, N * sizeof(int), cudaMemcpyHostToDevice); // Start Time clock_t start_h = clock(); // Calling kernels with N blocks and one thread per block, passing device pointers as parameters gpuAdd <<<10000, 1000 >>>(d_a, d_b, d_c); cudaThreadSynchronize(); // End Time clock_t end_h = clock(); // Copy result back to host memory from device memory cudaMemcpy(h_c, d_c, N * sizeof(int), cudaMemcpyDeviceToHost); printf("Vector addition on GPU \n"); // Execute Time double time_h = (double)(end_h - start_h) / CLOCKS_PER_SEC; printf("N = %d \n", N); printf("Execute time: %f seconds \n", time_h); // Printing result on console for (int i = 9999990; i < N; i++) { printf("The sum of %d element is %d + %d = %d\n", i, h_a[i], h_b[i], h_c[i]); } // Free up memory cudaFree(d_a); cudaFree(d_b); cudaFree(d_c); system("pause"); return 0; }
code for sm_80 Function : _Z6gpuAddPiS_S_ .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R6, SR_CTAID.X ; /* 0x0000000000067919 */ /* 0x000e280000002500 */ /*0020*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e240000002100 */ /*0030*/ IMAD R6, R6, 0x3e8, R3 ; /* 0x000003e806067824 */ /* 0x001fca00078e0203 */ /*0040*/ ISETP.GT.AND P0, PT, R6, 0x98967f, PT ; /* 0x0098967f0600780c */ /* 0x000fda0003f04270 */ /*0050*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0060*/ HFMA2.MMA R7, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff077435 */ /* 0x000fe200000001ff */ /*0070*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd20000000a00 */ /*0080*/ IMAD.WIDE R4, R6, R7, c[0x0][0x168] ; /* 0x00005a0006047625 */ /* 0x000fc800078e0207 */ /*0090*/ IMAD.WIDE R2, R6.reuse, R7.reuse, c[0x0][0x160] ; /* 0x0000580006027625 */ /* 0x0c0fe400078e0207 */ /*00a0*/ LDG.E R4, [R4.64] ; /* 0x0000000404047981 */ /* 0x000ea8000c1e1900 */ /*00b0*/ LDG.E R3, [R2.64] ; /* 0x0000000402037981 */ /* 0x000ea2000c1e1900 */ /*00c0*/ IMAD.WIDE R6, R6, R7, c[0x0][0x170] ; /* 0x00005c0006067625 */ /* 0x000fe200078e0207 */ /*00d0*/ IADD3 R9, R4, R3, RZ ; /* 0x0000000304097210 */ /* 0x004fca0007ffe0ff */ /*00e0*/ STG.E [R6.64], R9 ; /* 0x0000000906007986 */ /* 0x000fe2000c101904 */ /*00f0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0100*/ BRA 0x100; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0110*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0120*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0130*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0140*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0150*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0160*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0170*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0180*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0190*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include "stdio.h" #include "time.h" #include <iostream> #include <cuda.h> #include <cuda_runtime.h> #include "device_launch_parameters.h" // Defining number of elements in Array #define N 10000000 // Defining Kernel function for vector addition __global__ void gpuAdd(int *d_a, int *d_b, int *d_c) { // Getting block index of current kernel int tid = blockIdx.x * 1000 + threadIdx.x; if (tid < N) d_c[tid] = d_a[tid] + d_b[tid]; } int main(void) { // Defining host arrays int *h_a, *h_b, *h_c; h_a = (int *)malloc(sizeof(int) * N); h_b = (int *)malloc(sizeof(int) * N); h_c = (int *)malloc(sizeof(int) * N); // Defining device pointers int *d_a, *d_b, *d_c; // Initializing two arrays for addition for (int i = 0; i < N; i++) { h_a[i] = i + 1; h_b[i] = i - 1; } // Allocate the memory cudaMalloc((void **)&d_a, sizeof(int) * N); cudaMalloc((void **)&d_b, sizeof(int) * N); cudaMalloc((void **)&d_c, sizeof(int) * N); // Copy input arrays from host to device memory cudaMemcpy(d_a, h_a, N * sizeof(int), cudaMemcpyHostToDevice); cudaMemcpy(d_b, h_b, N * sizeof(int), cudaMemcpyHostToDevice); // Start Time clock_t start_h = clock(); // Calling kernels with N blocks and one thread per block, passing device pointers as parameters gpuAdd <<<10000, 1000 >>>(d_a, d_b, d_c); cudaThreadSynchronize(); // End Time clock_t end_h = clock(); // Copy result back to host memory from device memory cudaMemcpy(h_c, d_c, N * sizeof(int), cudaMemcpyDeviceToHost); printf("Vector addition on GPU \n"); // Execute Time double time_h = (double)(end_h - start_h) / CLOCKS_PER_SEC; printf("N = %d \n", N); printf("Execute time: %f seconds \n", time_h); // Printing result on console for (int i = 9999990; i < N; i++) { printf("The sum of %d element is %d + %d = %d\n", i, h_a[i], h_b[i], h_c[i]); } // Free up memory cudaFree(d_a); cudaFree(d_b); cudaFree(d_c); system("pause"); return 0; }
.file "tmpxft_0002beec_00000000-6_Add_gpu.cudafe1.cpp" .text #APP .globl _ZSt21ios_base_library_initv #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB3672: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3672: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_ .type _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_, @function _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_: .LFB3694: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movq %fs:40, %rax movq %rax, 120(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 120(%rsp), %rax subq %fs:40, %rax jne .L8 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 152 pushq 40(%rsp) .cfi_def_cfa_offset 160 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z6gpuAddPiS_S_(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE3694: .size _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_, .-_Z29__device_stub__Z6gpuAddPiS_S_PiS_S_ .globl _Z6gpuAddPiS_S_ .type _Z6gpuAddPiS_S_, @function _Z6gpuAddPiS_S_: .LFB3695: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3695: .size _Z6gpuAddPiS_S_, .-_Z6gpuAddPiS_S_ .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "Vector addition on GPU \n" .LC2: .string "N = %d \n" .LC3: .string "Execute time: %f seconds \n" .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC4: .string "The sum of %d element is %d + %d = %d\n" .section .rodata.str1.1 .LC5: .string "pause" .text .globl main .type main, @function main: .LFB3669: .cfi_startproc endbr64 pushq %r14 .cfi_def_cfa_offset 16 .cfi_offset 14, -16 pushq %r13 .cfi_def_cfa_offset 24 .cfi_offset 13, -24 pushq %r12 .cfi_def_cfa_offset 32 .cfi_offset 12, -32 pushq %rbp .cfi_def_cfa_offset 40 .cfi_offset 6, -40 pushq %rbx .cfi_def_cfa_offset 48 .cfi_offset 3, -48 subq $64, %rsp .cfi_def_cfa_offset 112 movq %fs:40, %rax movq %rax, 56(%rsp) xorl %eax, %eax movl $40000000, %edi call malloc@PLT movq %rax, %rbx movl $40000000, %edi call malloc@PLT movq %rax, %rbp movl $40000000, %edi call malloc@PLT movq %rax, %r13 movl $1, %eax .L12: movl %eax, -4(%rbx,%rax,4) leal -2(%rax), %edx movl %edx, -4(%rbp,%rax,4) addq $1, %rax cmpq $10000001, %rax jne .L12 leaq 8(%rsp), %rdi movl $40000000, %esi call cudaMalloc@PLT leaq 16(%rsp), %rdi movl $40000000, %esi call cudaMalloc@PLT leaq 24(%rsp), %rdi movl $40000000, %esi call cudaMalloc@PLT movl $1, %ecx movl $40000000, %edx movq %rbx, %rsi movq 8(%rsp), %rdi call cudaMemcpy@PLT movl $1, %ecx movl $40000000, %edx movq %rbp, %rsi movq 16(%rsp), %rdi call cudaMemcpy@PLT call clock@PLT movq %rax, %r14 movl $1000, 44(%rsp) movl $1, 48(%rsp) movl $10000, 32(%rsp) movl $1, 36(%rsp) movl $0, %r9d movl $0, %r8d movq 44(%rsp), %rdx movl $1, %ecx movq 32(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L19 .L13: call cudaThreadSynchronize@PLT call clock@PLT movq %rax, %r12 movl $2, %ecx movl $40000000, %edx movq 24(%rsp), %rsi movq %r13, %rdi call cudaMemcpy@PLT leaq .LC0(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT subq %r14, %r12 pxor %xmm0, %xmm0 cvtsi2sdq %r12, %xmm0 divsd .LC1(%rip), %xmm0 movq %xmm0, %r12 movl $10000000, %edx leaq .LC2(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq %r12, %xmm0 leaq .LC3(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movl $9999990, %r12d leaq .LC4(%rip), %r14 .L14: movl (%rbx,%r12,4), %ecx movl 0(%r13,%r12,4), %r9d movl 0(%rbp,%r12,4), %r8d movl %r12d, %edx movq %r14, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addq $1, %r12 cmpq $10000000, %r12 jne .L14 movq 8(%rsp), %rdi call cudaFree@PLT movq 16(%rsp), %rdi call cudaFree@PLT movq 24(%rsp), %rdi call cudaFree@PLT leaq .LC5(%rip), %rdi call system@PLT movq 56(%rsp), %rax subq %fs:40, %rax jne .L20 movl $0, %eax addq $64, %rsp .cfi_remember_state .cfi_def_cfa_offset 48 popq %rbx .cfi_def_cfa_offset 40 popq %rbp .cfi_def_cfa_offset 32 popq %r12 .cfi_def_cfa_offset 24 popq %r13 .cfi_def_cfa_offset 16 popq %r14 .cfi_def_cfa_offset 8 ret .L19: .cfi_restore_state movq 24(%rsp), %rdx movq 16(%rsp), %rsi movq 8(%rsp), %rdi call _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_ jmp .L13 .L20: call __stack_chk_fail@PLT .cfi_endproc .LFE3669: .size main, .-main .section .rodata.str1.1 .LC6: .string "_Z6gpuAddPiS_S_" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB3697: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC6(%rip), %rdx movq %rdx, %rcx leaq _Z6gpuAddPiS_S_(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3697: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst8,"aM",@progbits,8 .align 8 .LC1: .long 0 .long 1093567616 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include "stdio.h" #include "time.h" #include <iostream> #include <cuda.h> #include <cuda_runtime.h> #include "device_launch_parameters.h" // Defining number of elements in Array #define N 10000000 // Defining Kernel function for vector addition __global__ void gpuAdd(int *d_a, int *d_b, int *d_c) { // Getting block index of current kernel int tid = blockIdx.x * 1000 + threadIdx.x; if (tid < N) d_c[tid] = d_a[tid] + d_b[tid]; } int main(void) { // Defining host arrays int *h_a, *h_b, *h_c; h_a = (int *)malloc(sizeof(int) * N); h_b = (int *)malloc(sizeof(int) * N); h_c = (int *)malloc(sizeof(int) * N); // Defining device pointers int *d_a, *d_b, *d_c; // Initializing two arrays for addition for (int i = 0; i < N; i++) { h_a[i] = i + 1; h_b[i] = i - 1; } // Allocate the memory cudaMalloc((void **)&d_a, sizeof(int) * N); cudaMalloc((void **)&d_b, sizeof(int) * N); cudaMalloc((void **)&d_c, sizeof(int) * N); // Copy input arrays from host to device memory cudaMemcpy(d_a, h_a, N * sizeof(int), cudaMemcpyHostToDevice); cudaMemcpy(d_b, h_b, N * sizeof(int), cudaMemcpyHostToDevice); // Start Time clock_t start_h = clock(); // Calling kernels with N blocks and one thread per block, passing device pointers as parameters gpuAdd <<<10000, 1000 >>>(d_a, d_b, d_c); cudaThreadSynchronize(); // End Time clock_t end_h = clock(); // Copy result back to host memory from device memory cudaMemcpy(h_c, d_c, N * sizeof(int), cudaMemcpyDeviceToHost); printf("Vector addition on GPU \n"); // Execute Time double time_h = (double)(end_h - start_h) / CLOCKS_PER_SEC; printf("N = %d \n", N); printf("Execute time: %f seconds \n", time_h); // Printing result on console for (int i = 9999990; i < N; i++) { printf("The sum of %d element is %d + %d = %d\n", i, h_a[i], h_b[i], h_c[i]); } // Free up memory cudaFree(d_a); cudaFree(d_b); cudaFree(d_c); system("pause"); return 0; }
#include "stdio.h" #include "time.h" #include <iostream> #include <hip/hip_runtime.h> // Defining number of elements in Array #define N 10000000 // Defining Kernel function for vector addition __global__ void gpuAdd(int *d_a, int *d_b, int *d_c) { // Getting block index of current kernel int tid = blockIdx.x * 1000 + threadIdx.x; if (tid < N) d_c[tid] = d_a[tid] + d_b[tid]; } int main(void) { // Defining host arrays int *h_a, *h_b, *h_c; h_a = (int *)malloc(sizeof(int) * N); h_b = (int *)malloc(sizeof(int) * N); h_c = (int *)malloc(sizeof(int) * N); // Defining device pointers int *d_a, *d_b, *d_c; // Initializing two arrays for addition for (int i = 0; i < N; i++) { h_a[i] = i + 1; h_b[i] = i - 1; } // Allocate the memory hipMalloc((void **)&d_a, sizeof(int) * N); hipMalloc((void **)&d_b, sizeof(int) * N); hipMalloc((void **)&d_c, sizeof(int) * N); // Copy input arrays from host to device memory hipMemcpy(d_a, h_a, N * sizeof(int), hipMemcpyHostToDevice); hipMemcpy(d_b, h_b, N * sizeof(int), hipMemcpyHostToDevice); // Start Time clock_t start_h = clock(); // Calling kernels with N blocks and one thread per block, passing device pointers as parameters gpuAdd <<<10000, 1000 >>>(d_a, d_b, d_c); hipDeviceSynchronize(); // End Time clock_t end_h = clock(); // Copy result back to host memory from device memory hipMemcpy(h_c, d_c, N * sizeof(int), hipMemcpyDeviceToHost); printf("Vector addition on GPU \n"); // Execute Time double time_h = (double)(end_h - start_h) / CLOCKS_PER_SEC; printf("N = %d \n", N); printf("Execute time: %f seconds \n", time_h); // Printing result on console for (int i = 9999990; i < N; i++) { printf("The sum of %d element is %d + %d = %d\n", i, h_a[i], h_b[i], h_c[i]); } // Free up memory hipFree(d_a); hipFree(d_b); hipFree(d_c); system("pause"); return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include "stdio.h" #include "time.h" #include <iostream> #include <hip/hip_runtime.h> // Defining number of elements in Array #define N 10000000 // Defining Kernel function for vector addition __global__ void gpuAdd(int *d_a, int *d_b, int *d_c) { // Getting block index of current kernel int tid = blockIdx.x * 1000 + threadIdx.x; if (tid < N) d_c[tid] = d_a[tid] + d_b[tid]; } int main(void) { // Defining host arrays int *h_a, *h_b, *h_c; h_a = (int *)malloc(sizeof(int) * N); h_b = (int *)malloc(sizeof(int) * N); h_c = (int *)malloc(sizeof(int) * N); // Defining device pointers int *d_a, *d_b, *d_c; // Initializing two arrays for addition for (int i = 0; i < N; i++) { h_a[i] = i + 1; h_b[i] = i - 1; } // Allocate the memory hipMalloc((void **)&d_a, sizeof(int) * N); hipMalloc((void **)&d_b, sizeof(int) * N); hipMalloc((void **)&d_c, sizeof(int) * N); // Copy input arrays from host to device memory hipMemcpy(d_a, h_a, N * sizeof(int), hipMemcpyHostToDevice); hipMemcpy(d_b, h_b, N * sizeof(int), hipMemcpyHostToDevice); // Start Time clock_t start_h = clock(); // Calling kernels with N blocks and one thread per block, passing device pointers as parameters gpuAdd <<<10000, 1000 >>>(d_a, d_b, d_c); hipDeviceSynchronize(); // End Time clock_t end_h = clock(); // Copy result back to host memory from device memory hipMemcpy(h_c, d_c, N * sizeof(int), hipMemcpyDeviceToHost); printf("Vector addition on GPU \n"); // Execute Time double time_h = (double)(end_h - start_h) / CLOCKS_PER_SEC; printf("N = %d \n", N); printf("Execute time: %f seconds \n", time_h); // Printing result on console for (int i = 9999990; i < N; i++) { printf("The sum of %d element is %d + %d = %d\n", i, h_a[i], h_b[i], h_c[i]); } // Free up memory hipFree(d_a); hipFree(d_b); hipFree(d_c); system("pause"); return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z6gpuAddPiS_S_ .globl _Z6gpuAddPiS_S_ .p2align 8 .type _Z6gpuAddPiS_S_,@function _Z6gpuAddPiS_S_: v_mad_u64_u32 v[1:2], null, s15, 0x3e8, v[0:1] s_mov_b32 s2, exec_lo s_delay_alu instid0(VALU_DEP_1) v_cmpx_gt_i32_e32 0x989680, v1 s_cbranch_execz .LBB0_2 s_load_b128 s[4:7], s[0:1], 0x0 v_ashrrev_i32_e32 v2, 31, v1 s_load_b64 s[0:1], s[0:1], 0x10 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[0:1], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v2, vcc_lo, s4, v0 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v3, vcc_lo, s5, v1, vcc_lo v_add_co_u32 v4, vcc_lo, s6, v0 v_add_co_ci_u32_e32 v5, vcc_lo, s7, v1, vcc_lo v_add_co_u32 v0, vcc_lo, s0, v0 global_load_b32 v2, v[2:3], off global_load_b32 v3, v[4:5], off v_add_co_ci_u32_e32 v1, vcc_lo, s1, v1, vcc_lo s_waitcnt vmcnt(0) v_add_nc_u32_e32 v2, v3, v2 global_store_b32 v[0:1], v2, off .LBB0_2: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z6gpuAddPiS_S_ .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 24 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 6 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z6gpuAddPiS_S_, .Lfunc_end0-_Z6gpuAddPiS_S_ .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 24 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z6gpuAddPiS_S_ .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z6gpuAddPiS_S_.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 6 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include "stdio.h" #include "time.h" #include <iostream> #include <hip/hip_runtime.h> // Defining number of elements in Array #define N 10000000 // Defining Kernel function for vector addition __global__ void gpuAdd(int *d_a, int *d_b, int *d_c) { // Getting block index of current kernel int tid = blockIdx.x * 1000 + threadIdx.x; if (tid < N) d_c[tid] = d_a[tid] + d_b[tid]; } int main(void) { // Defining host arrays int *h_a, *h_b, *h_c; h_a = (int *)malloc(sizeof(int) * N); h_b = (int *)malloc(sizeof(int) * N); h_c = (int *)malloc(sizeof(int) * N); // Defining device pointers int *d_a, *d_b, *d_c; // Initializing two arrays for addition for (int i = 0; i < N; i++) { h_a[i] = i + 1; h_b[i] = i - 1; } // Allocate the memory hipMalloc((void **)&d_a, sizeof(int) * N); hipMalloc((void **)&d_b, sizeof(int) * N); hipMalloc((void **)&d_c, sizeof(int) * N); // Copy input arrays from host to device memory hipMemcpy(d_a, h_a, N * sizeof(int), hipMemcpyHostToDevice); hipMemcpy(d_b, h_b, N * sizeof(int), hipMemcpyHostToDevice); // Start Time clock_t start_h = clock(); // Calling kernels with N blocks and one thread per block, passing device pointers as parameters gpuAdd <<<10000, 1000 >>>(d_a, d_b, d_c); hipDeviceSynchronize(); // End Time clock_t end_h = clock(); // Copy result back to host memory from device memory hipMemcpy(h_c, d_c, N * sizeof(int), hipMemcpyDeviceToHost); printf("Vector addition on GPU \n"); // Execute Time double time_h = (double)(end_h - start_h) / CLOCKS_PER_SEC; printf("N = %d \n", N); printf("Execute time: %f seconds \n", time_h); // Printing result on console for (int i = 9999990; i < N; i++) { printf("The sum of %d element is %d + %d = %d\n", i, h_a[i], h_b[i], h_c[i]); } // Free up memory hipFree(d_a); hipFree(d_b); hipFree(d_c); system("pause"); return 0; }
.text .file "Add_gpu.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl _Z21__device_stub__gpuAddPiS_S_ # -- Begin function _Z21__device_stub__gpuAddPiS_S_ .p2align 4, 0x90 .type _Z21__device_stub__gpuAddPiS_S_,@function _Z21__device_stub__gpuAddPiS_S_: # @_Z21__device_stub__gpuAddPiS_S_ .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movq %rdx, 56(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z6gpuAddPiS_S_, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $120, %rsp .cfi_adjust_cfa_offset -120 retq .Lfunc_end0: .size _Z21__device_stub__gpuAddPiS_S_, .Lfunc_end0-_Z21__device_stub__gpuAddPiS_S_ .cfi_endproc # -- End function .section .rodata.cst8,"aM",@progbits,8 .p2align 3, 0x0 # -- Begin function main .LCPI1_0: .quad 0x412e848000000000 # double 1.0E+6 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %r15 .cfi_def_cfa_offset 16 pushq %r14 .cfi_def_cfa_offset 24 pushq %r13 .cfi_def_cfa_offset 32 pushq %r12 .cfi_def_cfa_offset 40 pushq %rbx .cfi_def_cfa_offset 48 subq $144, %rsp .cfi_def_cfa_offset 192 .cfi_offset %rbx, -48 .cfi_offset %r12, -40 .cfi_offset %r13, -32 .cfi_offset %r14, -24 .cfi_offset %r15, -16 movl $40000000, %edi # imm = 0x2625A00 callq malloc movq %rax, %rbx movl $40000000, %edi # imm = 0x2625A00 callq malloc movq %rax, %r14 movl $40000000, %edi # imm = 0x2625A00 callq malloc movq %rax, %r15 xorl %eax, %eax .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 leaq 1(%rax), %rcx movl %ecx, (%rbx,%rax,4) leal -1(%rax), %edx movl %edx, (%r14,%rax,4) movq %rcx, %rax cmpq $10000000, %rcx # imm = 0x989680 jne .LBB1_1 # %bb.2: leaq 24(%rsp), %rdi movl $40000000, %esi # imm = 0x2625A00 callq hipMalloc leaq 16(%rsp), %rdi movl $40000000, %esi # imm = 0x2625A00 callq hipMalloc leaq 8(%rsp), %rdi movl $40000000, %esi # imm = 0x2625A00 callq hipMalloc movq 24(%rsp), %rdi movl $40000000, %edx # imm = 0x2625A00 movq %rbx, %rsi movl $1, %ecx callq hipMemcpy movq 16(%rsp), %rdi movl $40000000, %edx # imm = 0x2625A00 movq %r14, %rsi movl $1, %ecx callq hipMemcpy callq clock movq %rax, %r12 movabsq $4294968296, %rdx # imm = 0x1000003E8 leaq 9000(%rdx), %rdi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_4 # %bb.3: movq 24(%rsp), %rax movq 16(%rsp), %rcx movq 8(%rsp), %rdx movq %rax, 104(%rsp) movq %rcx, 96(%rsp) movq %rdx, 88(%rsp) leaq 104(%rsp), %rax movq %rax, 112(%rsp) leaq 96(%rsp), %rax movq %rax, 120(%rsp) leaq 88(%rsp), %rax movq %rax, 128(%rsp) leaq 72(%rsp), %rdi leaq 56(%rsp), %rsi leaq 48(%rsp), %rdx leaq 40(%rsp), %rcx callq __hipPopCallConfiguration movq 72(%rsp), %rsi movl 80(%rsp), %edx movq 56(%rsp), %rcx movl 64(%rsp), %r8d leaq 112(%rsp), %r9 movl $_Z6gpuAddPiS_S_, %edi pushq 40(%rsp) .cfi_adjust_cfa_offset 8 pushq 56(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_4: callq hipDeviceSynchronize callq clock movq %rax, %r13 movq 8(%rsp), %rsi movl $40000000, %edx # imm = 0x2625A00 movq %r15, %rdi movl $2, %ecx callq hipMemcpy movl $.Lstr, %edi callq puts@PLT subq %r12, %r13 cvtsi2sd %r13, %xmm0 divsd .LCPI1_0(%rip), %xmm0 movsd %xmm0, 32(%rsp) # 8-byte Spill movl $.L.str.1, %edi movl $10000000, %esi # imm = 0x989680 xorl %eax, %eax callq printf movl $.L.str.2, %edi movsd 32(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero movb $1, %al callq printf movl $9999990, %r12d # imm = 0x989676 .p2align 4, 0x90 .LBB1_5: # =>This Inner Loop Header: Depth=1 movl (%rbx,%r12,4), %edx movl (%r14,%r12,4), %ecx movl (%r15,%r12,4), %r8d movl $.L.str.3, %edi movl %r12d, %esi xorl %eax, %eax callq printf incq %r12 cmpq $10000000, %r12 # imm = 0x989680 jne .LBB1_5 # %bb.6: movq 24(%rsp), %rdi callq hipFree movq 16(%rsp), %rdi callq hipFree movq 8(%rsp), %rdi callq hipFree movl $.L.str.4, %edi callq system xorl %eax, %eax addq $144, %rsp .cfi_def_cfa_offset 48 popq %rbx .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z6gpuAddPiS_S_, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z6gpuAddPiS_S_,@object # @_Z6gpuAddPiS_S_ .section .rodata,"a",@progbits .globl _Z6gpuAddPiS_S_ .p2align 3, 0x0 _Z6gpuAddPiS_S_: .quad _Z21__device_stub__gpuAddPiS_S_ .size _Z6gpuAddPiS_S_, 8 .type .L.str.1,@object # @.str.1 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.1: .asciz "N = %d \n" .size .L.str.1, 9 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "Execute time: %f seconds \n" .size .L.str.2, 27 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "The sum of %d element is %d + %d = %d\n" .size .L.str.3, 39 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "pause" .size .L.str.4, 6 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z6gpuAddPiS_S_" .size .L__unnamed_1, 16 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "Vector addition on GPU " .size .Lstr, 24 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z21__device_stub__gpuAddPiS_S_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z6gpuAddPiS_S_ .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z6gpuAddPiS_S_ .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R6, SR_CTAID.X ; /* 0x0000000000067919 */ /* 0x000e280000002500 */ /*0020*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e240000002100 */ /*0030*/ IMAD R6, R6, 0x3e8, R3 ; /* 0x000003e806067824 */ /* 0x001fca00078e0203 */ /*0040*/ ISETP.GT.AND P0, PT, R6, 0x98967f, PT ; /* 0x0098967f0600780c */ /* 0x000fda0003f04270 */ /*0050*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0060*/ HFMA2.MMA R7, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff077435 */ /* 0x000fe200000001ff */ /*0070*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd20000000a00 */ /*0080*/ IMAD.WIDE R4, R6, R7, c[0x0][0x168] ; /* 0x00005a0006047625 */ /* 0x000fc800078e0207 */ /*0090*/ IMAD.WIDE R2, R6.reuse, R7.reuse, c[0x0][0x160] ; /* 0x0000580006027625 */ /* 0x0c0fe400078e0207 */ /*00a0*/ LDG.E R4, [R4.64] ; /* 0x0000000404047981 */ /* 0x000ea8000c1e1900 */ /*00b0*/ LDG.E R3, [R2.64] ; /* 0x0000000402037981 */ /* 0x000ea2000c1e1900 */ /*00c0*/ IMAD.WIDE R6, R6, R7, c[0x0][0x170] ; /* 0x00005c0006067625 */ /* 0x000fe200078e0207 */ /*00d0*/ IADD3 R9, R4, R3, RZ ; /* 0x0000000304097210 */ /* 0x004fca0007ffe0ff */ /*00e0*/ STG.E [R6.64], R9 ; /* 0x0000000906007986 */ /* 0x000fe2000c101904 */ /*00f0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0100*/ BRA 0x100; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0110*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0120*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0130*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0140*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0150*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0160*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0170*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0180*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0190*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*01f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z6gpuAddPiS_S_ .globl _Z6gpuAddPiS_S_ .p2align 8 .type _Z6gpuAddPiS_S_,@function _Z6gpuAddPiS_S_: v_mad_u64_u32 v[1:2], null, s15, 0x3e8, v[0:1] s_mov_b32 s2, exec_lo s_delay_alu instid0(VALU_DEP_1) v_cmpx_gt_i32_e32 0x989680, v1 s_cbranch_execz .LBB0_2 s_load_b128 s[4:7], s[0:1], 0x0 v_ashrrev_i32_e32 v2, 31, v1 s_load_b64 s[0:1], s[0:1], 0x10 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[0:1], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v2, vcc_lo, s4, v0 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v3, vcc_lo, s5, v1, vcc_lo v_add_co_u32 v4, vcc_lo, s6, v0 v_add_co_ci_u32_e32 v5, vcc_lo, s7, v1, vcc_lo v_add_co_u32 v0, vcc_lo, s0, v0 global_load_b32 v2, v[2:3], off global_load_b32 v3, v[4:5], off v_add_co_ci_u32_e32 v1, vcc_lo, s1, v1, vcc_lo s_waitcnt vmcnt(0) v_add_nc_u32_e32 v2, v3, v2 global_store_b32 v[0:1], v2, off .LBB0_2: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z6gpuAddPiS_S_ .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 24 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 6 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z6gpuAddPiS_S_, .Lfunc_end0-_Z6gpuAddPiS_S_ .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 8 .size: 8 .value_kind: global_buffer - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 24 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z6gpuAddPiS_S_ .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z6gpuAddPiS_S_.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 6 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_0002beec_00000000-6_Add_gpu.cudafe1.cpp" .text #APP .globl _ZSt21ios_base_library_initv #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB3672: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3672: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_ .type _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_, @function _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_: .LFB3694: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movq %rdx, 8(%rsp) movq %fs:40, %rax movq %rax, 120(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 8(%rsp), %rax movq %rax, 112(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 120(%rsp), %rax subq %fs:40, %rax jne .L8 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 152 pushq 40(%rsp) .cfi_def_cfa_offset 160 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z6gpuAddPiS_S_(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE3694: .size _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_, .-_Z29__device_stub__Z6gpuAddPiS_S_PiS_S_ .globl _Z6gpuAddPiS_S_ .type _Z6gpuAddPiS_S_, @function _Z6gpuAddPiS_S_: .LFB3695: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3695: .size _Z6gpuAddPiS_S_, .-_Z6gpuAddPiS_S_ .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "Vector addition on GPU \n" .LC2: .string "N = %d \n" .LC3: .string "Execute time: %f seconds \n" .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC4: .string "The sum of %d element is %d + %d = %d\n" .section .rodata.str1.1 .LC5: .string "pause" .text .globl main .type main, @function main: .LFB3669: .cfi_startproc endbr64 pushq %r14 .cfi_def_cfa_offset 16 .cfi_offset 14, -16 pushq %r13 .cfi_def_cfa_offset 24 .cfi_offset 13, -24 pushq %r12 .cfi_def_cfa_offset 32 .cfi_offset 12, -32 pushq %rbp .cfi_def_cfa_offset 40 .cfi_offset 6, -40 pushq %rbx .cfi_def_cfa_offset 48 .cfi_offset 3, -48 subq $64, %rsp .cfi_def_cfa_offset 112 movq %fs:40, %rax movq %rax, 56(%rsp) xorl %eax, %eax movl $40000000, %edi call malloc@PLT movq %rax, %rbx movl $40000000, %edi call malloc@PLT movq %rax, %rbp movl $40000000, %edi call malloc@PLT movq %rax, %r13 movl $1, %eax .L12: movl %eax, -4(%rbx,%rax,4) leal -2(%rax), %edx movl %edx, -4(%rbp,%rax,4) addq $1, %rax cmpq $10000001, %rax jne .L12 leaq 8(%rsp), %rdi movl $40000000, %esi call cudaMalloc@PLT leaq 16(%rsp), %rdi movl $40000000, %esi call cudaMalloc@PLT leaq 24(%rsp), %rdi movl $40000000, %esi call cudaMalloc@PLT movl $1, %ecx movl $40000000, %edx movq %rbx, %rsi movq 8(%rsp), %rdi call cudaMemcpy@PLT movl $1, %ecx movl $40000000, %edx movq %rbp, %rsi movq 16(%rsp), %rdi call cudaMemcpy@PLT call clock@PLT movq %rax, %r14 movl $1000, 44(%rsp) movl $1, 48(%rsp) movl $10000, 32(%rsp) movl $1, 36(%rsp) movl $0, %r9d movl $0, %r8d movq 44(%rsp), %rdx movl $1, %ecx movq 32(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L19 .L13: call cudaThreadSynchronize@PLT call clock@PLT movq %rax, %r12 movl $2, %ecx movl $40000000, %edx movq 24(%rsp), %rsi movq %r13, %rdi call cudaMemcpy@PLT leaq .LC0(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT subq %r14, %r12 pxor %xmm0, %xmm0 cvtsi2sdq %r12, %xmm0 divsd .LC1(%rip), %xmm0 movq %xmm0, %r12 movl $10000000, %edx leaq .LC2(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq %r12, %xmm0 leaq .LC3(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movl $9999990, %r12d leaq .LC4(%rip), %r14 .L14: movl (%rbx,%r12,4), %ecx movl 0(%r13,%r12,4), %r9d movl 0(%rbp,%r12,4), %r8d movl %r12d, %edx movq %r14, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addq $1, %r12 cmpq $10000000, %r12 jne .L14 movq 8(%rsp), %rdi call cudaFree@PLT movq 16(%rsp), %rdi call cudaFree@PLT movq 24(%rsp), %rdi call cudaFree@PLT leaq .LC5(%rip), %rdi call system@PLT movq 56(%rsp), %rax subq %fs:40, %rax jne .L20 movl $0, %eax addq $64, %rsp .cfi_remember_state .cfi_def_cfa_offset 48 popq %rbx .cfi_def_cfa_offset 40 popq %rbp .cfi_def_cfa_offset 32 popq %r12 .cfi_def_cfa_offset 24 popq %r13 .cfi_def_cfa_offset 16 popq %r14 .cfi_def_cfa_offset 8 ret .L19: .cfi_restore_state movq 24(%rsp), %rdx movq 16(%rsp), %rsi movq 8(%rsp), %rdi call _Z29__device_stub__Z6gpuAddPiS_S_PiS_S_ jmp .L13 .L20: call __stack_chk_fail@PLT .cfi_endproc .LFE3669: .size main, .-main .section .rodata.str1.1 .LC6: .string "_Z6gpuAddPiS_S_" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB3697: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC6(%rip), %rdx movq %rdx, %rcx leaq _Z6gpuAddPiS_S_(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3697: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .section .rodata.cst8,"aM",@progbits,8 .align 8 .LC1: .long 0 .long 1093567616 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "Add_gpu.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl _Z21__device_stub__gpuAddPiS_S_ # -- Begin function _Z21__device_stub__gpuAddPiS_S_ .p2align 4, 0x90 .type _Z21__device_stub__gpuAddPiS_S_,@function _Z21__device_stub__gpuAddPiS_S_: # @_Z21__device_stub__gpuAddPiS_S_ .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movq %rdx, 56(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 56(%rsp), %rax movq %rax, 96(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z6gpuAddPiS_S_, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $120, %rsp .cfi_adjust_cfa_offset -120 retq .Lfunc_end0: .size _Z21__device_stub__gpuAddPiS_S_, .Lfunc_end0-_Z21__device_stub__gpuAddPiS_S_ .cfi_endproc # -- End function .section .rodata.cst8,"aM",@progbits,8 .p2align 3, 0x0 # -- Begin function main .LCPI1_0: .quad 0x412e848000000000 # double 1.0E+6 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .cfi_startproc # %bb.0: pushq %r15 .cfi_def_cfa_offset 16 pushq %r14 .cfi_def_cfa_offset 24 pushq %r13 .cfi_def_cfa_offset 32 pushq %r12 .cfi_def_cfa_offset 40 pushq %rbx .cfi_def_cfa_offset 48 subq $144, %rsp .cfi_def_cfa_offset 192 .cfi_offset %rbx, -48 .cfi_offset %r12, -40 .cfi_offset %r13, -32 .cfi_offset %r14, -24 .cfi_offset %r15, -16 movl $40000000, %edi # imm = 0x2625A00 callq malloc movq %rax, %rbx movl $40000000, %edi # imm = 0x2625A00 callq malloc movq %rax, %r14 movl $40000000, %edi # imm = 0x2625A00 callq malloc movq %rax, %r15 xorl %eax, %eax .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 leaq 1(%rax), %rcx movl %ecx, (%rbx,%rax,4) leal -1(%rax), %edx movl %edx, (%r14,%rax,4) movq %rcx, %rax cmpq $10000000, %rcx # imm = 0x989680 jne .LBB1_1 # %bb.2: leaq 24(%rsp), %rdi movl $40000000, %esi # imm = 0x2625A00 callq hipMalloc leaq 16(%rsp), %rdi movl $40000000, %esi # imm = 0x2625A00 callq hipMalloc leaq 8(%rsp), %rdi movl $40000000, %esi # imm = 0x2625A00 callq hipMalloc movq 24(%rsp), %rdi movl $40000000, %edx # imm = 0x2625A00 movq %rbx, %rsi movl $1, %ecx callq hipMemcpy movq 16(%rsp), %rdi movl $40000000, %edx # imm = 0x2625A00 movq %r14, %rsi movl $1, %ecx callq hipMemcpy callq clock movq %rax, %r12 movabsq $4294968296, %rdx # imm = 0x1000003E8 leaq 9000(%rdx), %rdi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_4 # %bb.3: movq 24(%rsp), %rax movq 16(%rsp), %rcx movq 8(%rsp), %rdx movq %rax, 104(%rsp) movq %rcx, 96(%rsp) movq %rdx, 88(%rsp) leaq 104(%rsp), %rax movq %rax, 112(%rsp) leaq 96(%rsp), %rax movq %rax, 120(%rsp) leaq 88(%rsp), %rax movq %rax, 128(%rsp) leaq 72(%rsp), %rdi leaq 56(%rsp), %rsi leaq 48(%rsp), %rdx leaq 40(%rsp), %rcx callq __hipPopCallConfiguration movq 72(%rsp), %rsi movl 80(%rsp), %edx movq 56(%rsp), %rcx movl 64(%rsp), %r8d leaq 112(%rsp), %r9 movl $_Z6gpuAddPiS_S_, %edi pushq 40(%rsp) .cfi_adjust_cfa_offset 8 pushq 56(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_4: callq hipDeviceSynchronize callq clock movq %rax, %r13 movq 8(%rsp), %rsi movl $40000000, %edx # imm = 0x2625A00 movq %r15, %rdi movl $2, %ecx callq hipMemcpy movl $.Lstr, %edi callq puts@PLT subq %r12, %r13 cvtsi2sd %r13, %xmm0 divsd .LCPI1_0(%rip), %xmm0 movsd %xmm0, 32(%rsp) # 8-byte Spill movl $.L.str.1, %edi movl $10000000, %esi # imm = 0x989680 xorl %eax, %eax callq printf movl $.L.str.2, %edi movsd 32(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero movb $1, %al callq printf movl $9999990, %r12d # imm = 0x989676 .p2align 4, 0x90 .LBB1_5: # =>This Inner Loop Header: Depth=1 movl (%rbx,%r12,4), %edx movl (%r14,%r12,4), %ecx movl (%r15,%r12,4), %r8d movl $.L.str.3, %edi movl %r12d, %esi xorl %eax, %eax callq printf incq %r12 cmpq $10000000, %r12 # imm = 0x989680 jne .LBB1_5 # %bb.6: movq 24(%rsp), %rdi callq hipFree movq 16(%rsp), %rdi callq hipFree movq 8(%rsp), %rdi callq hipFree movl $.L.str.4, %edi callq system xorl %eax, %eax addq $144, %rsp .cfi_def_cfa_offset 48 popq %rbx .cfi_def_cfa_offset 40 popq %r12 .cfi_def_cfa_offset 32 popq %r13 .cfi_def_cfa_offset 24 popq %r14 .cfi_def_cfa_offset 16 popq %r15 .cfi_def_cfa_offset 8 retq .Lfunc_end1: .size main, .Lfunc_end1-main .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB2_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB2_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z6gpuAddPiS_S_, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end2: .size __hip_module_ctor, .Lfunc_end2-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB3_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB3_2: retq .Lfunc_end3: .size __hip_module_dtor, .Lfunc_end3-__hip_module_dtor .cfi_endproc # -- End function .type _Z6gpuAddPiS_S_,@object # @_Z6gpuAddPiS_S_ .section .rodata,"a",@progbits .globl _Z6gpuAddPiS_S_ .p2align 3, 0x0 _Z6gpuAddPiS_S_: .quad _Z21__device_stub__gpuAddPiS_S_ .size _Z6gpuAddPiS_S_, 8 .type .L.str.1,@object # @.str.1 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.1: .asciz "N = %d \n" .size .L.str.1, 9 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "Execute time: %f seconds \n" .size .L.str.2, 27 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "The sum of %d element is %d + %d = %d\n" .size .L.str.3, 39 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "pause" .size .L.str.4, 6 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z6gpuAddPiS_S_" .size .L__unnamed_1, 16 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "Vector addition on GPU " .size .Lstr, 24 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z21__device_stub__gpuAddPiS_S_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z6gpuAddPiS_S_ .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
#include "includes.h" __global__ void kernelInitNablaW(float *nabla_w,int tws) { if ((blockIdx.x*blockDim.x+threadIdx.x)<tws) { nabla_w[blockIdx.x*blockDim.x+threadIdx.x]=0.0; } }
code for sm_80 Function : _Z16kernelInitNablaWPfi .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R2, SR_CTAID.X ; /* 0x0000000000027919 */ /* 0x000e280000002500 */ /*0020*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e240000002100 */ /*0030*/ IMAD R2, R2, c[0x0][0x0], R3 ; /* 0x0000000002027a24 */ /* 0x001fca00078e0203 */ /*0040*/ ISETP.GE.U32.AND P0, PT, R2, c[0x0][0x168], PT ; /* 0x00005a0002007a0c */ /* 0x000fda0003f06070 */ /*0050*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0060*/ HFMA2.MMA R3, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff037435 */ /* 0x000fe200000001ff */ /*0070*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd20000000a00 */ /*0080*/ IMAD.WIDE.U32 R2, R2, R3, c[0x0][0x160] ; /* 0x0000580002027625 */ /* 0x000fca00078e0003 */ /*0090*/ STG.E [R2.64], RZ ; /* 0x000000ff02007986 */ /* 0x000fe2000c101904 */ /*00a0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00b0*/ BRA 0xb0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0100*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0110*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0120*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0130*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0140*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0150*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0160*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0170*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include "includes.h" __global__ void kernelInitNablaW(float *nabla_w,int tws) { if ((blockIdx.x*blockDim.x+threadIdx.x)<tws) { nabla_w[blockIdx.x*blockDim.x+threadIdx.x]=0.0; } }
.file "tmpxft_0001a24f_00000000-6_kernelInitNablaW.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2029: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2029: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z37__device_stub__Z16kernelInitNablaWPfiPfi .type _Z37__device_stub__Z16kernelInitNablaWPfiPfi, @function _Z37__device_stub__Z16kernelInitNablaWPfiPfi: .LFB2051: .cfi_startproc endbr64 subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 8(%rsp) movl %esi, 4(%rsp) movq %fs:40, %rax movq %rax, 104(%rsp) xorl %eax, %eax leaq 8(%rsp), %rax movq %rax, 80(%rsp) leaq 4(%rsp), %rax movq %rax, 88(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) leaq 24(%rsp), %rcx leaq 16(%rsp), %rdx leaq 44(%rsp), %rsi leaq 32(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 104(%rsp), %rax subq %fs:40, %rax jne .L8 addq $120, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 24(%rsp) .cfi_def_cfa_offset 136 pushq 24(%rsp) .cfi_def_cfa_offset 144 leaq 96(%rsp), %r9 movq 60(%rsp), %rcx movl 68(%rsp), %r8d movq 48(%rsp), %rsi movl 56(%rsp), %edx leaq _Z16kernelInitNablaWPfi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 128 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2051: .size _Z37__device_stub__Z16kernelInitNablaWPfiPfi, .-_Z37__device_stub__Z16kernelInitNablaWPfiPfi .globl _Z16kernelInitNablaWPfi .type _Z16kernelInitNablaWPfi, @function _Z16kernelInitNablaWPfi: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z37__device_stub__Z16kernelInitNablaWPfiPfi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z16kernelInitNablaWPfi, .-_Z16kernelInitNablaWPfi .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z16kernelInitNablaWPfi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2054: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z16kernelInitNablaWPfi(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2054: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include "includes.h" __global__ void kernelInitNablaW(float *nabla_w,int tws) { if ((blockIdx.x*blockDim.x+threadIdx.x)<tws) { nabla_w[blockIdx.x*blockDim.x+threadIdx.x]=0.0; } }
#include <hip/hip_runtime.h> #include "includes.h" __global__ void kernelInitNablaW(float *nabla_w,int tws) { if ((blockIdx.x*blockDim.x+threadIdx.x)<tws) { nabla_w[blockIdx.x*blockDim.x+threadIdx.x]=0.0; } }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> #include "includes.h" __global__ void kernelInitNablaW(float *nabla_w,int tws) { if ((blockIdx.x*blockDim.x+threadIdx.x)<tws) { nabla_w[blockIdx.x*blockDim.x+threadIdx.x]=0.0; } }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z16kernelInitNablaWPfi .globl _Z16kernelInitNablaWPfi .p2align 8 .type _Z16kernelInitNablaWPfi,@function _Z16kernelInitNablaWPfi: s_clause 0x1 s_load_b32 s2, s[0:1], 0x1c s_load_b32 s3, s[0:1], 0x8 s_waitcnt lgkmcnt(0) s_and_b32 s2, s2, 0xffff s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_mad_u64_u32 v[1:2], null, s15, s2, v[0:1] s_mov_b32 s2, exec_lo v_cmpx_gt_u32_e64 s3, v1 s_cbranch_execz .LBB0_2 s_load_b64 s[0:1], s[0:1], 0x0 v_mov_b32_e32 v2, 0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[0:1], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v0, vcc_lo, s0, v0 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v1, vcc_lo, s1, v1, vcc_lo global_store_b32 v[0:1], v2, off .LBB0_2: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z16kernelInitNablaWPfi .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 272 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 3 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z16kernelInitNablaWPfi, .Lfunc_end0-_Z16kernelInitNablaWPfi .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .offset: 8 .size: 4 .value_kind: by_value - .offset: 16 .size: 4 .value_kind: hidden_block_count_x - .offset: 20 .size: 4 .value_kind: hidden_block_count_y - .offset: 24 .size: 4 .value_kind: hidden_block_count_z - .offset: 28 .size: 2 .value_kind: hidden_group_size_x - .offset: 30 .size: 2 .value_kind: hidden_group_size_y - .offset: 32 .size: 2 .value_kind: hidden_group_size_z - .offset: 34 .size: 2 .value_kind: hidden_remainder_x - .offset: 36 .size: 2 .value_kind: hidden_remainder_y - .offset: 38 .size: 2 .value_kind: hidden_remainder_z - .offset: 56 .size: 8 .value_kind: hidden_global_offset_x - .offset: 64 .size: 8 .value_kind: hidden_global_offset_y - .offset: 72 .size: 8 .value_kind: hidden_global_offset_z - .offset: 80 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 272 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z16kernelInitNablaWPfi .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z16kernelInitNablaWPfi.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 3 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <hip/hip_runtime.h> #include "includes.h" __global__ void kernelInitNablaW(float *nabla_w,int tws) { if ((blockIdx.x*blockDim.x+threadIdx.x)<tws) { nabla_w[blockIdx.x*blockDim.x+threadIdx.x]=0.0; } }
.text .file "kernelInitNablaW.hip" .globl _Z31__device_stub__kernelInitNablaWPfi # -- Begin function _Z31__device_stub__kernelInitNablaWPfi .p2align 4, 0x90 .type _Z31__device_stub__kernelInitNablaWPfi,@function _Z31__device_stub__kernelInitNablaWPfi: # @_Z31__device_stub__kernelInitNablaWPfi .cfi_startproc # %bb.0: subq $88, %rsp .cfi_def_cfa_offset 96 movq %rdi, 56(%rsp) movl %esi, 4(%rsp) leaq 56(%rsp), %rax movq %rax, 64(%rsp) leaq 4(%rsp), %rax movq %rax, 72(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 64(%rsp), %r9 movl $_Z16kernelInitNablaWPfi, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $104, %rsp .cfi_adjust_cfa_offset -104 retq .Lfunc_end0: .size _Z31__device_stub__kernelInitNablaWPfi, .Lfunc_end0-_Z31__device_stub__kernelInitNablaWPfi .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z16kernelInitNablaWPfi, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type _Z16kernelInitNablaWPfi,@object # @_Z16kernelInitNablaWPfi .section .rodata,"a",@progbits .globl _Z16kernelInitNablaWPfi .p2align 3, 0x0 _Z16kernelInitNablaWPfi: .quad _Z31__device_stub__kernelInitNablaWPfi .size _Z16kernelInitNablaWPfi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z16kernelInitNablaWPfi" .size .L__unnamed_1, 24 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z31__device_stub__kernelInitNablaWPfi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z16kernelInitNablaWPfi .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA device assembly to AMD device assembly.
code for sm_80 Function : _Z16kernelInitNablaWPfi .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ MOV R1, c[0x0][0x28] ; /* 0x00000a0000017a02 */ /* 0x000fe40000000f00 */ /*0010*/ S2R R2, SR_CTAID.X ; /* 0x0000000000027919 */ /* 0x000e280000002500 */ /*0020*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e240000002100 */ /*0030*/ IMAD R2, R2, c[0x0][0x0], R3 ; /* 0x0000000002027a24 */ /* 0x001fca00078e0203 */ /*0040*/ ISETP.GE.U32.AND P0, PT, R2, c[0x0][0x168], PT ; /* 0x00005a0002007a0c */ /* 0x000fda0003f06070 */ /*0050*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0060*/ HFMA2.MMA R3, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff037435 */ /* 0x000fe200000001ff */ /*0070*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd20000000a00 */ /*0080*/ IMAD.WIDE.U32 R2, R2, R3, c[0x0][0x160] ; /* 0x0000580002027625 */ /* 0x000fca00078e0003 */ /*0090*/ STG.E [R2.64], RZ ; /* 0x000000ff02007986 */ /* 0x000fe2000c101904 */ /*00a0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00b0*/ BRA 0xb0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*00c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*00f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0100*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0110*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0120*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0130*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0140*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0150*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0160*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0170*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z16kernelInitNablaWPfi .globl _Z16kernelInitNablaWPfi .p2align 8 .type _Z16kernelInitNablaWPfi,@function _Z16kernelInitNablaWPfi: s_clause 0x1 s_load_b32 s2, s[0:1], 0x1c s_load_b32 s3, s[0:1], 0x8 s_waitcnt lgkmcnt(0) s_and_b32 s2, s2, 0xffff s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_mad_u64_u32 v[1:2], null, s15, s2, v[0:1] s_mov_b32 s2, exec_lo v_cmpx_gt_u32_e64 s3, v1 s_cbranch_execz .LBB0_2 s_load_b64 s[0:1], s[0:1], 0x0 v_mov_b32_e32 v2, 0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[0:1], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v0, vcc_lo, s0, v0 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v1, vcc_lo, s1, v1, vcc_lo global_store_b32 v[0:1], v2, off .LBB0_2: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z16kernelInitNablaWPfi .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 272 .amdhsa_user_sgpr_count 15 .amdhsa_user_sgpr_dispatch_ptr 0 .amdhsa_user_sgpr_queue_ptr 0 .amdhsa_user_sgpr_kernarg_segment_ptr 1 .amdhsa_user_sgpr_dispatch_id 0 .amdhsa_user_sgpr_private_segment_size 0 .amdhsa_wavefront_size32 1 .amdhsa_uses_dynamic_stack 0 .amdhsa_enable_private_segment 0 .amdhsa_system_sgpr_workgroup_id_x 1 .amdhsa_system_sgpr_workgroup_id_y 0 .amdhsa_system_sgpr_workgroup_id_z 0 .amdhsa_system_sgpr_workgroup_info 0 .amdhsa_system_vgpr_workitem_id 0 .amdhsa_next_free_vgpr 3 .amdhsa_next_free_sgpr 16 .amdhsa_float_round_mode_32 0 .amdhsa_float_round_mode_16_64 0 .amdhsa_float_denorm_mode_32 3 .amdhsa_float_denorm_mode_16_64 3 .amdhsa_dx10_clamp 1 .amdhsa_ieee_mode 1 .amdhsa_fp16_overflow 0 .amdhsa_workgroup_processor_mode 1 .amdhsa_memory_ordered 1 .amdhsa_forward_progress 0 .amdhsa_shared_vgpr_count 0 .amdhsa_exception_fp_ieee_invalid_op 0 .amdhsa_exception_fp_denorm_src 0 .amdhsa_exception_fp_ieee_div_zero 0 .amdhsa_exception_fp_ieee_overflow 0 .amdhsa_exception_fp_ieee_underflow 0 .amdhsa_exception_fp_ieee_inexact 0 .amdhsa_exception_int_div_zero 0 .end_amdhsa_kernel .text .Lfunc_end0: .size _Z16kernelInitNablaWPfi, .Lfunc_end0-_Z16kernelInitNablaWPfi .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .type __hip_cuid_,@object .section .bss,"aw",@nobits .globl __hip_cuid_ __hip_cuid_: .byte 0 .size __hip_cuid_, 1 .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym __hip_cuid_ .amdgpu_metadata --- amdhsa.kernels: - .args: - .address_space: global .offset: 0 .size: 8 .value_kind: global_buffer - .offset: 8 .size: 4 .value_kind: by_value - .offset: 16 .size: 4 .value_kind: hidden_block_count_x - .offset: 20 .size: 4 .value_kind: hidden_block_count_y - .offset: 24 .size: 4 .value_kind: hidden_block_count_z - .offset: 28 .size: 2 .value_kind: hidden_group_size_x - .offset: 30 .size: 2 .value_kind: hidden_group_size_y - .offset: 32 .size: 2 .value_kind: hidden_group_size_z - .offset: 34 .size: 2 .value_kind: hidden_remainder_x - .offset: 36 .size: 2 .value_kind: hidden_remainder_y - .offset: 38 .size: 2 .value_kind: hidden_remainder_z - .offset: 56 .size: 8 .value_kind: hidden_global_offset_x - .offset: 64 .size: 8 .value_kind: hidden_global_offset_y - .offset: 72 .size: 8 .value_kind: hidden_global_offset_z - .offset: 80 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 272 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z16kernelInitNablaWPfi .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z16kernelInitNablaWPfi.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 3 .vgpr_spill_count: 0 .wavefront_size: 32 .workgroup_processor_mode: 1 amdhsa.target: amdgcn-amd-amdhsa--gfx1100 amdhsa.version: - 1 - 2 ... .end_amdgpu_metadata
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_0001a24f_00000000-6_kernelInitNablaW.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2029: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2029: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z37__device_stub__Z16kernelInitNablaWPfiPfi .type _Z37__device_stub__Z16kernelInitNablaWPfiPfi, @function _Z37__device_stub__Z16kernelInitNablaWPfiPfi: .LFB2051: .cfi_startproc endbr64 subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 8(%rsp) movl %esi, 4(%rsp) movq %fs:40, %rax movq %rax, 104(%rsp) xorl %eax, %eax leaq 8(%rsp), %rax movq %rax, 80(%rsp) leaq 4(%rsp), %rax movq %rax, 88(%rsp) movl $1, 32(%rsp) movl $1, 36(%rsp) movl $1, 40(%rsp) movl $1, 44(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) leaq 24(%rsp), %rcx leaq 16(%rsp), %rdx leaq 44(%rsp), %rsi leaq 32(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 104(%rsp), %rax subq %fs:40, %rax jne .L8 addq $120, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 24(%rsp) .cfi_def_cfa_offset 136 pushq 24(%rsp) .cfi_def_cfa_offset 144 leaq 96(%rsp), %r9 movq 60(%rsp), %rcx movl 68(%rsp), %r8d movq 48(%rsp), %rsi movl 56(%rsp), %edx leaq _Z16kernelInitNablaWPfi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 128 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2051: .size _Z37__device_stub__Z16kernelInitNablaWPfiPfi, .-_Z37__device_stub__Z16kernelInitNablaWPfiPfi .globl _Z16kernelInitNablaWPfi .type _Z16kernelInitNablaWPfi, @function _Z16kernelInitNablaWPfi: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z37__device_stub__Z16kernelInitNablaWPfiPfi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z16kernelInitNablaWPfi, .-_Z16kernelInitNablaWPfi .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z16kernelInitNablaWPfi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2054: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z16kernelInitNablaWPfi(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2054: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "kernelInitNablaW.hip" .globl _Z31__device_stub__kernelInitNablaWPfi # -- Begin function _Z31__device_stub__kernelInitNablaWPfi .p2align 4, 0x90 .type _Z31__device_stub__kernelInitNablaWPfi,@function _Z31__device_stub__kernelInitNablaWPfi: # @_Z31__device_stub__kernelInitNablaWPfi .cfi_startproc # %bb.0: subq $88, %rsp .cfi_def_cfa_offset 96 movq %rdi, 56(%rsp) movl %esi, 4(%rsp) leaq 56(%rsp), %rax movq %rax, 64(%rsp) leaq 4(%rsp), %rax movq %rax, 72(%rsp) leaq 40(%rsp), %rdi leaq 24(%rsp), %rsi leaq 16(%rsp), %rdx leaq 8(%rsp), %rcx callq __hipPopCallConfiguration movq 40(%rsp), %rsi movl 48(%rsp), %edx movq 24(%rsp), %rcx movl 32(%rsp), %r8d leaq 64(%rsp), %r9 movl $_Z16kernelInitNablaWPfi, %edi pushq 8(%rsp) .cfi_adjust_cfa_offset 8 pushq 24(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $104, %rsp .cfi_adjust_cfa_offset -104 retq .Lfunc_end0: .size _Z31__device_stub__kernelInitNablaWPfi, .Lfunc_end0-_Z31__device_stub__kernelInitNablaWPfi .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z16kernelInitNablaWPfi, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type _Z16kernelInitNablaWPfi,@object # @_Z16kernelInitNablaWPfi .section .rodata,"a",@progbits .globl _Z16kernelInitNablaWPfi .p2align 3, 0x0 _Z16kernelInitNablaWPfi: .quad _Z31__device_stub__kernelInitNablaWPfi .size _Z16kernelInitNablaWPfi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z16kernelInitNablaWPfi" .size .L__unnamed_1, 24 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z31__device_stub__kernelInitNablaWPfi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z16kernelInitNablaWPfi .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
#include <stdio.h> __global__ void binPacking (float *bins, float *items, float cap, int length) { int x = 0; for(int i = 0; i < length; i++) { x = 0; if(items[i] > cap) { printf ("Element %f je veci od kapaciteta spremnika koji je %f. PREKID!\n", items[i], cap); break; } if(bins[0 * length + x] >= items[i]) { bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } else { while(bins[0 * length + x] < items[i]) { x+=1; } bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } } }
code for sm_80 Function : _Z10binPackingPfS_fi .headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM80 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM80)" /*0000*/ IMAD.MOV.U32 R1, RZ, RZ, c[0x0][0x28] ; /* 0x00000a00ff017624 */ /* 0x000fc800078e00ff */ /*0010*/ IMAD.MOV.U32 R9, RZ, RZ, c[0x0][0x174] ; /* 0x00005d00ff097624 */ /* 0x000fe200078e00ff */ /*0020*/ IADD3 R1, R1, -0x10, RZ ; /* 0xfffffff001017810 */ /* 0x000fc80007ffe0ff */ /*0030*/ ISETP.GE.AND P0, PT, R9, 0x1, PT ; /* 0x000000010900780c */ /* 0x000fda0003f06270 */ /*0040*/ @!P0 EXIT ; /* 0x000000000000894d */ /* 0x000fea0003800000 */ /*0050*/ SHF.R.S32.HI R2, RZ, 0x1f, R9 ; /* 0x0000001fff027819 */ /* 0x000fe20000011409 */ /*0060*/ IMAD.MOV.U32 R4, RZ, RZ, 0x4 ; /* 0x00000004ff047424 */ /* 0x000fe200078e00ff */ /*0070*/ IADD3 R6, P0, R1, c[0x0][0x20], RZ ; /* 0x0000080001067a10 */ /* 0x000fe20007f1e0ff */ /*0080*/ IMAD.MOV.U32 R0, RZ, RZ, RZ ; /* 0x000000ffff007224 */ /* 0x000fe200078e00ff */ /*0090*/ SHF.L.U64.HI R19, R9.reuse, 0x2, R2 ; /* 0x0000000209137819 */ /* 0x040fe20000010202 */ /*00a0*/ IMAD.WIDE R2, R9, R4, c[0x0][0x160] ; /* 0x0000580009027625 */ /* 0x000fe200078e0204 */ /*00b0*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fc60000000a00 */ /*00c0*/ IMAD.X R7, RZ, RZ, c[0x0][0x24], P0 ; /* 0x00000900ff077624 */ /* 0x000fe400000e06ff */ /*00d0*/ IMAD.MOV.U32 R17, RZ, RZ, 0x4 ; /* 0x00000004ff117424 */ /* 0x001fc800078e00ff */ /*00e0*/ IMAD.WIDE R4, R0, R17, c[0x0][0x168] ; /* 0x00005a0000047625 */ /* 0x000fca00078e0211 */ /*00f0*/ LDG.E R8, [R4.64] ; /* 0x0000000404087981 */ /* 0x000ea4000c1e1900 */ /*0100*/ FSETP.GT.AND P0, PT, R8, c[0x0][0x170], PT ; /* 0x00005c0008007a0b */ /* 0x004fda0003f04000 */ /*0110*/ @P0 BRA 0x300 ; /* 0x000001e000000947 */ /* 0x000fea0003800000 */ /*0120*/ IMAD.MOV.U32 R10, RZ, RZ, c[0x0][0x160] ; /* 0x00005800ff0a7624 */ /* 0x000fe400078e00ff */ /*0130*/ IMAD.MOV.U32 R11, RZ, RZ, c[0x0][0x164] ; /* 0x00005900ff0b7624 */ /* 0x000fca00078e00ff */ /*0140*/ LDG.E R13, [R10.64] ; /* 0x000000040a0d7981 */ /* 0x000ea4000c1e1900 */ /*0150*/ FSETP.GE.AND P0, PT, R13, R8, PT ; /* 0x000000080d00720b */ /* 0x004fda0003f06000 */ /*0160*/ @!P0 BRA 0x1e0 ; /* 0x0000007000008947 */ /* 0x000fea0003800000 */ /*0170*/ FADD R13, -R8, R13 ; /* 0x0000000d080d7221 */ /* 0x000fca0000000100 */ /*0180*/ STG.E [R10.64], R13 ; /* 0x0000000d0a007986 */ /* 0x0001e8000c101904 */ /*0190*/ LDG.E R5, [R4.64] ; /* 0x0000000404057981 */ /* 0x000ea8000c1e1900 */ /*01a0*/ LDG.E R8, [R2.64] ; /* 0x0000000402087981 */ /* 0x000ea4000c1e1900 */ /*01b0*/ FADD R15, R8, R5 ; /* 0x00000005080f7221 */ /* 0x004fca0000000000 */ /*01c0*/ STG.E [R2.64], R15 ; /* 0x0000000f02007986 */ /* 0x0001e2000c101904 */ /*01d0*/ BRA 0x2c0 ; /* 0x000000e000007947 */ /* 0x000fea0003800000 */ /*01e0*/ IMAD.MOV.U32 R10, RZ, RZ, RZ ; /* 0x000000ffff0a7224 */ /* 0x000fc800078e00ff */ /*01f0*/ IMAD.WIDE R12, R10, R17, c[0x0][0x160] ; /* 0x000058000a0c7625 */ /* 0x000fca00078e0211 */ /*0200*/ LDG.E R15, [R12.64] ; /* 0x000000040c0f7981 */ /* 0x000ea2000c1e1900 */ /*0210*/ IADD3 R10, R10, 0x1, RZ ; /* 0x000000010a0a7810 */ /* 0x000fe40007ffe0ff */ /*0220*/ FSETP.GEU.AND P0, PT, R15, R8, PT ; /* 0x000000080f00720b */ /* 0x004fda0003f0e000 */ /*0230*/ @!P0 BRA 0x1f0 ; /* 0xffffffb000008947 */ /* 0x000fea000383ffff */ /*0240*/ LEA R10, P0, R9, R12, 0x2 ; /* 0x0000000c090a7211 */ /* 0x000fe200078010ff */ /*0250*/ FADD R15, -R8, R15 ; /* 0x0000000f080f7221 */ /* 0x000fc80000000100 */ /*0260*/ IMAD.X R11, R13, 0x1, R19, P0 ; /* 0x000000010d0b7824 */ /* 0x000fe200000e0613 */ /*0270*/ STG.E [R12.64], R15 ; /* 0x0000000f0c007986 */ /* 0x0001e8000c101904 */ /*0280*/ LDG.E R5, [R4.64] ; /* 0x0000000404057981 */ /* 0x000ea8000c1e1900 */ /*0290*/ LDG.E R8, [R10.64] ; /* 0x000000040a087981 */ /* 0x000ea4000c1e1900 */ /*02a0*/ FADD R17, R8, R5 ; /* 0x0000000508117221 */ /* 0x004fca0000000000 */ /*02b0*/ STG.E [R10.64], R17 ; /* 0x000000110a007986 */ /* 0x0001e4000c101904 */ /*02c0*/ IADD3 R0, R0, 0x1, RZ ; /* 0x0000000100007810 */ /* 0x000fc80007ffe0ff */ /*02d0*/ ISETP.GE.AND P0, PT, R0, c[0x0][0x174], PT ; /* 0x00005d0000007a0c */ /* 0x000fda0003f06270 */ /*02e0*/ @!P0 BRA 0xd0 ; /* 0xfffffde000008947 */ /* 0x000fea000383ffff */ /*02f0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0300*/ F2F.F64.F32 R8, R8 ; /* 0x0000000800087310 */ /* 0x000fe20000201800 */ /*0310*/ MOV R0, 0x0 ; /* 0x0000000000007802 */ /* 0x000fe20000000f00 */ /*0320*/ IMAD.MOV.U32 R4, RZ, RZ, c[0x4][0x8] ; /* 0x01000200ff047624 */ /* 0x000fe400078e00ff */ /*0330*/ IMAD.MOV.U32 R5, RZ, RZ, c[0x4][0xc] ; /* 0x01000300ff057624 */ /* 0x000fe200078e00ff */ /*0340*/ LDC.64 R2, c[0x4][R0] ; /* 0x0100000000027b82 */ /* 0x0000660000000a00 */ /*0350*/ F2F.F64.F32 R10, c[0x0][0x170] ; /* 0x00005c00000a7b10 */ /* 0x000ea40000201800 */ /*0360*/ STL.128 [R1], R8 ; /* 0x0000000801007387 */ /* 0x0041e40000100c00 */ /*0370*/ LEPC R8 ; /* 0x000000000008734e */ /* 0x001fe20000000000 */ /*0380*/ MOV R11, 0x3f0 ; /* 0x000003f0000b7802 */ /* 0x000fc40000000f00 */ /*0390*/ MOV R20, 0x370 ; /* 0x0000037000147802 */ /* 0x000fe40000000f00 */ /*03a0*/ MOV R21, 0x0 ; /* 0x0000000000157802 */ /* 0x000fe40000000f00 */ /*03b0*/ MOV R0, 0x0 ; /* 0x0000000000007802 */ /* 0x000fe40000000f00 */ /*03c0*/ IADD3 R20, P0, P1, -R20, R11, R8 ; /* 0x0000000b14147210 */ /* 0x000fc8000791e108 */ /*03d0*/ IADD3.X R21, ~R0, R21, R9, P0, P1 ; /* 0x0000001500157210 */ /* 0x000fc800007e2509 */ /*03e0*/ CALL.ABS.NOINC R2 ; /* 0x0000000002007343 */ /* 0x002fea0003c00000 */ /*03f0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0400*/ BRA 0x400; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0410*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0420*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0430*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0440*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0450*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0460*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0470*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0480*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0490*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*04a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*04b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*04c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*04d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*04e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*04f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include <stdio.h> __global__ void binPacking (float *bins, float *items, float cap, int length) { int x = 0; for(int i = 0; i < length; i++) { x = 0; if(items[i] > cap) { printf ("Element %f je veci od kapaciteta spremnika koji je %f. PREKID!\n", items[i], cap); break; } if(bins[0 * length + x] >= items[i]) { bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } else { while(bins[0 * length + x] < items[i]) { x+=1; } bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } } }
.file "tmpxft_00060c51_00000000-6_firstFitAlgorithm.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2059: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2059: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z34__device_stub__Z10binPackingPfS_fiPfS_fi .type _Z34__device_stub__Z10binPackingPfS_fiPfS_fi, @function _Z34__device_stub__Z10binPackingPfS_fiPfS_fi: .LFB2081: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movss %xmm0, 12(%rsp) movl %edx, 8(%rsp) movq %fs:40, %rax movq %rax, 136(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 12(%rsp), %rax movq %rax, 112(%rsp) leaq 8(%rsp), %rax movq %rax, 120(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 136(%rsp), %rax subq %fs:40, %rax jne .L8 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 168 pushq 40(%rsp) .cfi_def_cfa_offset 176 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z10binPackingPfS_fi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2081: .size _Z34__device_stub__Z10binPackingPfS_fiPfS_fi, .-_Z34__device_stub__Z10binPackingPfS_fiPfS_fi .globl _Z10binPackingPfS_fi .type _Z10binPackingPfS_fi, @function _Z10binPackingPfS_fi: .LFB2082: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z34__device_stub__Z10binPackingPfS_fiPfS_fi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2082: .size _Z10binPackingPfS_fi, .-_Z10binPackingPfS_fi .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z10binPackingPfS_fi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2084: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z10binPackingPfS_fi(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2084: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include <stdio.h> __global__ void binPacking (float *bins, float *items, float cap, int length) { int x = 0; for(int i = 0; i < length; i++) { x = 0; if(items[i] > cap) { printf ("Element %f je veci od kapaciteta spremnika koji je %f. PREKID!\n", items[i], cap); break; } if(bins[0 * length + x] >= items[i]) { bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } else { while(bins[0 * length + x] < items[i]) { x+=1; } bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } } }
#include <hip/hip_runtime.h> #include <stdio.h> __global__ void binPacking (float *bins, float *items, float cap, int length) { int x = 0; for(int i = 0; i < length; i++) { x = 0; if(items[i] > cap) { printf ("Element %f je veci od kapaciteta spremnika koji je %f. PREKID!\n", items[i], cap); break; } if(bins[0 * length + x] >= items[i]) { bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } else { while(bins[0 * length + x] < items[i]) { x+=1; } bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } } }
You are a helpful coding assistant specialized in converting HIP source code to AMD host assembly.
#include <hip/hip_runtime.h> #include <stdio.h> __global__ void binPacking (float *bins, float *items, float cap, int length) { int x = 0; for(int i = 0; i < length; i++) { x = 0; if(items[i] > cap) { printf ("Element %f je veci od kapaciteta spremnika koji je %f. PREKID!\n", items[i], cap); break; } if(bins[0 * length + x] >= items[i]) { bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } else { while(bins[0 * length + x] < items[i]) { x+=1; } bins[0 * length + x] -= items[i]; bins[1 * length + x] += items[i]; } } }
.text .file "firstFitAlgorithm.hip" .globl _Z25__device_stub__binPackingPfS_fi # -- Begin function _Z25__device_stub__binPackingPfS_fi .p2align 4, 0x90 .type _Z25__device_stub__binPackingPfS_fi,@function _Z25__device_stub__binPackingPfS_fi: # @_Z25__device_stub__binPackingPfS_fi .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movss %xmm0, 12(%rsp) movl %edx, 8(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 12(%rsp), %rax movq %rax, 96(%rsp) leaq 8(%rsp), %rax movq %rax, 104(%rsp) leaq 48(%rsp), %rdi leaq 32(%rsp), %rsi leaq 24(%rsp), %rdx leaq 16(%rsp), %rcx callq __hipPopCallConfiguration movq 48(%rsp), %rsi movl 56(%rsp), %edx movq 32(%rsp), %rcx movl 40(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z10binPackingPfS_fi, %edi pushq 16(%rsp) .cfi_adjust_cfa_offset 8 pushq 32(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $136, %rsp .cfi_adjust_cfa_offset -136 retq .Lfunc_end0: .size _Z25__device_stub__binPackingPfS_fi, .Lfunc_end0-_Z25__device_stub__binPackingPfS_fi .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z10binPackingPfS_fi, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type _Z10binPackingPfS_fi,@object # @_Z10binPackingPfS_fi .section .rodata,"a",@progbits .globl _Z10binPackingPfS_fi .p2align 3, 0x0 _Z10binPackingPfS_fi: .quad _Z25__device_stub__binPackingPfS_fi .size _Z10binPackingPfS_fi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z10binPackingPfS_fi" .size .L__unnamed_1, 21 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z25__device_stub__binPackingPfS_fi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z10binPackingPfS_fi .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting from CUDA host assembly to AMD host assembly.
.file "tmpxft_00060c51_00000000-6_firstFitAlgorithm.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2059: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaUnregisterFatBinary@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2059: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z34__device_stub__Z10binPackingPfS_fiPfS_fi .type _Z34__device_stub__Z10binPackingPfS_fiPfS_fi, @function _Z34__device_stub__Z10binPackingPfS_fiPfS_fi: .LFB2081: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movss %xmm0, 12(%rsp) movl %edx, 8(%rsp) movq %fs:40, %rax movq %rax, 136(%rsp) xorl %eax, %eax leaq 24(%rsp), %rax movq %rax, 96(%rsp) leaq 16(%rsp), %rax movq %rax, 104(%rsp) leaq 12(%rsp), %rax movq %rax, 112(%rsp) leaq 8(%rsp), %rax movq %rax, 120(%rsp) movl $1, 48(%rsp) movl $1, 52(%rsp) movl $1, 56(%rsp) movl $1, 60(%rsp) movl $1, 64(%rsp) movl $1, 68(%rsp) leaq 40(%rsp), %rcx leaq 32(%rsp), %rdx leaq 60(%rsp), %rsi leaq 48(%rsp), %rdi call __cudaPopCallConfiguration@PLT testl %eax, %eax je .L7 .L3: movq 136(%rsp), %rax subq %fs:40, %rax jne .L8 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L7: .cfi_restore_state pushq 40(%rsp) .cfi_def_cfa_offset 168 pushq 40(%rsp) .cfi_def_cfa_offset 176 leaq 112(%rsp), %r9 movq 76(%rsp), %rcx movl 84(%rsp), %r8d movq 64(%rsp), %rsi movl 72(%rsp), %edx leaq _Z10binPackingPfS_fi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2081: .size _Z34__device_stub__Z10binPackingPfS_fiPfS_fi, .-_Z34__device_stub__Z10binPackingPfS_fiPfS_fi .globl _Z10binPackingPfS_fi .type _Z10binPackingPfS_fi, @function _Z10binPackingPfS_fi: .LFB2082: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z34__device_stub__Z10binPackingPfS_fiPfS_fi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2082: .size _Z10binPackingPfS_fi, .-_Z10binPackingPfS_fi .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z10binPackingPfS_fi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2084: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 leaq _ZL15__fatDeviceText(%rip), %rdi call __cudaRegisterFatBinary@PLT movq %rax, %rdi movq %rax, _ZL20__cudaFatCubinHandle(%rip) pushq $0 .cfi_def_cfa_offset 24 pushq $0 .cfi_def_cfa_offset 32 pushq $0 .cfi_def_cfa_offset 40 pushq $0 .cfi_def_cfa_offset 48 movl $0, %r9d movl $-1, %r8d leaq .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z10binPackingPfS_fi(%rip), %rsi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 movq _ZL20__cudaFatCubinHandle(%rip), %rdi call __cudaRegisterFatBinaryEnd@PLT leaq _ZL26__cudaUnregisterBinaryUtilv(%rip), %rdi call atexit@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2084: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .nvFatBinSegment,"aw" .align 8 .type _ZL15__fatDeviceText, @object .size _ZL15__fatDeviceText, 24 _ZL15__fatDeviceText: .long 1180844977 .long 1 .quad fatbinData .quad 0 .local _ZL20__cudaFatCubinHandle .comm _ZL20__cudaFatCubinHandle,8,8 .ident "GCC: (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0" .section .note.GNU-stack,"",@progbits .section .note.gnu.property,"a" .align 8 .long 1f - 0f .long 4f - 1f .long 5 0: .string "GNU" 1: .align 8 .long 0xc0000002 .long 3f - 2f 2: .long 0x3 3: .align 8 4:
.text .file "firstFitAlgorithm.hip" .globl _Z25__device_stub__binPackingPfS_fi # -- Begin function _Z25__device_stub__binPackingPfS_fi .p2align 4, 0x90 .type _Z25__device_stub__binPackingPfS_fi,@function _Z25__device_stub__binPackingPfS_fi: # @_Z25__device_stub__binPackingPfS_fi .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movss %xmm0, 12(%rsp) movl %edx, 8(%rsp) leaq 72(%rsp), %rax movq %rax, 80(%rsp) leaq 64(%rsp), %rax movq %rax, 88(%rsp) leaq 12(%rsp), %rax movq %rax, 96(%rsp) leaq 8(%rsp), %rax movq %rax, 104(%rsp) leaq 48(%rsp), %rdi leaq 32(%rsp), %rsi leaq 24(%rsp), %rdx leaq 16(%rsp), %rcx callq __hipPopCallConfiguration movq 48(%rsp), %rsi movl 56(%rsp), %edx movq 32(%rsp), %rcx movl 40(%rsp), %r8d leaq 80(%rsp), %r9 movl $_Z10binPackingPfS_fi, %edi pushq 16(%rsp) .cfi_adjust_cfa_offset 8 pushq 32(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $136, %rsp .cfi_adjust_cfa_offset -136 retq .Lfunc_end0: .size _Z25__device_stub__binPackingPfS_fi, .Lfunc_end0-_Z25__device_stub__binPackingPfS_fi .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_ctor .type __hip_module_ctor,@function __hip_module_ctor: # @__hip_module_ctor .cfi_startproc # %bb.0: subq $40, %rsp .cfi_def_cfa_offset 48 cmpq $0, __hip_gpubin_handle(%rip) jne .LBB1_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB1_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z10binPackingPfS_fi, %esi movl $.L__unnamed_1, %edx movl $.L__unnamed_1, %ecx movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $__hip_module_dtor, %edi addq $40, %rsp .cfi_def_cfa_offset 8 jmp atexit # TAILCALL .Lfunc_end1: .size __hip_module_ctor, .Lfunc_end1-__hip_module_ctor .cfi_endproc # -- End function .p2align 4, 0x90 # -- Begin function __hip_module_dtor .type __hip_module_dtor,@function __hip_module_dtor: # @__hip_module_dtor .cfi_startproc # %bb.0: movq __hip_gpubin_handle(%rip), %rdi testq %rdi, %rdi je .LBB2_2 # %bb.1: pushq %rax .cfi_def_cfa_offset 16 callq __hipUnregisterFatBinary movq $0, __hip_gpubin_handle(%rip) addq $8, %rsp .cfi_def_cfa_offset 8 .LBB2_2: retq .Lfunc_end2: .size __hip_module_dtor, .Lfunc_end2-__hip_module_dtor .cfi_endproc # -- End function .type _Z10binPackingPfS_fi,@object # @_Z10binPackingPfS_fi .section .rodata,"a",@progbits .globl _Z10binPackingPfS_fi .p2align 3, 0x0 _Z10binPackingPfS_fi: .quad _Z25__device_stub__binPackingPfS_fi .size _Z10binPackingPfS_fi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z10binPackingPfS_fi" .size .L__unnamed_1, 21 .type __hip_fatbin_wrapper,@object # @__hip_fatbin_wrapper .section .hipFatBinSegment,"a",@progbits .p2align 3, 0x0 __hip_fatbin_wrapper: .long 1212764230 # 0x48495046 .long 1 # 0x1 .quad __hip_fatbin .quad 0 .size __hip_fatbin_wrapper, 24 .type __hip_gpubin_handle,@object # @__hip_gpubin_handle .local __hip_gpubin_handle .comm __hip_gpubin_handle,8,8 .section .init_array,"aw",@init_array .p2align 3, 0x0 .quad __hip_module_ctor .type __hip_cuid_,@object # @__hip_cuid_ .bss .globl __hip_cuid_ __hip_cuid_: .byte 0 # 0x0 .size __hip_cuid_, 1 .section ".linker-options","e",@llvm_linker_options .ident "AMD clang version 18.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-6.3.2 25012 e5bf7e55c91490b07c49d8960fa7983d864936c4)" .section ".note.GNU-stack","",@progbits .addrsig .addrsig_sym _Z25__device_stub__binPackingPfS_fi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z10binPackingPfS_fi .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_
You are a helpful coding assistant specialized in converting CUDA source code to CUDA device assembly.
/********************************************************************* * Copyright © 2011-2012, * Marwan Abdellah: <abdellah.marwan@gmail.com> * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation. * This library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, * MA 02110-1301, USA. ********************************************************************/ /*! * CUDA : This kernel adds a constant value to the input vector into the * output vector with length N. * * @param devArrayInput * Input vector. * * @param constVal * Constant value to be added to the input device vector. * * @param devArrayOutput * Sum vector. * * @param N * Vector length. * * @author * Marwan Abdellah <abdellah.marwan@gmail.com> * * @date * Created: August, 2012. * @date * Last Update: September, 2012. * * @note * Minimum CUDA version 3.2. * @note * Minimum Device Compute Capability 1.0. */ template <typename T> __global__ void Constant_Add_1D_Array_Kernel(T* devArrayInput, T constVal, T* devArrayOutput, int N) { int xThreadIdx = threadIdx.x; int blockWidth = blockDim.x; int index = blockIdx.x * blockWidth + xThreadIdx; #ifdef VEC_CHECK if (index < N) devArrayOutput[index] = (T) ((T) devArrayInput[index] + (T) constVal); #else devArrayOutput[index] = (T) ((T) devArrayInput[index] + (T) constVal); #endif }
code for sm_80