system
stringclasses
7 values
input
stringlengths
10
94.5k
output
stringlengths
14
55.5k
You are a helpful coding assistant specialized in converting CUDA source code to HIP/AMD source code.
#include "includes.h" __global__ void fourier_transform(float *in, float *out, int height, int width, int blockConfig) { // block elements and function variables int my_x, k, t; my_x = blockIdx.x * blockDim.x + threadIdx.x; // iterate through each element, going from frequency to time domain for (k = 0; k < height; k++) { // difference, which will be used to subtract off float realSum = 0.0; // iterate through the input element for (t = 0; t < width; t++) { // calculate the angle and update the sum float angle = 2 * M_PI * (my_x * height + t) * (my_x * width + k) / height; realSum += in[my_x * height + t] * cos(angle); } // each output element will be the current sum for that index out[my_x * height + k] = realSum; } }
#include <hip/hip_runtime.h> #include "includes.h" __global__ void fourier_transform(float *in, float *out, int height, int width, int blockConfig) { // block elements and function variables int my_x, k, t; my_x = blockIdx.x * blockDim.x + threadIdx.x; // iterate through each element, going from frequency to time domain for (k = 0; k < height; k++) { // difference, which will be used to subtract off float realSum = 0.0; // iterate through the input element for (t = 0; t < width; t++) { // calculate the angle and update the sum float angle = 2 * M_PI * (my_x * height + t) * (my_x * width + k) / height; realSum += in[my_x * height + t] * cos(angle); } // each output element will be the current sum for that index out[my_x * height + k] = realSum; } }
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 fourier_transform(float *in, float *out, int height, int width, int blockConfig) { // block elements and function variables int my_x, k, t; my_x = blockIdx.x * blockDim.x + threadIdx.x; // iterate through each element, going from frequency to time domain for (k = 0; k < height; k++) { // difference, which will be used to subtract off float realSum = 0.0; // iterate through the input element for (t = 0; t < width; t++) { // calculate the angle and update the sum float angle = 2 * M_PI * (my_x * height + t) * (my_x * width + k) / height; realSum += in[my_x * height + t] * cos(angle); } // each output element will be the current sum for that index out[my_x * height + k] = realSum; } }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z17fourier_transformPfS_iii .globl _Z17fourier_transformPfS_iii .p2align 8 .type _Z17fourier_transformPfS_iii,@function _Z17fourier_transformPfS_iii: s_load_b32 s6, s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_cmp_lt_i32 s6, 1 s_cbranch_scc1 .LBB0_11 v_cvt_f64_i32_e32 v[1:2], s6 s_clause 0x1 s_load_b32 s2, s[0:1], 0x2c s_load_b32 s7, s[0:1], 0x14 s_mov_b32 s8, 0 s_mov_b32 s5, 0x401921fb s_mov_b32 s4, 0x54442d18 s_mov_b32 s10, 0x7fffff s_mov_b32 s11, 0xb94c1982 s_mov_b32 s12, 0x37d75334 s_waitcnt lgkmcnt(0) s_and_b32 s2, s2, 0xffff s_cmp_gt_i32 s7, 0 s_cselect_b32 s9, -1, 0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_mad_u64_u32 v[5:6], null, s15, s2, v[0:1] s_load_b128 s[0:3], s[0:1], 0x0 v_mul_lo_u32 v3, v5, s6 v_mul_lo_u32 v0, v5, s7 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_ashrrev_i32_e32 v4, 31, v3 v_lshlrev_b64 v[6:7], 2, v[3:4] s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v4, vcc_lo, s0, v6 v_add_co_ci_u32_e32 v5, vcc_lo, s1, v7, vcc_lo s_branch .LBB0_4 .LBB0_2: v_mov_b32_e32 v10, 0 .LBB0_3: v_add_nc_u32_e32 v6, s8, v3 s_add_i32 s8, s8, 1 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) s_cmp_lg_u32 s8, s6 v_ashrrev_i32_e32 v7, 31, v6 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b64 v[6:7], 2, v[6:7] v_add_co_u32 v6, vcc_lo, s2, v6 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v7, vcc_lo, s3, v7, vcc_lo global_store_b32 v[6:7], v10, off s_cbranch_scc0 .LBB0_11 .LBB0_4: s_and_not1_b32 vcc_lo, exec_lo, s9 s_cbranch_vccnz .LBB0_2 v_dual_mov_b32 v11, v3 :: v_dual_add_nc_u32 v6, s8, v0 v_dual_mov_b32 v10, 0 :: v_dual_mov_b32 v9, v5 v_mov_b32_e32 v8, v4 s_delay_alu instid0(VALU_DEP_3) v_cvt_f64_i32_e32 v[6:7], v6 s_mov_b32 s13, s7 s_branch .LBB0_7 .LBB0_6: s_or_b32 exec_lo, exec_lo, s0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_2) v_dual_mul_f32 v16, v14, v14 :: v_dual_and_b32 v19, 1, v15 s_add_i32 s13, s13, -1 v_lshlrev_b32_e32 v15, 30, v15 s_cmp_lg_u32 s13, 0 v_fmaak_f32 v17, s11, v16, 0x3c0881c4 v_cmp_eq_u32_e32 vcc_lo, 0, v19 v_add_nc_u32_e32 v11, 1, v11 v_and_b32_e32 v15, 0x80000000, v15 s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_1) v_fmaak_f32 v17, v16, v17, 0xbe2aaa9d v_dual_fmaak_f32 v18, s12, v16, 0xbab64f3b :: v_dual_mul_f32 v17, v16, v17 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_fmaak_f32 v18, v16, v18, 0x3d2aabf7 v_fmac_f32_e32 v14, v14, v17 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_fmaak_f32 v18, v16, v18, 0xbf000004 v_fma_f32 v16, v16, v18, 1.0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_cndmask_b32_e64 v14, -v14, v16, vcc_lo v_cmp_class_f32_e64 vcc_lo, v13, 0x1f8 v_xor_b32_e32 v14, v15, v14 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_3) v_cndmask_b32_e32 v13, 0x7fc00000, v14, vcc_lo v_add_co_u32 v8, vcc_lo, v8, 4 v_add_co_ci_u32_e32 v9, vcc_lo, 0, v9, vcc_lo s_waitcnt vmcnt(0) v_fmac_f32_e32 v10, v12, v13 s_cbranch_scc0 .LBB0_3 .LBB0_7: v_cvt_f64_i32_e32 v[12:13], v11 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f64 v[12:13], v[12:13], s[4:5] v_mul_f64 v[13:14], v[12:13], v[6:7] global_load_b32 v12, v[8:9], off v_div_scale_f64 v[15:16], null, v[1:2], v[1:2], v[13:14] s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_rcp_f64_e32 v[17:18], v[15:16] s_waitcnt_depctr 0xfff v_fma_f64 v[19:20], -v[15:16], v[17:18], 1.0 v_fma_f64 v[17:18], v[17:18], v[19:20], v[17:18] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fma_f64 v[19:20], -v[15:16], v[17:18], 1.0 v_fma_f64 v[17:18], v[17:18], v[19:20], v[17:18] v_div_scale_f64 v[19:20], vcc_lo, v[13:14], v[1:2], v[13:14] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f64 v[21:22], v[19:20], v[17:18] v_fma_f64 v[15:16], -v[15:16], v[21:22], v[19:20] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_div_fmas_f64 v[15:16], v[15:16], v[17:18], v[21:22] v_div_fixup_f64 v[13:14], v[15:16], v[1:2], v[13:14] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cvt_f32_f64_e32 v13, v[13:14] v_cmp_ngt_f32_e64 s0, 0x48000000, |v13| s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_saveexec_b32 s1, s0 s_xor_b32 s14, exec_lo, s1 s_cbranch_execz .LBB0_9 v_dual_mov_b32 v16, 0 :: v_dual_and_b32 v21, 0x7fffffff, v13 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_and_or_b32 v24, v21, s10, 0x800000 v_lshrrev_b32_e32 v21, 23, v21 v_mad_u64_u32 v[14:15], null, v24, 0xfe5163ab, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_nc_u32_e32 v22, 0xffffff88, v21 v_cmp_lt_u32_e32 vcc_lo, 63, v22 s_delay_alu instid0(VALU_DEP_3) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_mad_u64_u32 v[17:18], null, v24, 0x3c439041, v[15:16] v_cndmask_b32_e64 v23, 0, 0xffffffc0, vcc_lo v_mov_b32_e32 v15, v18 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v23, v23, v22 v_mad_u64_u32 v[18:19], null, v24, 0xdb629599, v[15:16] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_lt_u32_e64 s0, 31, v23 v_cndmask_b32_e64 v25, 0, 0xffffffe0, s0 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2) v_dual_mov_b32 v15, v19 :: v_dual_cndmask_b32 v14, v18, v14 v_add_nc_u32_e32 v25, v25, v23 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mad_u64_u32 v[19:20], null, v24, 0xf534ddc0, v[15:16] v_cmp_lt_u32_e64 s1, 31, v25 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_mov_b32_e32 v15, v20 v_cndmask_b32_e32 v17, v19, v17, vcc_lo s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mad_u64_u32 v[20:21], null, v24, 0xfc2757d1, v[15:16] v_cndmask_b32_e64 v14, v17, v14, s0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_mov_b32_e32 v15, v21 v_mad_u64_u32 v[21:22], null, v24, 0x4e441529, v[15:16] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mov_b32_e32 v15, v22 v_mad_u64_u32 v[22:23], null, v24, 0xa2f9836e, v[15:16] v_cndmask_b32_e64 v15, 0, 0xffffffe0, s1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_3) v_dual_cndmask_b32 v16, v21, v19 :: v_dual_add_nc_u32 v15, v15, v25 v_dual_cndmask_b32 v22, v22, v20 :: v_dual_cndmask_b32 v21, v23, v21 v_cndmask_b32_e32 v20, v20, v18, vcc_lo s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_cmp_eq_u32_e32 vcc_lo, 0, v15 v_cndmask_b32_e64 v19, v22, v16, s0 s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4) v_cndmask_b32_e64 v21, v21, v22, s0 v_cndmask_b32_e64 v16, v16, v20, s0 v_sub_nc_u32_e32 v22, 32, v15 v_cndmask_b32_e64 v20, v20, v17, s0 s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4) v_cndmask_b32_e64 v21, v21, v19, s1 v_cndmask_b32_e64 v19, v19, v16, s1 s_delay_alu instid0(VALU_DEP_3) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_cndmask_b32_e64 v16, v16, v20, s1 v_cndmask_b32_e64 v14, v20, v14, s1 v_alignbit_b32 v23, v21, v19, v22 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2) v_alignbit_b32 v18, v19, v16, v22 v_cndmask_b32_e32 v15, v23, v21, vcc_lo s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_3) v_alignbit_b32 v21, v16, v14, v22 v_cndmask_b32_e32 v17, v18, v19, vcc_lo s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_bfe_u32 v18, v15, 29, 1 v_cndmask_b32_e32 v16, v21, v16, vcc_lo s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_alignbit_b32 v19, v15, v17, 30 v_sub_nc_u32_e32 v20, 0, v18 s_delay_alu instid0(VALU_DEP_3) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_alignbit_b32 v17, v17, v16, 30 v_alignbit_b32 v14, v16, v14, 30 v_xor_b32_e32 v19, v19, v20 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_xor_b32_e32 v16, v17, v20 v_xor_b32_e32 v14, v14, v20 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_clz_i32_u32_e32 v21, v19 v_min_u32_e32 v21, 32, v21 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_sub_nc_u32_e32 v17, 31, v21 v_lshlrev_b32_e32 v23, 23, v21 v_alignbit_b32 v19, v19, v16, v17 v_alignbit_b32 v14, v16, v14, v17 v_lshrrev_b32_e32 v17, 29, v15 v_lshrrev_b32_e32 v15, 30, v15 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_alignbit_b32 v16, v19, v14, 9 v_lshlrev_b32_e32 v17, 31, v17 v_lshrrev_b32_e32 v19, 9, v19 s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4) v_add_nc_u32_e32 v15, v18, v15 v_clz_i32_u32_e32 v20, v16 s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_2) v_or_b32_e32 v22, 0.5, v17 v_min_u32_e32 v20, 32, v20 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_sub_nc_u32_e32 v22, v22, v23 v_sub_nc_u32_e32 v24, 31, v20 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_3) v_alignbit_b32 v14, v16, v14, v24 v_or_b32_e32 v16, v19, v22 v_add_lshl_u32 v19, v20, v21, 23 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_lshrrev_b32_e32 v14, 9, v14 v_mul_f32_e32 v20, 0x3fc90fda, v16 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_sub_nc_u32_e32 v14, v14, v19 v_fma_f32 v19, v16, 0x3fc90fda, -v20 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v14, 0x33000000, v14 v_fmac_f32_e32 v19, 0x33a22168, v16 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_or_b32_e32 v14, v14, v17 v_fmac_f32_e32 v19, 0x3fc90fda, v14 s_delay_alu instid0(VALU_DEP_1) v_add_f32_e32 v14, v20, v19 .LBB0_9: s_and_not1_saveexec_b32 s0, s14 s_cbranch_execz .LBB0_6 v_mul_f32_e64 v14, 0x3f22f983, |v13| s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_rndne_f32_e32 v15, v14 v_fma_f32 v14, v15, 0xbfc90fda, |v13| s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fmac_f32_e32 v14, 0xb3a22168, v15 v_fmac_f32_e32 v14, 0xa7c234c4, v15 v_cvt_i32_f32_e32 v15, v15 s_branch .LBB0_6 .LBB0_11: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z17fourier_transformPfS_iii .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 288 .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 26 .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 _Z17fourier_transformPfS_iii, .Lfunc_end0-_Z17fourier_transformPfS_iii .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 - .offset: 24 .size: 4 .value_kind: by_value - .offset: 32 .size: 4 .value_kind: hidden_block_count_x - .offset: 36 .size: 4 .value_kind: hidden_block_count_y - .offset: 40 .size: 4 .value_kind: hidden_block_count_z - .offset: 44 .size: 2 .value_kind: hidden_group_size_x - .offset: 46 .size: 2 .value_kind: hidden_group_size_y - .offset: 48 .size: 2 .value_kind: hidden_group_size_z - .offset: 50 .size: 2 .value_kind: hidden_remainder_x - .offset: 52 .size: 2 .value_kind: hidden_remainder_y - .offset: 54 .size: 2 .value_kind: hidden_remainder_z - .offset: 72 .size: 8 .value_kind: hidden_global_offset_x - .offset: 80 .size: 8 .value_kind: hidden_global_offset_y - .offset: 88 .size: 8 .value_kind: hidden_global_offset_z - .offset: 96 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 288 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z17fourier_transformPfS_iii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z17fourier_transformPfS_iii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 26 .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 fourier_transform(float *in, float *out, int height, int width, int blockConfig) { // block elements and function variables int my_x, k, t; my_x = blockIdx.x * blockDim.x + threadIdx.x; // iterate through each element, going from frequency to time domain for (k = 0; k < height; k++) { // difference, which will be used to subtract off float realSum = 0.0; // iterate through the input element for (t = 0; t < width; t++) { // calculate the angle and update the sum float angle = 2 * M_PI * (my_x * height + t) * (my_x * width + k) / height; realSum += in[my_x * height + t] * cos(angle); } // each output element will be the current sum for that index out[my_x * height + k] = realSum; } }
.text .file "fourier_transform.hip" .globl _Z32__device_stub__fourier_transformPfS_iii # -- Begin function _Z32__device_stub__fourier_transformPfS_iii .p2align 4, 0x90 .type _Z32__device_stub__fourier_transformPfS_iii,@function _Z32__device_stub__fourier_transformPfS_iii: # @_Z32__device_stub__fourier_transformPfS_iii .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%rsp) movl %ecx, 8(%rsp) movl %r8d, 4(%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 4(%rsp), %rax movq %rax, 112(%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 $_Z17fourier_transformPfS_iii, %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 _Z32__device_stub__fourier_transformPfS_iii, .Lfunc_end0-_Z32__device_stub__fourier_transformPfS_iii .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 $_Z17fourier_transformPfS_iii, %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 _Z17fourier_transformPfS_iii,@object # @_Z17fourier_transformPfS_iii .section .rodata,"a",@progbits .globl _Z17fourier_transformPfS_iii .p2align 3, 0x0 _Z17fourier_transformPfS_iii: .quad _Z32__device_stub__fourier_transformPfS_iii .size _Z17fourier_transformPfS_iii, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z17fourier_transformPfS_iii" .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 _Z32__device_stub__fourier_transformPfS_iii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z17fourier_transformPfS_iii .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_0002886e_00000000-6_fourier_transform.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__Z17fourier_transformPfS_iiiPfS_iii .type _Z42__device_stub__Z17fourier_transformPfS_iiiPfS_iii, @function _Z42__device_stub__Z17fourier_transformPfS_iiiPfS_iii: .LFB2051: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%rsp) movl %ecx, 8(%rsp) movl %r8d, 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 12(%rsp), %rax movq %rax, 112(%rsp) leaq 8(%rsp), %rax movq %rax, 120(%rsp) leaq 4(%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 _Z17fourier_transformPfS_iii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2051: .size _Z42__device_stub__Z17fourier_transformPfS_iiiPfS_iii, .-_Z42__device_stub__Z17fourier_transformPfS_iiiPfS_iii .globl _Z17fourier_transformPfS_iii .type _Z17fourier_transformPfS_iii, @function _Z17fourier_transformPfS_iii: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z42__device_stub__Z17fourier_transformPfS_iiiPfS_iii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z17fourier_transformPfS_iii, .-_Z17fourier_transformPfS_iii .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z17fourier_transformPfS_iii" .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 _Z17fourier_transformPfS_iii(%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 "fourier_transform.hip" .globl _Z32__device_stub__fourier_transformPfS_iii # -- Begin function _Z32__device_stub__fourier_transformPfS_iii .p2align 4, 0x90 .type _Z32__device_stub__fourier_transformPfS_iii,@function _Z32__device_stub__fourier_transformPfS_iii: # @_Z32__device_stub__fourier_transformPfS_iii .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%rsp) movl %ecx, 8(%rsp) movl %r8d, 4(%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 4(%rsp), %rax movq %rax, 112(%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 $_Z17fourier_transformPfS_iii, %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 _Z32__device_stub__fourier_transformPfS_iii, .Lfunc_end0-_Z32__device_stub__fourier_transformPfS_iii .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 $_Z17fourier_transformPfS_iii, %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 _Z17fourier_transformPfS_iii,@object # @_Z17fourier_transformPfS_iii .section .rodata,"a",@progbits .globl _Z17fourier_transformPfS_iii .p2align 3, 0x0 _Z17fourier_transformPfS_iii: .quad _Z32__device_stub__fourier_transformPfS_iii .size _Z17fourier_transformPfS_iii, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z17fourier_transformPfS_iii" .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 _Z32__device_stub__fourier_transformPfS_iii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z17fourier_transformPfS_iii .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.
//This code is a modification of L1 cache benchmark from //"Dissecting the NVIDIA Volta GPU Architecture via Microbenchmarking": https://arxiv.org/pdf/1804.06826.pdf //This benchmark measures the latency of L1 cache //This code have been tested on Volta V100 architecture #include <stdio.h> #include <stdlib.h> #include <cuda.h> #define SHARED_MEM_SIZE_BYTE (48*1024) //size in bytes, max 96KB for v100 #define SHARED_MEM_SIZE (SHARED_MEM_SIZE_BYTE/8) #define THREADS_NUM 32 //Launch only one thread to calcaulte the latency using a pointer-chasing array technique #define WARP_SIZE 32 #define ITERS 2048 //iterate over the array ITERS times // GPU error check #define gpuErrchk(ans) { gpuAssert((ans), __FILE__, __LINE__); } inline void gpuAssert(cudaError_t code, const char *file, int line, bool abort=true){ if (code != cudaSuccess) { fprintf(stderr,"GPUassert: %s %s %d\n", cudaGetErrorString(code), file, line); if (abort) exit(code); } } //TO DO: @Jason, please change the code to be similar to the L2/DRAM latency format //Measure latency of ITERS reads. __global__ void shared_lat(uint32_t *startClk, uint32_t *stopClk, uint64_t *dsink, uint32_t stride){ // thread index uint32_t tid = threadIdx.x; uint32_t bid = blockIdx.x; uint32_t uid = bid*blockDim.x+tid; uint32_t n_threads = blockDim.x * gridDim.x; __shared__ uint64_t s[SHARED_MEM_SIZE]; //static shared memory // one thread to initialize the pointer-chasing array for (uint32_t i=uid; i<(SHARED_MEM_SIZE-stride); i+=n_threads) s[i] = (i+stride)%SHARED_MEM_SIZE; if(uid == 0){ //initalize pointer chaser uint64_t p_chaser = 0; // start timing uint32_t start = 0; asm volatile ("mov.u32 %0, %%clock;" : "=r"(start) :: "memory"); // pointer-chasing ITERS times // use ca modifier to cache the load in L1 for(uint32_t i=0; i<ITERS; ++i) { p_chaser = s[p_chaser]; } // stop timing uint32_t stop = 0; asm volatile("mov.u32 %0, %%clock;" : "=r"(stop) :: "memory"); // write time and data back to memory startClk[uid] = start; stopClk[uid] = stop; dsink[uid] = p_chaser; } } int main(){ uint32_t *startClk = (uint32_t*) malloc(sizeof(uint32_t)); uint32_t *stopClk = (uint32_t*) malloc(sizeof(uint32_t)); uint64_t *dsink = (uint64_t*) malloc(sizeof(uint64_t)); uint32_t *startClk_g; uint32_t *stopClk_g; uint64_t *dsink_g; gpuErrchk( cudaMalloc(&startClk_g, sizeof(uint32_t)) ); gpuErrchk( cudaMalloc(&stopClk_g, sizeof(uint32_t)) ); gpuErrchk( cudaMalloc(&dsink_g, sizeof(uint64_t)) ); shared_lat<<<1,THREADS_NUM>>>(startClk_g, stopClk_g, dsink_g, 1); gpuErrchk( cudaPeekAtLastError() ); gpuErrchk( cudaMemcpy(startClk, startClk_g, sizeof(uint32_t), cudaMemcpyDeviceToHost) ); gpuErrchk( cudaMemcpy(stopClk, stopClk_g, sizeof(uint32_t), cudaMemcpyDeviceToHost) ); gpuErrchk( cudaMemcpy(dsink, dsink_g, sizeof(uint64_t), cudaMemcpyDeviceToHost) ); printf("Shared Memory Latency = %f cycles\n", (float)(stopClk[0]-startClk[0])/ITERS ); printf("Total Clk number = %u \n", stopClk[0]-startClk[0]); printf("start clk = %u \n", startClk[0]); printf("stop clk = %u \n", stopClk[0]); return 0; }
code for sm_80 Function : _Z10shared_latPjS_Pmj .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 */ /* 0x000e220000002500 */ /*0020*/ ULDC UR4, c[0x0][0x178] ; /* 0x00005e0000047ab9 */ /* 0x000fe20000000800 */ /*0030*/ BSSY B0, 0x150 ; /* 0x0000011000007945 */ /* 0x000fe20003800000 */ /*0040*/ UIADD3 UR4, -UR4, 0x1800, URZ ; /* 0x0000180004047890 */ /* 0x000fe2000fffe13f */ /*0050*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e240000002100 */ /*0060*/ IMAD R0, R0, c[0x0][0x0], R3 ; /* 0x0000000000007a24 */ /* 0x001fca00078e0203 */ /*0070*/ ISETP.GE.U32.AND P1, PT, R0.reuse, UR4, PT ; /* 0x0000000400007c0c */ /* 0x040fe4000bf26070 */ /*0080*/ ISETP.NE.AND P0, PT, R0, RZ, PT ; /* 0x000000ff0000720c */ /* 0x000fd60003f05270 */ /*0090*/ @P1 BRA 0x140 ; /* 0x000000a000001947 */ /* 0x000fea0003800000 */ /*00a0*/ IMAD.MOV.U32 R5, RZ, RZ, c[0x0][0x0] ; /* 0x00000000ff057624 */ /* 0x000fe400078e00ff */ /*00b0*/ IADD3 R4, R0, c[0x0][0x178], RZ ; /* 0x00005e0000047a10 */ /* 0x000fca0007ffe0ff */ /*00c0*/ IMAD.WIDE.U32 R2, R4, -0x55555555, RZ ; /* 0xaaaaaaab04027825 */ /* 0x000fca00078e00ff */ /*00d0*/ SHF.R.U32.HI R3, RZ, 0xc, R3 ; /* 0x0000000cff037819 */ /* 0x000fca0000011603 */ /*00e0*/ IMAD R2, R3, -0x1800, R4 ; /* 0xffffe80003027824 */ /* 0x000fe200078e0204 */ /*00f0*/ HFMA2.MMA R3, -RZ, RZ, 0, 0 ; /* 0x00000000ff037435 */ /* 0x000fce00000001ff */ /*0100*/ STS.64 [R0.X8], R2 ; /* 0x0000000200007388 */ /* 0x0001e40000008a00 */ /*0110*/ IMAD R0, R5, c[0x0][0xc], R0 ; /* 0x0000030005007a24 */ /* 0x001fca00078e0200 */ /*0120*/ ISETP.GE.U32.AND P1, PT, R0, UR4, PT ; /* 0x0000000400007c0c */ /* 0x000fda000bf26070 */ /*0130*/ @!P1 BRA 0xb0 ; /* 0xffffff7000009947 */ /* 0x000fea000383ffff */ /*0140*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*0150*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0160*/ CS2R.32 R0, SR_CLOCKLO ; /* 0x0000000000007805 */ /* 0x000fe40000005000 */ /*0170*/ IMAD.MOV.U32 R2, RZ, RZ, RZ ; /* 0x000000ffff027224 */ /* 0x000fe200078e00ff */ /*0180*/ CS2R R4, SRZ ; /* 0x0000000000047805 */ /* 0x000fe2000001ff00 */ /*0190*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fca0000000a00 */ /*01a0*/ IMAD.SHL.U32 R4, R4, 0x8, RZ ; /* 0x0000000804047824 */ /* 0x001fe200078e00ff */ /*01b0*/ IADD3 R2, R2, 0x40, RZ ; /* 0x0000004002027810 */ /* 0x000fc80007ffe0ff */ /*01c0*/ ISETP.NE.AND P0, PT, R2, 0x800, PT ; /* 0x000008000200780c */ /* 0x000fe20003f05270 */ /*01d0*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*01e0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*01f0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0200*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0210*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0220*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0230*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0240*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0250*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0260*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0270*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0280*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0290*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*02a0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*02b0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*02c0*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*02d0*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*02e0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*02f0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0300*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0310*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0320*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0330*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0340*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0350*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*0360*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*0370*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0380*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0390*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*03a0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*03b0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*03c0*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*03d0*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*03e0*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*03f0*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0400*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0410*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*0420*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*0430*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0440*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0450*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0460*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0470*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0480*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0490*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*04a0*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*04b0*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*04c0*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*04d0*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*04e0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*04f0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0500*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0510*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0520*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0530*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0540*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0550*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0560*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0570*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0580*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0590*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*05a0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*05b0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*05c0*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*05d0*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*05e0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*05f0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0600*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0610*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0620*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0630*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0640*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0650*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*0660*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*0670*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0680*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0690*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*06a0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*06b0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*06c0*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*06d0*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*06e0*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*06f0*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0700*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0710*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*0720*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*0730*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0740*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0750*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0760*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0770*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0780*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0790*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*07a0*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*07b0*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*07c0*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*07d0*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*07e0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*07f0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0800*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0810*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0820*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0830*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0840*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0850*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0860*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0870*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0880*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0890*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*08a0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*08b0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*08c0*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*08d0*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*08e0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*08f0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0900*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0910*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0920*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0930*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0940*/ IMAD.SHL.U32 R9, R8, 0x8, RZ ; /* 0x0000000808097824 */ /* 0x001fcc00078e00ff */ /*0950*/ LDS R9, [R9] ; /* 0x0000000009097984 */ /* 0x000e240000000800 */ /*0960*/ SHF.L.U32 R3, R9, 0x3, RZ ; /* 0x0000000309037819 */ /* 0x001fcc00000006ff */ /*0970*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0980*/ IMAD.SHL.U32 R10, R3, 0x8, RZ ; /* 0x00000008030a7824 */ /* 0x001fcc00078e00ff */ /*0990*/ LDS R10, [R10] ; /* 0x000000000a0a7984 */ /* 0x000e240000000800 */ /*09a0*/ SHF.L.U32 R4, R10, 0x3, RZ ; /* 0x000000030a047819 */ /* 0x001fcc00000006ff */ /*09b0*/ LDS.64 R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e220000000a00 */ /*09c0*/ @P0 BRA 0x1a0 ; /* 0xfffff7d000000947 */ /* 0x000fea000383ffff */ /*09d0*/ CS2R.32 R11, SR_CLOCKLO ; /* 0x00000000000b7805 */ /* 0x000fe40000005000 */ /*09e0*/ IMAD.MOV.U32 R2, RZ, RZ, c[0x0][0x160] ; /* 0x00005800ff027624 */ /* 0x000fe200078e00ff */ /*09f0*/ MOV R3, c[0x0][0x164] ; /* 0x0000590000037a02 */ /* 0x000fe20000000f00 */ /*0a00*/ IMAD.MOV.U32 R6, RZ, RZ, c[0x0][0x168] ; /* 0x00005a00ff067624 */ /* 0x000fe200078e00ff */ /*0a10*/ MOV R7, c[0x0][0x16c] ; /* 0x00005b0000077a02 */ /* 0x000fe20000000f00 */ /*0a20*/ IMAD.MOV.U32 R8, RZ, RZ, c[0x0][0x170] ; /* 0x00005c00ff087624 */ /* 0x000fe200078e00ff */ /*0a30*/ MOV R9, c[0x0][0x174] ; /* 0x00005d0000097a02 */ /* 0x000fe20000000f00 */ /*0a40*/ STG.E [R2.64], R0 ; /* 0x0000000002007986 */ /* 0x000fe8000c101904 */ /*0a50*/ STG.E [R6.64], R11 ; /* 0x0000000b06007986 */ /* 0x000fe8000c101904 */ /*0a60*/ STG.E.64 [R8.64], R4 ; /* 0x0000000408007986 */ /* 0x001fe2000c101b04 */ /*0a70*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0a80*/ BRA 0xa80; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*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 */ /*0b00*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b10*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b20*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b30*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b40*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b50*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b60*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b70*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
//This code is a modification of L1 cache benchmark from //"Dissecting the NVIDIA Volta GPU Architecture via Microbenchmarking": https://arxiv.org/pdf/1804.06826.pdf //This benchmark measures the latency of L1 cache //This code have been tested on Volta V100 architecture #include <stdio.h> #include <stdlib.h> #include <cuda.h> #define SHARED_MEM_SIZE_BYTE (48*1024) //size in bytes, max 96KB for v100 #define SHARED_MEM_SIZE (SHARED_MEM_SIZE_BYTE/8) #define THREADS_NUM 32 //Launch only one thread to calcaulte the latency using a pointer-chasing array technique #define WARP_SIZE 32 #define ITERS 2048 //iterate over the array ITERS times // GPU error check #define gpuErrchk(ans) { gpuAssert((ans), __FILE__, __LINE__); } inline void gpuAssert(cudaError_t code, const char *file, int line, bool abort=true){ if (code != cudaSuccess) { fprintf(stderr,"GPUassert: %s %s %d\n", cudaGetErrorString(code), file, line); if (abort) exit(code); } } //TO DO: @Jason, please change the code to be similar to the L2/DRAM latency format //Measure latency of ITERS reads. __global__ void shared_lat(uint32_t *startClk, uint32_t *stopClk, uint64_t *dsink, uint32_t stride){ // thread index uint32_t tid = threadIdx.x; uint32_t bid = blockIdx.x; uint32_t uid = bid*blockDim.x+tid; uint32_t n_threads = blockDim.x * gridDim.x; __shared__ uint64_t s[SHARED_MEM_SIZE]; //static shared memory // one thread to initialize the pointer-chasing array for (uint32_t i=uid; i<(SHARED_MEM_SIZE-stride); i+=n_threads) s[i] = (i+stride)%SHARED_MEM_SIZE; if(uid == 0){ //initalize pointer chaser uint64_t p_chaser = 0; // start timing uint32_t start = 0; asm volatile ("mov.u32 %0, %%clock;" : "=r"(start) :: "memory"); // pointer-chasing ITERS times // use ca modifier to cache the load in L1 for(uint32_t i=0; i<ITERS; ++i) { p_chaser = s[p_chaser]; } // stop timing uint32_t stop = 0; asm volatile("mov.u32 %0, %%clock;" : "=r"(stop) :: "memory"); // write time and data back to memory startClk[uid] = start; stopClk[uid] = stop; dsink[uid] = p_chaser; } } int main(){ uint32_t *startClk = (uint32_t*) malloc(sizeof(uint32_t)); uint32_t *stopClk = (uint32_t*) malloc(sizeof(uint32_t)); uint64_t *dsink = (uint64_t*) malloc(sizeof(uint64_t)); uint32_t *startClk_g; uint32_t *stopClk_g; uint64_t *dsink_g; gpuErrchk( cudaMalloc(&startClk_g, sizeof(uint32_t)) ); gpuErrchk( cudaMalloc(&stopClk_g, sizeof(uint32_t)) ); gpuErrchk( cudaMalloc(&dsink_g, sizeof(uint64_t)) ); shared_lat<<<1,THREADS_NUM>>>(startClk_g, stopClk_g, dsink_g, 1); gpuErrchk( cudaPeekAtLastError() ); gpuErrchk( cudaMemcpy(startClk, startClk_g, sizeof(uint32_t), cudaMemcpyDeviceToHost) ); gpuErrchk( cudaMemcpy(stopClk, stopClk_g, sizeof(uint32_t), cudaMemcpyDeviceToHost) ); gpuErrchk( cudaMemcpy(dsink, dsink_g, sizeof(uint64_t), cudaMemcpyDeviceToHost) ); printf("Shared Memory Latency = %f cycles\n", (float)(stopClk[0]-startClk[0])/ITERS ); printf("Total Clk number = %u \n", stopClk[0]-startClk[0]); printf("start clk = %u \n", startClk[0]); printf("stop clk = %u \n", stopClk[0]); return 0; }
.file "tmpxft_00137c55_00000000-6_shared_lat.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._Z9gpuAssert9cudaErrorPKcib.str1.1,"aMS",@progbits,1 .LC0: .string "GPUassert: %s %s %d\n" .section .text._Z9gpuAssert9cudaErrorPKcib,"axG",@progbits,_Z9gpuAssert9cudaErrorPKcib,comdat .weak _Z9gpuAssert9cudaErrorPKcib .type _Z9gpuAssert9cudaErrorPKcib, @function _Z9gpuAssert9cudaErrorPKcib: .LFB2057: .cfi_startproc endbr64 testl %edi, %edi jne .L9 ret .L9: 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 $8, %rsp .cfi_def_cfa_offset 48 movl %edi, %ebx movq %rsi, %r13 movl %edx, %r12d movl %ecx, %ebp call cudaGetErrorString@PLT movq %rax, %rcx movl %r12d, %r9d movq %r13, %r8 leaq .LC0(%rip), %rdx movl $2, %esi movq stderr(%rip), %rdi movl $0, %eax call __fprintf_chk@PLT testb %bpl, %bpl jne .L10 addq $8, %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 .L10: .cfi_restore_state movl %ebx, %edi call exit@PLT .cfi_endproc .LFE2057: .size _Z9gpuAssert9cudaErrorPKcib, .-_Z9gpuAssert9cudaErrorPKcib .text .globl _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj .type _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj, @function _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj: .LFB2083: .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 .L15 .L11: movq 136(%rsp), %rax subq %fs:40, %rax jne .L16 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L15: .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 _Z10shared_latPjS_Pmj(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L11 .L16: call __stack_chk_fail@PLT .cfi_endproc .LFE2083: .size _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj, .-_Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj .globl _Z10shared_latPjS_Pmj .type _Z10shared_latPjS_Pmj, @function _Z10shared_latPjS_Pmj: .LFB2084: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2084: .size _Z10shared_latPjS_Pmj, .-_Z10shared_latPjS_Pmj .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC1: .string "/home/ubuntu/Datasets/stackv2/train-structured/purdue-aalp/gpgpu-sim_simulations/master/benchmarks/src/cuda/GPU_Microbenchmark/shared_lat/shared_lat.cu" .align 8 .LC3: .string "Shared Memory Latency = %f cycles\n" .section .rodata.str1.1,"aMS",@progbits,1 .LC4: .string "Total Clk number = %u \n" .LC5: .string "start clk = %u \n" .LC6: .string "stop clk = %u \n" .text .globl main .type main, @function main: .LFB2058: .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 $72, %rsp .cfi_def_cfa_offset 112 movq %fs:40, %rax movq %rax, 56(%rsp) xorl %eax, %eax movl $4, %edi call malloc@PLT movq %rax, %rbp movl $4, %edi call malloc@PLT movq %rax, %rbx movl $8, %edi call malloc@PLT movq %rax, %r13 leaq 8(%rsp), %rdi movl $4, %esi call cudaMalloc@PLT movl %eax, %edi movl $1, %ecx movl $79, %edx leaq .LC1(%rip), %r12 movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib leaq 16(%rsp), %rdi movl $4, %esi call cudaMalloc@PLT movl %eax, %edi movl $1, %ecx movl $80, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib leaq 24(%rsp), %rdi movl $8, %esi call cudaMalloc@PLT movl %eax, %edi movl $1, %ecx movl $81, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl $32, 44(%rsp) movl $1, 48(%rsp) movl $1, 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 .L25 .L20: call cudaPeekAtLastError@PLT movl %eax, %edi movl $1, %ecx movl $84, %edx leaq .LC1(%rip), %r12 movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl $2, %ecx movl $4, %edx movq 8(%rsp), %rsi movq %rbp, %rdi call cudaMemcpy@PLT movl %eax, %edi movl $1, %ecx movl $86, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl $2, %ecx movl $4, %edx movq 16(%rsp), %rsi movq %rbx, %rdi call cudaMemcpy@PLT movl %eax, %edi movl $1, %ecx movl $87, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl $2, %ecx movl $8, %edx movq 24(%rsp), %rsi movq %r13, %rdi call cudaMemcpy@PLT movl %eax, %edi movl $1, %ecx movl $88, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl (%rbx), %eax subl 0(%rbp), %eax pxor %xmm0, %xmm0 cvtsi2ssq %rax, %xmm0 mulss .LC2(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 leaq .LC3(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movl (%rbx), %edx subl 0(%rbp), %edx leaq .LC4(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl 0(%rbp), %edx leaq .LC5(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl (%rbx), %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq 56(%rsp), %rax subq %fs:40, %rax jne .L26 movl $0, %eax addq $72, %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 .L25: .cfi_restore_state movl $1, %ecx movq 24(%rsp), %rdx movq 16(%rsp), %rsi movq 8(%rsp), %rdi call _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj jmp .L20 .L26: call __stack_chk_fail@PLT .cfi_endproc .LFE2058: .size main, .-main .section .rodata.str1.1 .LC7: .string "_Z10shared_latPjS_Pmj" .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 .LC7(%rip), %rdx movq %rdx, %rcx leaq _Z10shared_latPjS_Pmj(%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 .LC2: .long 973078528 .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.
//This code is a modification of L1 cache benchmark from //"Dissecting the NVIDIA Volta GPU Architecture via Microbenchmarking": https://arxiv.org/pdf/1804.06826.pdf //This benchmark measures the latency of L1 cache //This code have been tested on Volta V100 architecture #include <stdio.h> #include <stdlib.h> #include <cuda.h> #define SHARED_MEM_SIZE_BYTE (48*1024) //size in bytes, max 96KB for v100 #define SHARED_MEM_SIZE (SHARED_MEM_SIZE_BYTE/8) #define THREADS_NUM 32 //Launch only one thread to calcaulte the latency using a pointer-chasing array technique #define WARP_SIZE 32 #define ITERS 2048 //iterate over the array ITERS times // GPU error check #define gpuErrchk(ans) { gpuAssert((ans), __FILE__, __LINE__); } inline void gpuAssert(cudaError_t code, const char *file, int line, bool abort=true){ if (code != cudaSuccess) { fprintf(stderr,"GPUassert: %s %s %d\n", cudaGetErrorString(code), file, line); if (abort) exit(code); } } //TO DO: @Jason, please change the code to be similar to the L2/DRAM latency format //Measure latency of ITERS reads. __global__ void shared_lat(uint32_t *startClk, uint32_t *stopClk, uint64_t *dsink, uint32_t stride){ // thread index uint32_t tid = threadIdx.x; uint32_t bid = blockIdx.x; uint32_t uid = bid*blockDim.x+tid; uint32_t n_threads = blockDim.x * gridDim.x; __shared__ uint64_t s[SHARED_MEM_SIZE]; //static shared memory // one thread to initialize the pointer-chasing array for (uint32_t i=uid; i<(SHARED_MEM_SIZE-stride); i+=n_threads) s[i] = (i+stride)%SHARED_MEM_SIZE; if(uid == 0){ //initalize pointer chaser uint64_t p_chaser = 0; // start timing uint32_t start = 0; asm volatile ("mov.u32 %0, %%clock;" : "=r"(start) :: "memory"); // pointer-chasing ITERS times // use ca modifier to cache the load in L1 for(uint32_t i=0; i<ITERS; ++i) { p_chaser = s[p_chaser]; } // stop timing uint32_t stop = 0; asm volatile("mov.u32 %0, %%clock;" : "=r"(stop) :: "memory"); // write time and data back to memory startClk[uid] = start; stopClk[uid] = stop; dsink[uid] = p_chaser; } } int main(){ uint32_t *startClk = (uint32_t*) malloc(sizeof(uint32_t)); uint32_t *stopClk = (uint32_t*) malloc(sizeof(uint32_t)); uint64_t *dsink = (uint64_t*) malloc(sizeof(uint64_t)); uint32_t *startClk_g; uint32_t *stopClk_g; uint64_t *dsink_g; gpuErrchk( cudaMalloc(&startClk_g, sizeof(uint32_t)) ); gpuErrchk( cudaMalloc(&stopClk_g, sizeof(uint32_t)) ); gpuErrchk( cudaMalloc(&dsink_g, sizeof(uint64_t)) ); shared_lat<<<1,THREADS_NUM>>>(startClk_g, stopClk_g, dsink_g, 1); gpuErrchk( cudaPeekAtLastError() ); gpuErrchk( cudaMemcpy(startClk, startClk_g, sizeof(uint32_t), cudaMemcpyDeviceToHost) ); gpuErrchk( cudaMemcpy(stopClk, stopClk_g, sizeof(uint32_t), cudaMemcpyDeviceToHost) ); gpuErrchk( cudaMemcpy(dsink, dsink_g, sizeof(uint64_t), cudaMemcpyDeviceToHost) ); printf("Shared Memory Latency = %f cycles\n", (float)(stopClk[0]-startClk[0])/ITERS ); printf("Total Clk number = %u \n", stopClk[0]-startClk[0]); printf("start clk = %u \n", startClk[0]); printf("stop clk = %u \n", stopClk[0]); return 0; }
//This code is a modification of L1 cache benchmark from //"Dissecting the NVIDIA Volta GPU Architecture via Microbenchmarking": https://arxiv.org/pdf/1804.06826.pdf //This benchmark measures the latency of L1 cache //This code have been tested on Volta V100 architecture #include <stdio.h> #include <stdlib.h> #include <hip/hip_runtime.h> #define SHARED_MEM_SIZE_BYTE (48*1024) //size in bytes, max 96KB for v100 #define SHARED_MEM_SIZE (SHARED_MEM_SIZE_BYTE/8) #define THREADS_NUM 32 //Launch only one thread to calcaulte the latency using a pointer-chasing array technique #define WARP_SIZE 32 #define ITERS 2048 //iterate over the array ITERS times // GPU error check #define gpuErrchk(ans) { gpuAssert((ans), __FILE__, __LINE__); } inline void gpuAssert(hipError_t code, const char *file, int line, bool abort=true){ if (code != hipSuccess) { fprintf(stderr,"GPUassert: %s %s %d\n", hipGetErrorString(code), file, line); if (abort) exit(code); } } //TO DO: @Jason, please change the code to be similar to the L2/DRAM latency format //Measure latency of ITERS reads. __global__ void shared_lat(uint32_t *startClk, uint32_t *stopClk, uint64_t *dsink, uint32_t stride){ // thread index uint32_t tid = threadIdx.x; uint32_t bid = blockIdx.x; uint32_t uid = bid*blockDim.x+tid; uint32_t n_threads = blockDim.x * gridDim.x; __shared__ uint64_t s[SHARED_MEM_SIZE]; //static shared memory // one thread to initialize the pointer-chasing array for (uint32_t i=uid; i<(SHARED_MEM_SIZE-stride); i+=n_threads) s[i] = (i+stride)%SHARED_MEM_SIZE; if(uid == 0){ //initalize pointer chaser uint64_t p_chaser = 0; // start timing uint32_t start = 0; asm volatile ("mov.u32 %0, %%clock;" : "=r"(start) :: "memory"); // pointer-chasing ITERS times // use ca modifier to cache the load in L1 for(uint32_t i=0; i<ITERS; ++i) { p_chaser = s[p_chaser]; } // stop timing uint32_t stop = 0; asm volatile("mov.u32 %0, %%clock;" : "=r"(stop) :: "memory"); // write time and data back to memory startClk[uid] = start; stopClk[uid] = stop; dsink[uid] = p_chaser; } } int main(){ uint32_t *startClk = (uint32_t*) malloc(sizeof(uint32_t)); uint32_t *stopClk = (uint32_t*) malloc(sizeof(uint32_t)); uint64_t *dsink = (uint64_t*) malloc(sizeof(uint64_t)); uint32_t *startClk_g; uint32_t *stopClk_g; uint64_t *dsink_g; gpuErrchk( hipMalloc(&startClk_g, sizeof(uint32_t)) ); gpuErrchk( hipMalloc(&stopClk_g, sizeof(uint32_t)) ); gpuErrchk( hipMalloc(&dsink_g, sizeof(uint64_t)) ); shared_lat<<<1,THREADS_NUM>>>(startClk_g, stopClk_g, dsink_g, 1); gpuErrchk( hipPeekAtLastError() ); gpuErrchk( hipMemcpy(startClk, startClk_g, sizeof(uint32_t), hipMemcpyDeviceToHost) ); gpuErrchk( hipMemcpy(stopClk, stopClk_g, sizeof(uint32_t), hipMemcpyDeviceToHost) ); gpuErrchk( hipMemcpy(dsink, dsink_g, sizeof(uint64_t), hipMemcpyDeviceToHost) ); printf("Shared Memory Latency = %f cycles\n", (float)(stopClk[0]-startClk[0])/ITERS ); printf("Total Clk number = %u \n", stopClk[0]-startClk[0]); printf("start clk = %u \n", startClk[0]); printf("stop clk = %u \n", stopClk[0]); return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
//This code is a modification of L1 cache benchmark from //"Dissecting the NVIDIA Volta GPU Architecture via Microbenchmarking": https://arxiv.org/pdf/1804.06826.pdf //This benchmark measures the latency of L1 cache //This code have been tested on Volta V100 architecture #include <stdio.h> #include <stdlib.h> #include <hip/hip_runtime.h> #define SHARED_MEM_SIZE_BYTE (48*1024) //size in bytes, max 96KB for v100 #define SHARED_MEM_SIZE (SHARED_MEM_SIZE_BYTE/8) #define THREADS_NUM 32 //Launch only one thread to calcaulte the latency using a pointer-chasing array technique #define WARP_SIZE 32 #define ITERS 2048 //iterate over the array ITERS times // GPU error check #define gpuErrchk(ans) { gpuAssert((ans), __FILE__, __LINE__); } inline void gpuAssert(hipError_t code, const char *file, int line, bool abort=true){ if (code != hipSuccess) { fprintf(stderr,"GPUassert: %s %s %d\n", hipGetErrorString(code), file, line); if (abort) exit(code); } } //TO DO: @Jason, please change the code to be similar to the L2/DRAM latency format //Measure latency of ITERS reads. __global__ void shared_lat(uint32_t *startClk, uint32_t *stopClk, uint64_t *dsink, uint32_t stride){ // thread index uint32_t tid = threadIdx.x; uint32_t bid = blockIdx.x; uint32_t uid = bid*blockDim.x+tid; uint32_t n_threads = blockDim.x * gridDim.x; __shared__ uint64_t s[SHARED_MEM_SIZE]; //static shared memory // one thread to initialize the pointer-chasing array for (uint32_t i=uid; i<(SHARED_MEM_SIZE-stride); i+=n_threads) s[i] = (i+stride)%SHARED_MEM_SIZE; if(uid == 0){ //initalize pointer chaser uint64_t p_chaser = 0; // start timing uint32_t start = 0; asm volatile ("mov.u32 %0, %%clock;" : "=r"(start) :: "memory"); // pointer-chasing ITERS times // use ca modifier to cache the load in L1 for(uint32_t i=0; i<ITERS; ++i) { p_chaser = s[p_chaser]; } // stop timing uint32_t stop = 0; asm volatile("mov.u32 %0, %%clock;" : "=r"(stop) :: "memory"); // write time and data back to memory startClk[uid] = start; stopClk[uid] = stop; dsink[uid] = p_chaser; } } int main(){ uint32_t *startClk = (uint32_t*) malloc(sizeof(uint32_t)); uint32_t *stopClk = (uint32_t*) malloc(sizeof(uint32_t)); uint64_t *dsink = (uint64_t*) malloc(sizeof(uint64_t)); uint32_t *startClk_g; uint32_t *stopClk_g; uint64_t *dsink_g; gpuErrchk( hipMalloc(&startClk_g, sizeof(uint32_t)) ); gpuErrchk( hipMalloc(&stopClk_g, sizeof(uint32_t)) ); gpuErrchk( hipMalloc(&dsink_g, sizeof(uint64_t)) ); shared_lat<<<1,THREADS_NUM>>>(startClk_g, stopClk_g, dsink_g, 1); gpuErrchk( hipPeekAtLastError() ); gpuErrchk( hipMemcpy(startClk, startClk_g, sizeof(uint32_t), hipMemcpyDeviceToHost) ); gpuErrchk( hipMemcpy(stopClk, stopClk_g, sizeof(uint32_t), hipMemcpyDeviceToHost) ); gpuErrchk( hipMemcpy(dsink, dsink_g, sizeof(uint64_t), hipMemcpyDeviceToHost) ); printf("Shared Memory Latency = %f cycles\n", (float)(stopClk[0]-startClk[0])/ITERS ); printf("Total Clk number = %u \n", stopClk[0]-startClk[0]); printf("start clk = %u \n", startClk[0]); printf("stop clk = %u \n", stopClk[0]); return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z10shared_latPjS_Pmj .globl _Z10shared_latPjS_Pmj .p2align 8 .type _Z10shared_latPjS_Pmj,@function _Z10shared_latPjS_Pmj: s_clause 0x1 s_load_b32 s5, s[0:1], 0x2c s_load_b32 s4, s[0:1], 0x18 s_add_u32 s2, s0, 32 s_addc_u32 s3, s1, 0 s_mov_b32 s6, exec_lo s_waitcnt lgkmcnt(0) s_and_b32 s7, s5, 0xffff s_sub_i32 s5, 0x1800, s4 s_mul_i32 s15, s15, s7 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_nc_u32_e32 v2, s15, v0 v_cmpx_gt_u32_e64 s5, v2 s_cbranch_execz .LBB0_3 s_load_b32 s2, s[2:3], 0x0 v_add3_u32 v3, s15, s4, v0 v_dual_mov_b32 v1, 0 :: v_dual_lshlrev_b32 v4, 3, v2 s_mov_b32 s3, 0 s_mov_b32 s8, 0 s_waitcnt lgkmcnt(0) s_mul_i32 s2, s2, s7 s_delay_alu instid0(SALU_CYCLE_1) s_lshl_b32 s7, s2, 3 .p2align 6 .LBB0_2: v_mul_hi_u32 v0, v3, 0xaaaaaaab v_add_nc_u32_e32 v3, s2, v3 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshrrev_b32_e32 v0, 12, v0 v_mul_u32_u24_e32 v0, 0x1800, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_sub_nc_u32_e32 v0, s4, v0 v_add3_u32 v0, v2, s8, v0 s_add_i32 s8, s8, s2 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_4) | instid1(SALU_CYCLE_1) v_add_nc_u32_e32 v5, s8, v2 ds_store_b64 v4, v[0:1] v_add_nc_u32_e32 v4, s7, v4 v_cmp_le_u32_e32 vcc_lo, s5, v5 s_or_b32 s3, vcc_lo, s3 s_and_not1_b32 exec_lo, exec_lo, s3 s_cbranch_execnz .LBB0_2 .LBB0_3: s_or_b32 exec_lo, exec_lo, s6 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s2, exec_lo v_cmpx_eq_u32_e32 0, v2 s_cbranch_execz .LBB0_7 mov.u32 s2, %clock s_movk_i32 s3, 0x800 v_mov_b32_e32 v0, 0 v_mov_b32_e32 v1, 0 .LBB0_5: s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_lshlrev_b32_e32 v0, 3, v0 s_add_i32 s3, s3, -1 s_cmp_eq_u32 s3, 0 ds_load_b64 v[0:1], v0 s_cbranch_scc0 .LBB0_5 s_clause 0x1 s_load_b128 s[4:7], s[0:1], 0x0 s_load_b64 s[0:1], s[0:1], 0x10 v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v3, s2 mov.u32 s3, %clock v_mov_b32_e32 v4, s3 s_waitcnt lgkmcnt(0) s_clause 0x2 global_store_b32 v2, v3, s[4:5] global_store_b32 v2, v4, s[6:7] global_store_b64 v2, v[0:1], s[0:1] .LBB0_7: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z10shared_latPjS_Pmj .amdhsa_group_segment_fixed_size 49152 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 288 .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 _Z10shared_latPjS_Pmj, .Lfunc_end0-_Z10shared_latPjS_Pmj .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 - .offset: 32 .size: 4 .value_kind: hidden_block_count_x - .offset: 36 .size: 4 .value_kind: hidden_block_count_y - .offset: 40 .size: 4 .value_kind: hidden_block_count_z - .offset: 44 .size: 2 .value_kind: hidden_group_size_x - .offset: 46 .size: 2 .value_kind: hidden_group_size_y - .offset: 48 .size: 2 .value_kind: hidden_group_size_z - .offset: 50 .size: 2 .value_kind: hidden_remainder_x - .offset: 52 .size: 2 .value_kind: hidden_remainder_y - .offset: 54 .size: 2 .value_kind: hidden_remainder_z - .offset: 72 .size: 8 .value_kind: hidden_global_offset_x - .offset: 80 .size: 8 .value_kind: hidden_global_offset_y - .offset: 88 .size: 8 .value_kind: hidden_global_offset_z - .offset: 96 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 49152 .kernarg_segment_align: 8 .kernarg_segment_size: 288 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z10shared_latPjS_Pmj .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z10shared_latPjS_Pmj.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.
//This code is a modification of L1 cache benchmark from //"Dissecting the NVIDIA Volta GPU Architecture via Microbenchmarking": https://arxiv.org/pdf/1804.06826.pdf //This benchmark measures the latency of L1 cache //This code have been tested on Volta V100 architecture #include <stdio.h> #include <stdlib.h> #include <hip/hip_runtime.h> #define SHARED_MEM_SIZE_BYTE (48*1024) //size in bytes, max 96KB for v100 #define SHARED_MEM_SIZE (SHARED_MEM_SIZE_BYTE/8) #define THREADS_NUM 32 //Launch only one thread to calcaulte the latency using a pointer-chasing array technique #define WARP_SIZE 32 #define ITERS 2048 //iterate over the array ITERS times // GPU error check #define gpuErrchk(ans) { gpuAssert((ans), __FILE__, __LINE__); } inline void gpuAssert(hipError_t code, const char *file, int line, bool abort=true){ if (code != hipSuccess) { fprintf(stderr,"GPUassert: %s %s %d\n", hipGetErrorString(code), file, line); if (abort) exit(code); } } //TO DO: @Jason, please change the code to be similar to the L2/DRAM latency format //Measure latency of ITERS reads. __global__ void shared_lat(uint32_t *startClk, uint32_t *stopClk, uint64_t *dsink, uint32_t stride){ // thread index uint32_t tid = threadIdx.x; uint32_t bid = blockIdx.x; uint32_t uid = bid*blockDim.x+tid; uint32_t n_threads = blockDim.x * gridDim.x; __shared__ uint64_t s[SHARED_MEM_SIZE]; //static shared memory // one thread to initialize the pointer-chasing array for (uint32_t i=uid; i<(SHARED_MEM_SIZE-stride); i+=n_threads) s[i] = (i+stride)%SHARED_MEM_SIZE; if(uid == 0){ //initalize pointer chaser uint64_t p_chaser = 0; // start timing uint32_t start = 0; asm volatile ("mov.u32 %0, %%clock;" : "=r"(start) :: "memory"); // pointer-chasing ITERS times // use ca modifier to cache the load in L1 for(uint32_t i=0; i<ITERS; ++i) { p_chaser = s[p_chaser]; } // stop timing uint32_t stop = 0; asm volatile("mov.u32 %0, %%clock;" : "=r"(stop) :: "memory"); // write time and data back to memory startClk[uid] = start; stopClk[uid] = stop; dsink[uid] = p_chaser; } } int main(){ uint32_t *startClk = (uint32_t*) malloc(sizeof(uint32_t)); uint32_t *stopClk = (uint32_t*) malloc(sizeof(uint32_t)); uint64_t *dsink = (uint64_t*) malloc(sizeof(uint64_t)); uint32_t *startClk_g; uint32_t *stopClk_g; uint64_t *dsink_g; gpuErrchk( hipMalloc(&startClk_g, sizeof(uint32_t)) ); gpuErrchk( hipMalloc(&stopClk_g, sizeof(uint32_t)) ); gpuErrchk( hipMalloc(&dsink_g, sizeof(uint64_t)) ); shared_lat<<<1,THREADS_NUM>>>(startClk_g, stopClk_g, dsink_g, 1); gpuErrchk( hipPeekAtLastError() ); gpuErrchk( hipMemcpy(startClk, startClk_g, sizeof(uint32_t), hipMemcpyDeviceToHost) ); gpuErrchk( hipMemcpy(stopClk, stopClk_g, sizeof(uint32_t), hipMemcpyDeviceToHost) ); gpuErrchk( hipMemcpy(dsink, dsink_g, sizeof(uint64_t), hipMemcpyDeviceToHost) ); printf("Shared Memory Latency = %f cycles\n", (float)(stopClk[0]-startClk[0])/ITERS ); printf("Total Clk number = %u \n", stopClk[0]-startClk[0]); printf("start clk = %u \n", startClk[0]); printf("stop clk = %u \n", stopClk[0]); return 0; }
.text .file "shared_lat.hip" .globl _Z25__device_stub__shared_latPjS_Pmj # -- Begin function _Z25__device_stub__shared_latPjS_Pmj .p2align 4, 0x90 .type _Z25__device_stub__shared_latPjS_Pmj,@function _Z25__device_stub__shared_latPjS_Pmj: # @_Z25__device_stub__shared_latPjS_Pmj .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 $_Z10shared_latPjS_Pmj, %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 _Z25__device_stub__shared_latPjS_Pmj, .Lfunc_end0-_Z25__device_stub__shared_latPjS_Pmj .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function main .LCPI1_0: .long 0x3a000000 # float 4.8828125E-4 .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 %rbx .cfi_def_cfa_offset 40 subq $152, %rsp .cfi_def_cfa_offset 192 .cfi_offset %rbx, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl $4, %edi callq malloc movq %rax, %r14 movl $4, %edi callq malloc movq %rax, %rbx movl $8, %edi callq malloc movq %rax, %r15 leaq 32(%rsp), %rdi movl $4, %esi callq hipMalloc testl %eax, %eax jne .LBB1_1 # %bb.3: # %_Z9gpuAssert10hipError_tPKcib.exit leaq 24(%rsp), %rdi movl $4, %esi callq hipMalloc testl %eax, %eax jne .LBB1_4 # %bb.5: # %_Z9gpuAssert10hipError_tPKcib.exit12 leaq 16(%rsp), %rdi movl $8, %esi callq hipMalloc testl %eax, %eax jne .LBB1_6 # %bb.7: # %_Z9gpuAssert10hipError_tPKcib.exit14 movabsq $4294967297, %rdi # imm = 0x100000001 leaq 31(%rdi), %rdx movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_9 # %bb.8: movq 32(%rsp), %rax movq 24(%rsp), %rcx movq 16(%rsp), %rdx movq %rax, 104(%rsp) movq %rcx, 96(%rsp) movq %rdx, 88(%rsp) movl $1, 12(%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 12(%rsp), %rax movq %rax, 136(%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 $_Z10shared_latPjS_Pmj, %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_9: callq hipPeekAtLastError testl %eax, %eax jne .LBB1_10 # %bb.11: # %_Z9gpuAssert10hipError_tPKcib.exit16 movq 32(%rsp), %rsi movl $4, %edx movq %r14, %rdi movl $2, %ecx callq hipMemcpy testl %eax, %eax jne .LBB1_12 # %bb.13: # %_Z9gpuAssert10hipError_tPKcib.exit18 movq 24(%rsp), %rsi movl $4, %edx movq %rbx, %rdi movl $2, %ecx callq hipMemcpy testl %eax, %eax jne .LBB1_14 # %bb.15: # %_Z9gpuAssert10hipError_tPKcib.exit20 movq 16(%rsp), %rsi movl $8, %edx movq %r15, %rdi movl $2, %ecx callq hipMemcpy testl %eax, %eax jne .LBB1_16 # %bb.17: # %_Z9gpuAssert10hipError_tPKcib.exit22 movl (%rbx), %eax subl (%r14), %eax cvtsi2ss %rax, %xmm0 mulss .LCPI1_0(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 movl $.L.str.1, %edi movb $1, %al callq printf movl (%rbx), %esi subl (%r14), %esi movl $.L.str.2, %edi xorl %eax, %eax callq printf movl (%r14), %esi movl $.L.str.3, %edi xorl %eax, %eax callq printf movl (%rbx), %esi movl $.L.str.4, %edi xorl %eax, %eax callq printf xorl %eax, %eax addq $152, %rsp .cfi_def_cfa_offset 40 popq %rbx .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 .LBB1_1: .cfi_def_cfa_offset 192 movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $79, %r8d jmp .LBB1_2 .LBB1_4: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $80, %r8d jmp .LBB1_2 .LBB1_6: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $81, %r8d jmp .LBB1_2 .LBB1_10: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $84, %r8d jmp .LBB1_2 .LBB1_12: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $86, %r8d jmp .LBB1_2 .LBB1_14: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $87, %r8d jmp .LBB1_2 .LBB1_16: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $88, %r8d .LBB1_2: xorl %eax, %eax callq fprintf movl %ebp, %edi callq exit .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 $_Z10shared_latPjS_Pmj, %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 _Z10shared_latPjS_Pmj,@object # @_Z10shared_latPjS_Pmj .section .rodata,"a",@progbits .globl _Z10shared_latPjS_Pmj .p2align 3, 0x0 _Z10shared_latPjS_Pmj: .quad _Z25__device_stub__shared_latPjS_Pmj .size _Z10shared_latPjS_Pmj, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "/home/ubuntu/Datasets/stackv2/train-structured-repos-hip/purdue-aalp/gpgpu-sim_simulations/master/benchmarks/src/cuda/GPU_Microbenchmark/shared_lat/shared_lat.hip" .size .L.str, 163 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz "Shared Memory Latency = %f cycles\n" .size .L.str.1, 36 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "Total Clk number = %u \n" .size .L.str.2, 24 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "start clk = %u \n" .size .L.str.3, 17 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "stop clk = %u \n" .size .L.str.4, 16 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "GPUassert: %s %s %d\n" .size .L.str.5, 21 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z10shared_latPjS_Pmj" .size .L__unnamed_1, 22 .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__shared_latPjS_Pmj .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z10shared_latPjS_Pmj .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 : _Z10shared_latPjS_Pmj .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 */ /* 0x000e220000002500 */ /*0020*/ ULDC UR4, c[0x0][0x178] ; /* 0x00005e0000047ab9 */ /* 0x000fe20000000800 */ /*0030*/ BSSY B0, 0x150 ; /* 0x0000011000007945 */ /* 0x000fe20003800000 */ /*0040*/ UIADD3 UR4, -UR4, 0x1800, URZ ; /* 0x0000180004047890 */ /* 0x000fe2000fffe13f */ /*0050*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e240000002100 */ /*0060*/ IMAD R0, R0, c[0x0][0x0], R3 ; /* 0x0000000000007a24 */ /* 0x001fca00078e0203 */ /*0070*/ ISETP.GE.U32.AND P1, PT, R0.reuse, UR4, PT ; /* 0x0000000400007c0c */ /* 0x040fe4000bf26070 */ /*0080*/ ISETP.NE.AND P0, PT, R0, RZ, PT ; /* 0x000000ff0000720c */ /* 0x000fd60003f05270 */ /*0090*/ @P1 BRA 0x140 ; /* 0x000000a000001947 */ /* 0x000fea0003800000 */ /*00a0*/ IMAD.MOV.U32 R5, RZ, RZ, c[0x0][0x0] ; /* 0x00000000ff057624 */ /* 0x000fe400078e00ff */ /*00b0*/ IADD3 R4, R0, c[0x0][0x178], RZ ; /* 0x00005e0000047a10 */ /* 0x000fca0007ffe0ff */ /*00c0*/ IMAD.WIDE.U32 R2, R4, -0x55555555, RZ ; /* 0xaaaaaaab04027825 */ /* 0x000fca00078e00ff */ /*00d0*/ SHF.R.U32.HI R3, RZ, 0xc, R3 ; /* 0x0000000cff037819 */ /* 0x000fca0000011603 */ /*00e0*/ IMAD R2, R3, -0x1800, R4 ; /* 0xffffe80003027824 */ /* 0x000fe200078e0204 */ /*00f0*/ HFMA2.MMA R3, -RZ, RZ, 0, 0 ; /* 0x00000000ff037435 */ /* 0x000fce00000001ff */ /*0100*/ STS.64 [R0.X8], R2 ; /* 0x0000000200007388 */ /* 0x0001e40000008a00 */ /*0110*/ IMAD R0, R5, c[0x0][0xc], R0 ; /* 0x0000030005007a24 */ /* 0x001fca00078e0200 */ /*0120*/ ISETP.GE.U32.AND P1, PT, R0, UR4, PT ; /* 0x0000000400007c0c */ /* 0x000fda000bf26070 */ /*0130*/ @!P1 BRA 0xb0 ; /* 0xffffff7000009947 */ /* 0x000fea000383ffff */ /*0140*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*0150*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0160*/ CS2R.32 R0, SR_CLOCKLO ; /* 0x0000000000007805 */ /* 0x000fe40000005000 */ /*0170*/ IMAD.MOV.U32 R2, RZ, RZ, RZ ; /* 0x000000ffff027224 */ /* 0x000fe200078e00ff */ /*0180*/ CS2R R4, SRZ ; /* 0x0000000000047805 */ /* 0x000fe2000001ff00 */ /*0190*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fca0000000a00 */ /*01a0*/ IMAD.SHL.U32 R4, R4, 0x8, RZ ; /* 0x0000000804047824 */ /* 0x001fe200078e00ff */ /*01b0*/ IADD3 R2, R2, 0x40, RZ ; /* 0x0000004002027810 */ /* 0x000fc80007ffe0ff */ /*01c0*/ ISETP.NE.AND P0, PT, R2, 0x800, PT ; /* 0x000008000200780c */ /* 0x000fe20003f05270 */ /*01d0*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*01e0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*01f0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0200*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0210*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0220*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0230*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0240*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0250*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0260*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0270*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0280*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0290*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*02a0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*02b0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*02c0*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*02d0*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*02e0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*02f0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0300*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0310*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0320*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0330*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0340*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0350*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*0360*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*0370*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0380*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0390*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*03a0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*03b0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*03c0*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*03d0*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*03e0*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*03f0*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0400*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0410*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*0420*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*0430*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0440*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0450*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0460*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0470*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0480*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0490*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*04a0*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*04b0*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*04c0*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*04d0*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*04e0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*04f0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0500*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0510*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0520*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0530*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0540*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0550*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0560*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0570*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0580*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0590*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*05a0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*05b0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*05c0*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*05d0*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*05e0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*05f0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0600*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0610*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0620*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0630*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0640*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0650*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*0660*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*0670*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0680*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0690*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*06a0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*06b0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*06c0*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*06d0*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*06e0*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*06f0*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0700*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0710*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*0720*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*0730*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0740*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0750*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0760*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0770*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0780*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0790*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*07a0*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*07b0*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*07c0*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*07d0*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*07e0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*07f0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0800*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*0810*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*0820*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*0830*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0840*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0850*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0860*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0870*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0880*/ IMAD.SHL.U32 R4, R8, 0x8, RZ ; /* 0x0000000808047824 */ /* 0x001fcc00078e00ff */ /*0890*/ LDS R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e240000000800 */ /*08a0*/ SHF.L.U32 R3, R4, 0x3, RZ ; /* 0x0000000304037819 */ /* 0x001fcc00000006ff */ /*08b0*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*08c0*/ IMAD.SHL.U32 R5, R3, 0x8, RZ ; /* 0x0000000803057824 */ /* 0x001fcc00078e00ff */ /*08d0*/ LDS R5, [R5] ; /* 0x0000000005057984 */ /* 0x000e240000000800 */ /*08e0*/ SHF.L.U32 R6, R5, 0x3, RZ ; /* 0x0000000305067819 */ /* 0x001fcc00000006ff */ /*08f0*/ LDS R6, [R6] ; /* 0x0000000006067984 */ /* 0x000e240000000800 */ /*0900*/ IMAD.SHL.U32 R7, R6, 0x8, RZ ; /* 0x0000000806077824 */ /* 0x001fcc00078e00ff */ /*0910*/ LDS R7, [R7] ; /* 0x0000000007077984 */ /* 0x000e240000000800 */ /*0920*/ SHF.L.U32 R8, R7, 0x3, RZ ; /* 0x0000000307087819 */ /* 0x001fcc00000006ff */ /*0930*/ LDS R8, [R8] ; /* 0x0000000008087984 */ /* 0x000e240000000800 */ /*0940*/ IMAD.SHL.U32 R9, R8, 0x8, RZ ; /* 0x0000000808097824 */ /* 0x001fcc00078e00ff */ /*0950*/ LDS R9, [R9] ; /* 0x0000000009097984 */ /* 0x000e240000000800 */ /*0960*/ SHF.L.U32 R3, R9, 0x3, RZ ; /* 0x0000000309037819 */ /* 0x001fcc00000006ff */ /*0970*/ LDS R3, [R3] ; /* 0x0000000003037984 */ /* 0x000e240000000800 */ /*0980*/ IMAD.SHL.U32 R10, R3, 0x8, RZ ; /* 0x00000008030a7824 */ /* 0x001fcc00078e00ff */ /*0990*/ LDS R10, [R10] ; /* 0x000000000a0a7984 */ /* 0x000e240000000800 */ /*09a0*/ SHF.L.U32 R4, R10, 0x3, RZ ; /* 0x000000030a047819 */ /* 0x001fcc00000006ff */ /*09b0*/ LDS.64 R4, [R4] ; /* 0x0000000004047984 */ /* 0x000e220000000a00 */ /*09c0*/ @P0 BRA 0x1a0 ; /* 0xfffff7d000000947 */ /* 0x000fea000383ffff */ /*09d0*/ CS2R.32 R11, SR_CLOCKLO ; /* 0x00000000000b7805 */ /* 0x000fe40000005000 */ /*09e0*/ IMAD.MOV.U32 R2, RZ, RZ, c[0x0][0x160] ; /* 0x00005800ff027624 */ /* 0x000fe200078e00ff */ /*09f0*/ MOV R3, c[0x0][0x164] ; /* 0x0000590000037a02 */ /* 0x000fe20000000f00 */ /*0a00*/ IMAD.MOV.U32 R6, RZ, RZ, c[0x0][0x168] ; /* 0x00005a00ff067624 */ /* 0x000fe200078e00ff */ /*0a10*/ MOV R7, c[0x0][0x16c] ; /* 0x00005b0000077a02 */ /* 0x000fe20000000f00 */ /*0a20*/ IMAD.MOV.U32 R8, RZ, RZ, c[0x0][0x170] ; /* 0x00005c00ff087624 */ /* 0x000fe200078e00ff */ /*0a30*/ MOV R9, c[0x0][0x174] ; /* 0x00005d0000097a02 */ /* 0x000fe20000000f00 */ /*0a40*/ STG.E [R2.64], R0 ; /* 0x0000000002007986 */ /* 0x000fe8000c101904 */ /*0a50*/ STG.E [R6.64], R11 ; /* 0x0000000b06007986 */ /* 0x000fe8000c101904 */ /*0a60*/ STG.E.64 [R8.64], R4 ; /* 0x0000000408007986 */ /* 0x001fe2000c101b04 */ /*0a70*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0a80*/ BRA 0xa80; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*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 */ /*0b00*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b10*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b20*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b30*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b40*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b50*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b60*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0b70*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z10shared_latPjS_Pmj .globl _Z10shared_latPjS_Pmj .p2align 8 .type _Z10shared_latPjS_Pmj,@function _Z10shared_latPjS_Pmj: s_clause 0x1 s_load_b32 s5, s[0:1], 0x2c s_load_b32 s4, s[0:1], 0x18 s_add_u32 s2, s0, 32 s_addc_u32 s3, s1, 0 s_mov_b32 s6, exec_lo s_waitcnt lgkmcnt(0) s_and_b32 s7, s5, 0xffff s_sub_i32 s5, 0x1800, s4 s_mul_i32 s15, s15, s7 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_nc_u32_e32 v2, s15, v0 v_cmpx_gt_u32_e64 s5, v2 s_cbranch_execz .LBB0_3 s_load_b32 s2, s[2:3], 0x0 v_add3_u32 v3, s15, s4, v0 v_dual_mov_b32 v1, 0 :: v_dual_lshlrev_b32 v4, 3, v2 s_mov_b32 s3, 0 s_mov_b32 s8, 0 s_waitcnt lgkmcnt(0) s_mul_i32 s2, s2, s7 s_delay_alu instid0(SALU_CYCLE_1) s_lshl_b32 s7, s2, 3 .p2align 6 .LBB0_2: v_mul_hi_u32 v0, v3, 0xaaaaaaab v_add_nc_u32_e32 v3, s2, v3 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshrrev_b32_e32 v0, 12, v0 v_mul_u32_u24_e32 v0, 0x1800, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_sub_nc_u32_e32 v0, s4, v0 v_add3_u32 v0, v2, s8, v0 s_add_i32 s8, s8, s2 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_4) | instid1(SALU_CYCLE_1) v_add_nc_u32_e32 v5, s8, v2 ds_store_b64 v4, v[0:1] v_add_nc_u32_e32 v4, s7, v4 v_cmp_le_u32_e32 vcc_lo, s5, v5 s_or_b32 s3, vcc_lo, s3 s_and_not1_b32 exec_lo, exec_lo, s3 s_cbranch_execnz .LBB0_2 .LBB0_3: s_or_b32 exec_lo, exec_lo, s6 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s2, exec_lo v_cmpx_eq_u32_e32 0, v2 s_cbranch_execz .LBB0_7 mov.u32 s2, %clock s_movk_i32 s3, 0x800 v_mov_b32_e32 v0, 0 v_mov_b32_e32 v1, 0 .LBB0_5: s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_lshlrev_b32_e32 v0, 3, v0 s_add_i32 s3, s3, -1 s_cmp_eq_u32 s3, 0 ds_load_b64 v[0:1], v0 s_cbranch_scc0 .LBB0_5 s_clause 0x1 s_load_b128 s[4:7], s[0:1], 0x0 s_load_b64 s[0:1], s[0:1], 0x10 v_dual_mov_b32 v2, 0 :: v_dual_mov_b32 v3, s2 mov.u32 s3, %clock v_mov_b32_e32 v4, s3 s_waitcnt lgkmcnt(0) s_clause 0x2 global_store_b32 v2, v3, s[4:5] global_store_b32 v2, v4, s[6:7] global_store_b64 v2, v[0:1], s[0:1] .LBB0_7: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z10shared_latPjS_Pmj .amdhsa_group_segment_fixed_size 49152 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 288 .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 _Z10shared_latPjS_Pmj, .Lfunc_end0-_Z10shared_latPjS_Pmj .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 - .offset: 32 .size: 4 .value_kind: hidden_block_count_x - .offset: 36 .size: 4 .value_kind: hidden_block_count_y - .offset: 40 .size: 4 .value_kind: hidden_block_count_z - .offset: 44 .size: 2 .value_kind: hidden_group_size_x - .offset: 46 .size: 2 .value_kind: hidden_group_size_y - .offset: 48 .size: 2 .value_kind: hidden_group_size_z - .offset: 50 .size: 2 .value_kind: hidden_remainder_x - .offset: 52 .size: 2 .value_kind: hidden_remainder_y - .offset: 54 .size: 2 .value_kind: hidden_remainder_z - .offset: 72 .size: 8 .value_kind: hidden_global_offset_x - .offset: 80 .size: 8 .value_kind: hidden_global_offset_y - .offset: 88 .size: 8 .value_kind: hidden_global_offset_z - .offset: 96 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 49152 .kernarg_segment_align: 8 .kernarg_segment_size: 288 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z10shared_latPjS_Pmj .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z10shared_latPjS_Pmj.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_00137c55_00000000-6_shared_lat.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._Z9gpuAssert9cudaErrorPKcib.str1.1,"aMS",@progbits,1 .LC0: .string "GPUassert: %s %s %d\n" .section .text._Z9gpuAssert9cudaErrorPKcib,"axG",@progbits,_Z9gpuAssert9cudaErrorPKcib,comdat .weak _Z9gpuAssert9cudaErrorPKcib .type _Z9gpuAssert9cudaErrorPKcib, @function _Z9gpuAssert9cudaErrorPKcib: .LFB2057: .cfi_startproc endbr64 testl %edi, %edi jne .L9 ret .L9: 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 $8, %rsp .cfi_def_cfa_offset 48 movl %edi, %ebx movq %rsi, %r13 movl %edx, %r12d movl %ecx, %ebp call cudaGetErrorString@PLT movq %rax, %rcx movl %r12d, %r9d movq %r13, %r8 leaq .LC0(%rip), %rdx movl $2, %esi movq stderr(%rip), %rdi movl $0, %eax call __fprintf_chk@PLT testb %bpl, %bpl jne .L10 addq $8, %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 .L10: .cfi_restore_state movl %ebx, %edi call exit@PLT .cfi_endproc .LFE2057: .size _Z9gpuAssert9cudaErrorPKcib, .-_Z9gpuAssert9cudaErrorPKcib .text .globl _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj .type _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj, @function _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj: .LFB2083: .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 .L15 .L11: movq 136(%rsp), %rax subq %fs:40, %rax jne .L16 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L15: .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 _Z10shared_latPjS_Pmj(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L11 .L16: call __stack_chk_fail@PLT .cfi_endproc .LFE2083: .size _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj, .-_Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj .globl _Z10shared_latPjS_Pmj .type _Z10shared_latPjS_Pmj, @function _Z10shared_latPjS_Pmj: .LFB2084: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2084: .size _Z10shared_latPjS_Pmj, .-_Z10shared_latPjS_Pmj .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC1: .string "/home/ubuntu/Datasets/stackv2/train-structured/purdue-aalp/gpgpu-sim_simulations/master/benchmarks/src/cuda/GPU_Microbenchmark/shared_lat/shared_lat.cu" .align 8 .LC3: .string "Shared Memory Latency = %f cycles\n" .section .rodata.str1.1,"aMS",@progbits,1 .LC4: .string "Total Clk number = %u \n" .LC5: .string "start clk = %u \n" .LC6: .string "stop clk = %u \n" .text .globl main .type main, @function main: .LFB2058: .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 $72, %rsp .cfi_def_cfa_offset 112 movq %fs:40, %rax movq %rax, 56(%rsp) xorl %eax, %eax movl $4, %edi call malloc@PLT movq %rax, %rbp movl $4, %edi call malloc@PLT movq %rax, %rbx movl $8, %edi call malloc@PLT movq %rax, %r13 leaq 8(%rsp), %rdi movl $4, %esi call cudaMalloc@PLT movl %eax, %edi movl $1, %ecx movl $79, %edx leaq .LC1(%rip), %r12 movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib leaq 16(%rsp), %rdi movl $4, %esi call cudaMalloc@PLT movl %eax, %edi movl $1, %ecx movl $80, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib leaq 24(%rsp), %rdi movl $8, %esi call cudaMalloc@PLT movl %eax, %edi movl $1, %ecx movl $81, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl $32, 44(%rsp) movl $1, 48(%rsp) movl $1, 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 .L25 .L20: call cudaPeekAtLastError@PLT movl %eax, %edi movl $1, %ecx movl $84, %edx leaq .LC1(%rip), %r12 movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl $2, %ecx movl $4, %edx movq 8(%rsp), %rsi movq %rbp, %rdi call cudaMemcpy@PLT movl %eax, %edi movl $1, %ecx movl $86, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl $2, %ecx movl $4, %edx movq 16(%rsp), %rsi movq %rbx, %rdi call cudaMemcpy@PLT movl %eax, %edi movl $1, %ecx movl $87, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl $2, %ecx movl $8, %edx movq 24(%rsp), %rsi movq %r13, %rdi call cudaMemcpy@PLT movl %eax, %edi movl $1, %ecx movl $88, %edx movq %r12, %rsi call _Z9gpuAssert9cudaErrorPKcib movl (%rbx), %eax subl 0(%rbp), %eax pxor %xmm0, %xmm0 cvtsi2ssq %rax, %xmm0 mulss .LC2(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 leaq .LC3(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movl (%rbx), %edx subl 0(%rbp), %edx leaq .LC4(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl 0(%rbp), %edx leaq .LC5(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl (%rbx), %edx leaq .LC6(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq 56(%rsp), %rax subq %fs:40, %rax jne .L26 movl $0, %eax addq $72, %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 .L25: .cfi_restore_state movl $1, %ecx movq 24(%rsp), %rdx movq 16(%rsp), %rsi movq 8(%rsp), %rdi call _Z35__device_stub__Z10shared_latPjS_PmjPjS_Pmj jmp .L20 .L26: call __stack_chk_fail@PLT .cfi_endproc .LFE2058: .size main, .-main .section .rodata.str1.1 .LC7: .string "_Z10shared_latPjS_Pmj" .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 .LC7(%rip), %rdx movq %rdx, %rcx leaq _Z10shared_latPjS_Pmj(%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 .LC2: .long 973078528 .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 "shared_lat.hip" .globl _Z25__device_stub__shared_latPjS_Pmj # -- Begin function _Z25__device_stub__shared_latPjS_Pmj .p2align 4, 0x90 .type _Z25__device_stub__shared_latPjS_Pmj,@function _Z25__device_stub__shared_latPjS_Pmj: # @_Z25__device_stub__shared_latPjS_Pmj .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 $_Z10shared_latPjS_Pmj, %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 _Z25__device_stub__shared_latPjS_Pmj, .Lfunc_end0-_Z25__device_stub__shared_latPjS_Pmj .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function main .LCPI1_0: .long 0x3a000000 # float 4.8828125E-4 .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 %rbx .cfi_def_cfa_offset 40 subq $152, %rsp .cfi_def_cfa_offset 192 .cfi_offset %rbx, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl $4, %edi callq malloc movq %rax, %r14 movl $4, %edi callq malloc movq %rax, %rbx movl $8, %edi callq malloc movq %rax, %r15 leaq 32(%rsp), %rdi movl $4, %esi callq hipMalloc testl %eax, %eax jne .LBB1_1 # %bb.3: # %_Z9gpuAssert10hipError_tPKcib.exit leaq 24(%rsp), %rdi movl $4, %esi callq hipMalloc testl %eax, %eax jne .LBB1_4 # %bb.5: # %_Z9gpuAssert10hipError_tPKcib.exit12 leaq 16(%rsp), %rdi movl $8, %esi callq hipMalloc testl %eax, %eax jne .LBB1_6 # %bb.7: # %_Z9gpuAssert10hipError_tPKcib.exit14 movabsq $4294967297, %rdi # imm = 0x100000001 leaq 31(%rdi), %rdx movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_9 # %bb.8: movq 32(%rsp), %rax movq 24(%rsp), %rcx movq 16(%rsp), %rdx movq %rax, 104(%rsp) movq %rcx, 96(%rsp) movq %rdx, 88(%rsp) movl $1, 12(%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 12(%rsp), %rax movq %rax, 136(%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 $_Z10shared_latPjS_Pmj, %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_9: callq hipPeekAtLastError testl %eax, %eax jne .LBB1_10 # %bb.11: # %_Z9gpuAssert10hipError_tPKcib.exit16 movq 32(%rsp), %rsi movl $4, %edx movq %r14, %rdi movl $2, %ecx callq hipMemcpy testl %eax, %eax jne .LBB1_12 # %bb.13: # %_Z9gpuAssert10hipError_tPKcib.exit18 movq 24(%rsp), %rsi movl $4, %edx movq %rbx, %rdi movl $2, %ecx callq hipMemcpy testl %eax, %eax jne .LBB1_14 # %bb.15: # %_Z9gpuAssert10hipError_tPKcib.exit20 movq 16(%rsp), %rsi movl $8, %edx movq %r15, %rdi movl $2, %ecx callq hipMemcpy testl %eax, %eax jne .LBB1_16 # %bb.17: # %_Z9gpuAssert10hipError_tPKcib.exit22 movl (%rbx), %eax subl (%r14), %eax cvtsi2ss %rax, %xmm0 mulss .LCPI1_0(%rip), %xmm0 cvtss2sd %xmm0, %xmm0 movl $.L.str.1, %edi movb $1, %al callq printf movl (%rbx), %esi subl (%r14), %esi movl $.L.str.2, %edi xorl %eax, %eax callq printf movl (%r14), %esi movl $.L.str.3, %edi xorl %eax, %eax callq printf movl (%rbx), %esi movl $.L.str.4, %edi xorl %eax, %eax callq printf xorl %eax, %eax addq $152, %rsp .cfi_def_cfa_offset 40 popq %rbx .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 .LBB1_1: .cfi_def_cfa_offset 192 movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $79, %r8d jmp .LBB1_2 .LBB1_4: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $80, %r8d jmp .LBB1_2 .LBB1_6: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $81, %r8d jmp .LBB1_2 .LBB1_10: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $84, %r8d jmp .LBB1_2 .LBB1_12: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $86, %r8d jmp .LBB1_2 .LBB1_14: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $87, %r8d jmp .LBB1_2 .LBB1_16: movq stderr(%rip), %rbx movl %eax, %edi movl %eax, %ebp callq hipGetErrorString movl $.L.str.5, %esi movl $.L.str, %ecx movq %rbx, %rdi movq %rax, %rdx movl $88, %r8d .LBB1_2: xorl %eax, %eax callq fprintf movl %ebp, %edi callq exit .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 $_Z10shared_latPjS_Pmj, %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 _Z10shared_latPjS_Pmj,@object # @_Z10shared_latPjS_Pmj .section .rodata,"a",@progbits .globl _Z10shared_latPjS_Pmj .p2align 3, 0x0 _Z10shared_latPjS_Pmj: .quad _Z25__device_stub__shared_latPjS_Pmj .size _Z10shared_latPjS_Pmj, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "/home/ubuntu/Datasets/stackv2/train-structured-repos-hip/purdue-aalp/gpgpu-sim_simulations/master/benchmarks/src/cuda/GPU_Microbenchmark/shared_lat/shared_lat.hip" .size .L.str, 163 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz "Shared Memory Latency = %f cycles\n" .size .L.str.1, 36 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "Total Clk number = %u \n" .size .L.str.2, 24 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "start clk = %u \n" .size .L.str.3, 17 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "stop clk = %u \n" .size .L.str.4, 16 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "GPUassert: %s %s %d\n" .size .L.str.5, 21 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z10shared_latPjS_Pmj" .size .L__unnamed_1, 22 .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__shared_latPjS_Pmj .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z10shared_latPjS_Pmj .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 <stdlib.h> #include <iostream> // both M and N must be evenly divisible by SIZE, M must be evenly divisible by CHKSIZE #define SIZE 2 #define N 2 #define M 3 #define CHKSIZE 1 //naive kernel __global__ void EuclideanDistancesNaive(float *A, float *B, float *C, int n, int m) { int idx = threadIdx.x + blockDim.x * blockIdx.x; int idy = threadIdx.y + blockDim.y * blockIdx.y; float result = 0.0f; if ((idx < n) && (idy < m)) { for (int i = 0; i < SIZE; i++) { float temp = A[(idx * SIZE) + i] - B[(idy * SIZE) + i]; result += temp * temp; } C[(idx * m) + idy] = result; } } //optimized kernel __global__ void EuclideanDistancesFast(const float *A, const float *B, float *C, const int n, const int m) { // n, A, 4000 this kernel assumes A is column-major A(SIZE, n) // m, B, 20000 this kernel assumes B is row-major B(m, SIZE) // this kernel assumes C is column-major C(m,n) // this kernel assumes number of threads per threadblock == SIZE // CHKSIZE is the number of B vectors that will be compute per block __shared__ float my_sB[CHKSIZE * SIZE]; // enough shared storage for CHKSIZE vectors of B int bx = blockIdx.x; // one block per CHKSIZE rows of B (the larger input matrix) while ((bx * CHKSIZE) < m) { // not used, this while loop could be used to extend a block to multiple chunks int tx = threadIdx.x; for (int i = 0; i < CHKSIZE; i++) // load vectors of B into shared memory my_sB[(i * SIZE) + tx] = B[(((bx * CHKSIZE) + i) * SIZE) + tx]; __syncthreads(); while (tx < n) { //loop across all vectors in A float result[CHKSIZE]; for (int i = 0; i < CHKSIZE; i++) result[i] = 0.0f; for (int i = 0; i < SIZE; i++) { float Atemp = A[(n * i) + tx]; for (int j = 0; j < CHKSIZE; j++) { // compute all CHKSIZE B vectors with read of A float temp = Atemp - my_sB[i + (j * SIZE)]; result[j] += temp * temp; } } for (int i = 0; i < CHKSIZE; i++) // store CHKSIZE results C[((i + (bx * CHKSIZE)) * n) + tx] = result[i]; tx += blockDim.x; } // continue looping across vectors in A __syncthreads(); // necessary to prevent warps from racing ahead, if block looping is used bx += gridDim.x; } } float comp_euclid_sq(const float *rA, const float *rB, const int size) { float result = 0.0f; float temp; for (int i = 0; i < size; i++) { temp = (rA[i] - rB[i]); result += temp * temp; } return result; } int main_ed_gpu() { float cpu_time = 0.0f, et1 = 0.0f, et2 = 0.0f, et_mem = 0.0f; cudaEvent_t start1, start2, stop1, stop2, start_mem_copy, stop_mem_copy; cudaEventCreate(&start1); cudaEventCreate(&start2); cudaEventCreate(&start_mem_copy); cudaEventCreate(&stop1); cudaEventCreate(&stop2); cudaEventCreate(&stop_mem_copy); int n = N; //MatrixA size : n * SIZE int m = M; //MatrixB size : m * SIZE srand((unsigned) time(0)); // Host Allocations float *matrixA = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n * SIZE; i++) matrixA[i] = (float) i + 1; float *matrixB = (float *) malloc(m * SIZE * sizeof(float)); for (int i = 0; i < m * SIZE; i++) matrixB[i] = (float) i + i; const clock_t begin_time = clock(); float *results_kernel = (float *) malloc(n * m * sizeof(float)); float *cpu_results_kernel = (float *) malloc(n * m * sizeof(float)); for (int i = 0; i < n * m; i++) cpu_results_kernel[i] = comp_euclid_sq(matrixA + ((i / m) * SIZE), matrixB + (i % m) * SIZE, SIZE); cpu_time = float( clock () - begin_time ) / 1000; //Device Allocation cudaEventRecord(start_mem_copy); float *d_matrixA; float *d_matrixB; cudaMalloc((void **) &d_matrixA, n * SIZE * sizeof(float)); cudaMalloc((void **) &d_matrixB, m * SIZE * sizeof(float)); cudaMemcpy(d_matrixA, matrixA, n * SIZE * sizeof(float), cudaMemcpyHostToDevice); cudaMemcpy(d_matrixB, matrixB, m * SIZE * sizeof(float), cudaMemcpyHostToDevice); cudaEventElapsedTime(&et_mem, start_mem_copy, stop_mem_copy); float *d_results_kernel; cudaMalloc((void **) &d_results_kernel, n * m * sizeof(float)); cudaFuncSetCacheConfig(EuclideanDistancesNaive, cudaFuncCachePreferL1); dim3 threads3(8, 32); // 1024 threads per block (maximum) dim3 blocks3(n / threads3.x, m / threads3.y); // assumes evenly divisible cudaEventRecord(start1); EuclideanDistancesNaive<<<blocks3, threads3>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); cudaEventRecord(stop1); cudaMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), cudaMemcpyDeviceToHost); // for (int i = 0; i < n * m; i++) { // if (results_kernel[i] != cpu_results_kernel[i]) { // printf("cpu/kernel3 mismatch at %d, cpu: %f, kernel3: %f\n", i, // cpu_results_kernel[i], results_kernel[i]); // return 1; // } // } cudaMemset(d_results_kernel, 0, n * m * sizeof(float)); cudaEventSynchronize(stop1); cudaEventElapsedTime(&et1, start1, stop1); // transpose matrix A float *matrixA_T = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n; i++) for (int j = 0; j < SIZE; j++) matrixA_T[(j * n) + i] = matrixA[(i * SIZE) + j]; cudaMemcpy(d_matrixA, matrixA_T, n * SIZE * sizeof(float), cudaMemcpyHostToDevice); cudaFuncSetCacheConfig(EuclideanDistancesFast, cudaFuncCachePreferL1); dim3 threads4(SIZE); // one thread per vector element dim3 blocks4(m / CHKSIZE); cudaEventRecord(start2); EuclideanDistancesFast<<<blocks4, threads4>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); cudaEventRecord(stop2); cudaMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), cudaMemcpyDeviceToHost); // test for correct transposed result C(m,n) for (int i = 0; i < n; i++) { for (int j = 0; j < m; j++) { printf("%f ", results_kernel[(j * n) + i]); if (results_kernel[(j * n) + i] != cpu_results_kernel[(i * m) + j]) { printf("cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n", i, j, cpu_results_kernel[(i * m) + j], results_kernel[(j * n) + i]); return 1; } } printf("\n"); } cudaEventSynchronize(stop2); cudaEventElapsedTime(&et2, start2, stop2); cudaFree(d_results_kernel); printf("Success!\n"); printf("CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n", cpu_time, et1, et2, et_mem); free(matrixA); free(matrixB); free(results_kernel); return 0; }
code for sm_80 Function : _Z22EuclideanDistancesFastPKfS0_Pfii .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.X ; /* 0x0000000000007919 */ /* 0x000e240000002500 */ /*0020*/ ISETP.GE.AND P0, PT, R0, c[0x0][0x17c], PT ; /* 0x00005f0000007a0c */ /* 0x001fda0003f06270 */ /*0030*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0040*/ S2R R15, SR_TID.X ; /* 0x00000000000f7919 */ /* 0x000e220000002100 */ /*0050*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fc60000000a00 */ /*0060*/ HFMA2.MMA R17, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff117435 */ /* 0x000fe200000001ff */ /*0070*/ LEA R2, R0, R15, 0x1 ; /* 0x0000000f00027211 */ /* 0x001fd200078e08ff */ /*0080*/ IMAD.WIDE R2, R2, R17, c[0x0][0x168] ; /* 0x00005a0002027625 */ /* 0x000fcc00078e0211 */ /*0090*/ LDG.E R2, [R2.64] ; /* 0x0000000402027981 */ /* 0x000ea2000c1e1900 */ /*00a0*/ ISETP.GE.AND P0, PT, R15, c[0x0][0x178], PT ; /* 0x00005e000f007a0c */ /* 0x000fe20003f06270 */ /*00b0*/ BSSY B0, 0x200 ; /* 0x0000014000007945 */ /* 0x000fe40003800000 */ /*00c0*/ STS [R15.X4], R2 ; /* 0x000000020f007388 */ /* 0x0041e80000004800 */ /*00d0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fec0000010000 */ /*00e0*/ @P0 BRA 0x1f0 ; /* 0x0000010000000947 */ /* 0x000fea0003800000 */ /*00f0*/ LDS.64 R2, [RZ] ; /* 0x00000000ff027984 */ /* 0x001e220000000a00 */ /*0100*/ MOV R11, R15 ; /* 0x0000000f000b7202 */ /* 0x000fca0000000f00 */ /*0110*/ IMAD.WIDE R4, R11, R17, c[0x0][0x160] ; /* 0x000058000b047625 */ /* 0x000fca00078e0211 */ /*0120*/ LDG.E R9, [R4.64] ; /* 0x0000000404097981 */ /* 0x001ea2000c1e1900 */ /*0130*/ IMAD.WIDE R6, R17, c[0x0][0x178], R4 ; /* 0x00005e0011067a25 */ /* 0x000fcc00078e0204 */ /*0140*/ LDG.E R6, [R6.64] ; /* 0x0000000406067981 */ /* 0x000ee2000c1e1900 */ /*0150*/ IMAD R8, R0, c[0x0][0x178], R11 ; /* 0x00005e0000087a24 */ /* 0x000fe200078e020b */ /*0160*/ IADD3 R11, R11, c[0x0][0x0], RZ ; /* 0x000000000b0b7a10 */ /* 0x000fc80007ffe0ff */ /*0170*/ ISETP.GE.AND P0, PT, R11, c[0x0][0x178], PT ; /* 0x00005e000b007a0c */ /* 0x000fe20003f06270 */ /*0180*/ FADD R9, -R2, R9 ; /* 0x0000000902097221 */ /* 0x005fc80000000100 */ /*0190*/ FFMA R10, R9, R9, RZ ; /* 0x00000009090a7223 */ /* 0x000fe400000000ff */ /*01a0*/ IMAD.WIDE R8, R8, R17, c[0x0][0x170] ; /* 0x00005c0008087625 */ /* 0x000fc800078e0211 */ /*01b0*/ FADD R13, R6, -R3 ; /* 0x80000003060d7221 */ /* 0x008fc80000000000 */ /*01c0*/ FFMA R13, R13, R13, R10 ; /* 0x0000000d0d0d7223 */ /* 0x000fca000000000a */ /*01d0*/ STG.E [R8.64], R13 ; /* 0x0000000d08007986 */ /* 0x0001e2000c101904 */ /*01e0*/ @!P0 BRA 0x110 ; /* 0xffffff2000008947 */ /* 0x000fea000383ffff */ /*01f0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x001fea0003800000 */ /*0200*/ IADD3 R0, R0, c[0x0][0xc], RZ ; /* 0x0000030000007a10 */ /* 0x000fe20007ffe0ff */ /*0210*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe60000010000 */ /*0220*/ ISETP.GE.AND P0, PT, R0, c[0x0][0x17c], PT ; /* 0x00005f0000007a0c */ /* 0x000fda0003f06270 */ /*0230*/ @!P0 BRA 0x60 ; /* 0xfffffe2000008947 */ /* 0x000fea000383ffff */ /*0240*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0250*/ BRA 0x250; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0260*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0270*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0280*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0290*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ .......... Function : _Z23EuclideanDistancesNaivePfS_S_ii .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 R7, SR_CTAID.Y ; /* 0x0000000000077919 */ /* 0x000e280000002600 */ /*0020*/ S2R R2, SR_TID.Y ; /* 0x0000000000027919 */ /* 0x000e280000002200 */ /*0030*/ S2R R0, SR_CTAID.X ; /* 0x0000000000007919 */ /* 0x000e680000002500 */ /*0040*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e620000002100 */ /*0050*/ IMAD R7, R7, c[0x0][0x4], R2 ; /* 0x0000010007077a24 */ /* 0x001fca00078e0202 */ /*0060*/ ISETP.GE.AND P0, PT, R7, c[0x0][0x17c], PT ; /* 0x00005f0007007a0c */ /* 0x000fe20003f06270 */ /*0070*/ IMAD R0, R0, c[0x0][0x0], R3 ; /* 0x0000000000007a24 */ /* 0x002fca00078e0203 */ /*0080*/ ISETP.GE.OR P0, PT, R0, c[0x0][0x178], P0 ; /* 0x00005e0000007a0c */ /* 0x000fda0000706670 */ /*0090*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*00a0*/ HFMA2.MMA R13, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff0d7435 */ /* 0x000fe200000001ff */ /*00b0*/ SHF.L.U32 R4, R7, 0x1, RZ ; /* 0x0000000107047819 */ /* 0x000fe200000006ff */ /*00c0*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe20000000a00 */ /*00d0*/ SHF.L.U32 R2, R0, 0x1, RZ ; /* 0x0000000100027819 */ /* 0x000fce00000006ff */ /*00e0*/ IMAD.WIDE R4, R4, R13, c[0x0][0x168] ; /* 0x00005a0004047625 */ /* 0x000fc800078e020d */ /*00f0*/ IMAD.WIDE R2, R2, R13, c[0x0][0x160] ; /* 0x0000580002027625 */ /* 0x000fe200078e020d */ /*0100*/ LDG.E R6, [R4.64] ; /* 0x0000000404067981 */ /* 0x000ea8000c1e1900 */ /*0110*/ LDG.E R9, [R2.64] ; /* 0x0000000402097981 */ /* 0x000ea8000c1e1900 */ /*0120*/ LDG.E R8, [R4.64+0x4] ; /* 0x0000040404087981 */ /* 0x000ee8000c1e1900 */ /*0130*/ LDG.E R11, [R2.64+0x4] ; /* 0x00000404020b7981 */ /* 0x000ee2000c1e1900 */ /*0140*/ IMAD R7, R0, c[0x0][0x17c], R7 ; /* 0x00005f0000077a24 */ /* 0x000fc400078e0207 */ /*0150*/ FADD R6, -R6, R9 ; /* 0x0000000906067221 */ /* 0x004fc80000000100 */ /*0160*/ FFMA R0, R6, R6, RZ ; /* 0x0000000606007223 */ /* 0x000fe400000000ff */ /*0170*/ IMAD.WIDE R6, R7, R13, c[0x0][0x170] ; /* 0x00005c0007067625 */ /* 0x000fc800078e020d */ /*0180*/ FADD R11, -R8, R11 ; /* 0x0000000b080b7221 */ /* 0x008fc80000000100 */ /*0190*/ FFMA R11, R11, R11, R0 ; /* 0x0000000b0b0b7223 */ /* 0x000fca0000000000 */ /*01a0*/ STG.E [R6.64], R11 ; /* 0x0000000b06007986 */ /* 0x000fe2000c101904 */ /*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.
#include <stdio.h> #include <stdlib.h> #include <iostream> // both M and N must be evenly divisible by SIZE, M must be evenly divisible by CHKSIZE #define SIZE 2 #define N 2 #define M 3 #define CHKSIZE 1 //naive kernel __global__ void EuclideanDistancesNaive(float *A, float *B, float *C, int n, int m) { int idx = threadIdx.x + blockDim.x * blockIdx.x; int idy = threadIdx.y + blockDim.y * blockIdx.y; float result = 0.0f; if ((idx < n) && (idy < m)) { for (int i = 0; i < SIZE; i++) { float temp = A[(idx * SIZE) + i] - B[(idy * SIZE) + i]; result += temp * temp; } C[(idx * m) + idy] = result; } } //optimized kernel __global__ void EuclideanDistancesFast(const float *A, const float *B, float *C, const int n, const int m) { // n, A, 4000 this kernel assumes A is column-major A(SIZE, n) // m, B, 20000 this kernel assumes B is row-major B(m, SIZE) // this kernel assumes C is column-major C(m,n) // this kernel assumes number of threads per threadblock == SIZE // CHKSIZE is the number of B vectors that will be compute per block __shared__ float my_sB[CHKSIZE * SIZE]; // enough shared storage for CHKSIZE vectors of B int bx = blockIdx.x; // one block per CHKSIZE rows of B (the larger input matrix) while ((bx * CHKSIZE) < m) { // not used, this while loop could be used to extend a block to multiple chunks int tx = threadIdx.x; for (int i = 0; i < CHKSIZE; i++) // load vectors of B into shared memory my_sB[(i * SIZE) + tx] = B[(((bx * CHKSIZE) + i) * SIZE) + tx]; __syncthreads(); while (tx < n) { //loop across all vectors in A float result[CHKSIZE]; for (int i = 0; i < CHKSIZE; i++) result[i] = 0.0f; for (int i = 0; i < SIZE; i++) { float Atemp = A[(n * i) + tx]; for (int j = 0; j < CHKSIZE; j++) { // compute all CHKSIZE B vectors with read of A float temp = Atemp - my_sB[i + (j * SIZE)]; result[j] += temp * temp; } } for (int i = 0; i < CHKSIZE; i++) // store CHKSIZE results C[((i + (bx * CHKSIZE)) * n) + tx] = result[i]; tx += blockDim.x; } // continue looping across vectors in A __syncthreads(); // necessary to prevent warps from racing ahead, if block looping is used bx += gridDim.x; } } float comp_euclid_sq(const float *rA, const float *rB, const int size) { float result = 0.0f; float temp; for (int i = 0; i < size; i++) { temp = (rA[i] - rB[i]); result += temp * temp; } return result; } int main_ed_gpu() { float cpu_time = 0.0f, et1 = 0.0f, et2 = 0.0f, et_mem = 0.0f; cudaEvent_t start1, start2, stop1, stop2, start_mem_copy, stop_mem_copy; cudaEventCreate(&start1); cudaEventCreate(&start2); cudaEventCreate(&start_mem_copy); cudaEventCreate(&stop1); cudaEventCreate(&stop2); cudaEventCreate(&stop_mem_copy); int n = N; //MatrixA size : n * SIZE int m = M; //MatrixB size : m * SIZE srand((unsigned) time(0)); // Host Allocations float *matrixA = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n * SIZE; i++) matrixA[i] = (float) i + 1; float *matrixB = (float *) malloc(m * SIZE * sizeof(float)); for (int i = 0; i < m * SIZE; i++) matrixB[i] = (float) i + i; const clock_t begin_time = clock(); float *results_kernel = (float *) malloc(n * m * sizeof(float)); float *cpu_results_kernel = (float *) malloc(n * m * sizeof(float)); for (int i = 0; i < n * m; i++) cpu_results_kernel[i] = comp_euclid_sq(matrixA + ((i / m) * SIZE), matrixB + (i % m) * SIZE, SIZE); cpu_time = float( clock () - begin_time ) / 1000; //Device Allocation cudaEventRecord(start_mem_copy); float *d_matrixA; float *d_matrixB; cudaMalloc((void **) &d_matrixA, n * SIZE * sizeof(float)); cudaMalloc((void **) &d_matrixB, m * SIZE * sizeof(float)); cudaMemcpy(d_matrixA, matrixA, n * SIZE * sizeof(float), cudaMemcpyHostToDevice); cudaMemcpy(d_matrixB, matrixB, m * SIZE * sizeof(float), cudaMemcpyHostToDevice); cudaEventElapsedTime(&et_mem, start_mem_copy, stop_mem_copy); float *d_results_kernel; cudaMalloc((void **) &d_results_kernel, n * m * sizeof(float)); cudaFuncSetCacheConfig(EuclideanDistancesNaive, cudaFuncCachePreferL1); dim3 threads3(8, 32); // 1024 threads per block (maximum) dim3 blocks3(n / threads3.x, m / threads3.y); // assumes evenly divisible cudaEventRecord(start1); EuclideanDistancesNaive<<<blocks3, threads3>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); cudaEventRecord(stop1); cudaMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), cudaMemcpyDeviceToHost); // for (int i = 0; i < n * m; i++) { // if (results_kernel[i] != cpu_results_kernel[i]) { // printf("cpu/kernel3 mismatch at %d, cpu: %f, kernel3: %f\n", i, // cpu_results_kernel[i], results_kernel[i]); // return 1; // } // } cudaMemset(d_results_kernel, 0, n * m * sizeof(float)); cudaEventSynchronize(stop1); cudaEventElapsedTime(&et1, start1, stop1); // transpose matrix A float *matrixA_T = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n; i++) for (int j = 0; j < SIZE; j++) matrixA_T[(j * n) + i] = matrixA[(i * SIZE) + j]; cudaMemcpy(d_matrixA, matrixA_T, n * SIZE * sizeof(float), cudaMemcpyHostToDevice); cudaFuncSetCacheConfig(EuclideanDistancesFast, cudaFuncCachePreferL1); dim3 threads4(SIZE); // one thread per vector element dim3 blocks4(m / CHKSIZE); cudaEventRecord(start2); EuclideanDistancesFast<<<blocks4, threads4>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); cudaEventRecord(stop2); cudaMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), cudaMemcpyDeviceToHost); // test for correct transposed result C(m,n) for (int i = 0; i < n; i++) { for (int j = 0; j < m; j++) { printf("%f ", results_kernel[(j * n) + i]); if (results_kernel[(j * n) + i] != cpu_results_kernel[(i * m) + j]) { printf("cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n", i, j, cpu_results_kernel[(i * m) + j], results_kernel[(j * n) + i]); return 1; } } printf("\n"); } cudaEventSynchronize(stop2); cudaEventElapsedTime(&et2, start2, stop2); cudaFree(d_results_kernel); printf("Success!\n"); printf("CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n", cpu_time, et1, et2, et_mem); free(matrixA); free(matrixB); free(results_kernel); return 0; }
.file "tmpxft_000df931_00000000-6_euclidean_distance_fast.cudafe1.cpp" .text #APP .globl _ZSt21ios_base_library_initv #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB3673: .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 .LFE3673: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z14comp_euclid_sqPKfS0_i .type _Z14comp_euclid_sqPKfS0_i, @function _Z14comp_euclid_sqPKfS0_i: .LFB3669: .cfi_startproc endbr64 testl %edx, %edx jle .L6 movslq %edx, %rdx salq $2, %rdx movl $0, %eax pxor %xmm1, %xmm1 .L5: movss (%rdi,%rax), %xmm0 subss (%rsi,%rax), %xmm0 mulss %xmm0, %xmm0 addss %xmm0, %xmm1 addq $4, %rax cmpq %rdx, %rax jne .L5 .L3: movaps %xmm1, %xmm0 ret .L6: pxor %xmm1, %xmm1 jmp .L3 .cfi_endproc .LFE3669: .size _Z14comp_euclid_sqPKfS0_i, .-_Z14comp_euclid_sqPKfS0_i .globl _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii .type _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii, @function _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii: .LFB3695: .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) movl %r8d, (%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) movq %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 .L12 .L8: movq 136(%rsp), %rax subq %fs:40, %rax jne .L13 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L12: .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 _Z23EuclideanDistancesNaivePfS_S_ii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L8 .L13: call __stack_chk_fail@PLT .cfi_endproc .LFE3695: .size _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii, .-_Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii .globl _Z23EuclideanDistancesNaivePfS_S_ii .type _Z23EuclideanDistancesNaivePfS_S_ii, @function _Z23EuclideanDistancesNaivePfS_S_ii: .LFB3696: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3696: .size _Z23EuclideanDistancesNaivePfS_S_ii, .-_Z23EuclideanDistancesNaivePfS_S_ii .globl _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii .type _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii, @function _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii: .LFB3697: .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) movl %r8d, (%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) movq %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 .L20 .L16: movq 136(%rsp), %rax subq %fs:40, %rax jne .L21 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L20: .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 _Z22EuclideanDistancesFastPKfS0_Pfii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L16 .L21: call __stack_chk_fail@PLT .cfi_endproc .LFE3697: .size _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii, .-_Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii .globl _Z22EuclideanDistancesFastPKfS0_Pfii .type _Z22EuclideanDistancesFastPKfS0_Pfii, @function _Z22EuclideanDistancesFastPKfS0_Pfii: .LFB3698: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3698: .size _Z22EuclideanDistancesFastPKfS0_Pfii, .-_Z22EuclideanDistancesFastPKfS0_Pfii .section .rodata.str1.1,"aMS",@progbits,1 .LC6: .string "%f " .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC7: .string "cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n" .section .rodata.str1.1 .LC8: .string "\n" .LC9: .string "Success!\n" .section .rodata.str1.8 .align 8 .LC10: .string "CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n" .text .globl _Z11main_ed_gpuv .type _Z11main_ed_gpuv, @function _Z11main_ed_gpuv: .LFB3670: .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 $168, %rsp .cfi_def_cfa_offset 224 movq %fs:40, %rax movq %rax, 152(%rsp) xorl %eax, %eax movl $0x00000000, 20(%rsp) movl $0x00000000, 24(%rsp) movl $0x00000000, 28(%rsp) leaq 32(%rsp), %rdi call cudaEventCreate@PLT leaq 40(%rsp), %rdi call cudaEventCreate@PLT leaq 64(%rsp), %rdi call cudaEventCreate@PLT leaq 48(%rsp), %rdi call cudaEventCreate@PLT leaq 56(%rsp), %rdi call cudaEventCreate@PLT leaq 72(%rsp), %rdi call cudaEventCreate@PLT movl $0, %edi call time@PLT movl %eax, %edi call srand@PLT movl $16, %edi call malloc@PLT movq %rax, %rbp movl $0x3f800000, (%rax) movl $0x40000000, 4(%rax) movl $0x40400000, 8(%rax) movl $0x40800000, 12(%rax) movl $24, %edi call malloc@PLT movq %rax, %r13 movl $0, %eax .L25: pxor %xmm0, %xmm0 cvtsi2ssl %eax, %xmm0 addss %xmm0, %xmm0 movss %xmm0, 0(%r13,%rax,4) addq $1, %rax cmpq $6, %rax jne .L25 call clock@PLT movq %rax, %r14 movl $24, %edi call malloc@PLT movq %rax, (%rsp) movl $24, %edi call malloc@PLT movq %rax, %r12 movl $0, %ebx .L26: movslq %ebx, %rax imulq $1431655766, %rax, %rax shrq $32, %rax movl %ebx, %edx sarl $31, %edx subl %edx, %eax leal (%rax,%rax), %ecx addl %ecx, %eax movl %ebx, %edx subl %eax, %edx leal (%rdx,%rdx), %eax cltq leaq 0(%r13,%rax,4), %rsi movslq %ecx, %rcx leaq 0(%rbp,%rcx,4), %rdi movl $2, %edx call _Z14comp_euclid_sqPKfS0_i movss %xmm0, (%r12,%rbx,4) addq $1, %rbx cmpq $6, %rbx jne .L26 call clock@PLT subq %r14, %rax pxor %xmm0, %xmm0 cvtsi2ssq %rax, %xmm0 divss .LC5(%rip), %xmm0 movss %xmm0, 12(%rsp) movl $0, %esi movq 64(%rsp), %rdi call cudaEventRecord@PLT leaq 80(%rsp), %rdi movl $16, %esi call cudaMalloc@PLT leaq 88(%rsp), %rdi movl $24, %esi call cudaMalloc@PLT movl $1, %ecx movl $16, %edx movq %rbp, %rsi movq 80(%rsp), %rdi call cudaMemcpy@PLT movl $1, %ecx movl $24, %edx movq %r13, %rsi movq 88(%rsp), %rdi call cudaMemcpy@PLT leaq 28(%rsp), %rdi movq 72(%rsp), %rdx movq 64(%rsp), %rsi call cudaEventElapsedTime@PLT leaq 96(%rsp), %rdi movl $24, %esi call cudaMalloc@PLT movl $2, %esi leaq _Z23EuclideanDistancesNaivePfS_S_ii(%rip), %rdi call cudaFuncSetCacheConfig@PLT movl $1, 112(%rsp) movl $0, 116(%rsp) movl $0, 120(%rsp) movl $1, 124(%rsp) movl $0, %esi movq 32(%rsp), %rdi call cudaEventRecord@PLT movl $8, 104(%rsp) movl $32, 108(%rsp) movl 112(%rsp), %ecx movl $0, %r9d movl $0, %r8d movq 104(%rsp), %rdx movq 116(%rsp), %rdi movl 124(%rsp), %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L41 .L27: movl $0, %esi movq 48(%rsp), %rdi call cudaEventRecord@PLT movl $2, %ecx movl $24, %edx movq 96(%rsp), %rsi movq (%rsp), %rdi call cudaMemcpy@PLT movl $24, %edx movl $0, %esi movq 96(%rsp), %rdi call cudaMemset@PLT movq 48(%rsp), %rdi call cudaEventSynchronize@PLT leaq 20(%rsp), %rdi movq 48(%rsp), %rdx movq 32(%rsp), %rsi call cudaEventElapsedTime@PLT movl $16, %edi call malloc@PLT movq %rax, %rsi movss 0(%rbp), %xmm0 movss %xmm0, (%rax) movss 4(%rbp), %xmm0 movss %xmm0, 8(%rax) movss 8(%rbp), %xmm0 movss %xmm0, 4(%rax) movss 12(%rbp), %xmm0 movss %xmm0, 12(%rax) movl $1, %ecx movl $16, %edx movq 80(%rsp), %rdi call cudaMemcpy@PLT movl $2, %esi leaq _Z22EuclideanDistancesFastPKfS0_Pfii(%rip), %rdi call cudaFuncSetCacheConfig@PLT movl $2, 128(%rsp) movl $1, 132(%rsp) movl $1, 136(%rsp) movl $3, 140(%rsp) movl $1, 144(%rsp) movl $1, 148(%rsp) movl $0, %esi movq 40(%rsp), %rdi call cudaEventRecord@PLT movl 136(%rsp), %ecx movl $0, %r9d movl $0, %r8d movq 128(%rsp), %rdx movq 140(%rsp), %rdi movl 148(%rsp), %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L42 .L28: movl $0, %esi movq 56(%rsp), %rdi call cudaEventRecord@PLT movl $2, %ecx movl $24, %edx movq 96(%rsp), %rsi movq (%rsp), %r14 movq %r14, %rdi call cudaMemcpy@PLT movl $0, %r15d .L29: movl $0, %ebx .L33: pxor %xmm0, %xmm0 cvtss2sd (%r14,%rbx,8), %xmm0 leaq .LC6(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movss (%r14,%rbx,8), %xmm1 movss (%r12,%rbx,4), %xmm0 ucomiss %xmm0, %xmm1 jp .L36 jne .L36 addq $1, %rbx cmpq $3, %rbx jne .L33 leaq .LC8(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addl $1, %r15d addq $4, %r14 addq $12, %r12 cmpl $2, %r15d jne .L29 movq 56(%rsp), %rdi call cudaEventSynchronize@PLT leaq 24(%rsp), %rdi movq 56(%rsp), %rdx movq 40(%rsp), %rsi call cudaEventElapsedTime@PLT movq 96(%rsp), %rdi call cudaFree@PLT leaq .LC9(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT pxor %xmm0, %xmm0 cvtss2sd 12(%rsp), %xmm0 pxor %xmm3, %xmm3 cvtss2sd 28(%rsp), %xmm3 pxor %xmm2, %xmm2 cvtss2sd 24(%rsp), %xmm2 pxor %xmm1, %xmm1 cvtss2sd 20(%rsp), %xmm1 leaq .LC10(%rip), %rsi movl $2, %edi movl $4, %eax call __printf_chk@PLT movq %rbp, %rdi call free@PLT movq %r13, %rdi call free@PLT movq (%rsp), %rdi call free@PLT movl $0, %eax .L24: movq 152(%rsp), %rdx subq %fs:40, %rdx jne .L43 addq $168, %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 .L41: .cfi_restore_state movl $3, %r8d movl $2, %ecx movq 96(%rsp), %rdx movq 88(%rsp), %rsi movq 80(%rsp), %rdi call _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii jmp .L27 .L42: movl $3, %r8d movl $2, %ecx movq 96(%rsp), %rdx movq 88(%rsp), %rsi movq 80(%rsp), %rdi call _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii jmp .L28 .L36: cvtss2sd %xmm0, %xmm0 cvtss2sd %xmm1, %xmm1 movl %ebx, %ecx movl %r15d, %edx leaq .LC7(%rip), %rsi movl $2, %edi movl $2, %eax call __printf_chk@PLT movl $1, %eax jmp .L24 .L43: call __stack_chk_fail@PLT .cfi_endproc .LFE3670: .size _Z11main_ed_gpuv, .-_Z11main_ed_gpuv .section .rodata.str1.8 .align 8 .LC11: .string "_Z22EuclideanDistancesFastPKfS0_Pfii" .align 8 .LC12: .string "_Z23EuclideanDistancesNaivePfS_S_ii" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB3700: .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 _Z22EuclideanDistancesFastPKfS0_Pfii(%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 _Z23EuclideanDistancesNaivePfS_S_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 .LFE3700: .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 .LC5: .long 1148846080 .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 <stdlib.h> #include <iostream> // both M and N must be evenly divisible by SIZE, M must be evenly divisible by CHKSIZE #define SIZE 2 #define N 2 #define M 3 #define CHKSIZE 1 //naive kernel __global__ void EuclideanDistancesNaive(float *A, float *B, float *C, int n, int m) { int idx = threadIdx.x + blockDim.x * blockIdx.x; int idy = threadIdx.y + blockDim.y * blockIdx.y; float result = 0.0f; if ((idx < n) && (idy < m)) { for (int i = 0; i < SIZE; i++) { float temp = A[(idx * SIZE) + i] - B[(idy * SIZE) + i]; result += temp * temp; } C[(idx * m) + idy] = result; } } //optimized kernel __global__ void EuclideanDistancesFast(const float *A, const float *B, float *C, const int n, const int m) { // n, A, 4000 this kernel assumes A is column-major A(SIZE, n) // m, B, 20000 this kernel assumes B is row-major B(m, SIZE) // this kernel assumes C is column-major C(m,n) // this kernel assumes number of threads per threadblock == SIZE // CHKSIZE is the number of B vectors that will be compute per block __shared__ float my_sB[CHKSIZE * SIZE]; // enough shared storage for CHKSIZE vectors of B int bx = blockIdx.x; // one block per CHKSIZE rows of B (the larger input matrix) while ((bx * CHKSIZE) < m) { // not used, this while loop could be used to extend a block to multiple chunks int tx = threadIdx.x; for (int i = 0; i < CHKSIZE; i++) // load vectors of B into shared memory my_sB[(i * SIZE) + tx] = B[(((bx * CHKSIZE) + i) * SIZE) + tx]; __syncthreads(); while (tx < n) { //loop across all vectors in A float result[CHKSIZE]; for (int i = 0; i < CHKSIZE; i++) result[i] = 0.0f; for (int i = 0; i < SIZE; i++) { float Atemp = A[(n * i) + tx]; for (int j = 0; j < CHKSIZE; j++) { // compute all CHKSIZE B vectors with read of A float temp = Atemp - my_sB[i + (j * SIZE)]; result[j] += temp * temp; } } for (int i = 0; i < CHKSIZE; i++) // store CHKSIZE results C[((i + (bx * CHKSIZE)) * n) + tx] = result[i]; tx += blockDim.x; } // continue looping across vectors in A __syncthreads(); // necessary to prevent warps from racing ahead, if block looping is used bx += gridDim.x; } } float comp_euclid_sq(const float *rA, const float *rB, const int size) { float result = 0.0f; float temp; for (int i = 0; i < size; i++) { temp = (rA[i] - rB[i]); result += temp * temp; } return result; } int main_ed_gpu() { float cpu_time = 0.0f, et1 = 0.0f, et2 = 0.0f, et_mem = 0.0f; cudaEvent_t start1, start2, stop1, stop2, start_mem_copy, stop_mem_copy; cudaEventCreate(&start1); cudaEventCreate(&start2); cudaEventCreate(&start_mem_copy); cudaEventCreate(&stop1); cudaEventCreate(&stop2); cudaEventCreate(&stop_mem_copy); int n = N; //MatrixA size : n * SIZE int m = M; //MatrixB size : m * SIZE srand((unsigned) time(0)); // Host Allocations float *matrixA = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n * SIZE; i++) matrixA[i] = (float) i + 1; float *matrixB = (float *) malloc(m * SIZE * sizeof(float)); for (int i = 0; i < m * SIZE; i++) matrixB[i] = (float) i + i; const clock_t begin_time = clock(); float *results_kernel = (float *) malloc(n * m * sizeof(float)); float *cpu_results_kernel = (float *) malloc(n * m * sizeof(float)); for (int i = 0; i < n * m; i++) cpu_results_kernel[i] = comp_euclid_sq(matrixA + ((i / m) * SIZE), matrixB + (i % m) * SIZE, SIZE); cpu_time = float( clock () - begin_time ) / 1000; //Device Allocation cudaEventRecord(start_mem_copy); float *d_matrixA; float *d_matrixB; cudaMalloc((void **) &d_matrixA, n * SIZE * sizeof(float)); cudaMalloc((void **) &d_matrixB, m * SIZE * sizeof(float)); cudaMemcpy(d_matrixA, matrixA, n * SIZE * sizeof(float), cudaMemcpyHostToDevice); cudaMemcpy(d_matrixB, matrixB, m * SIZE * sizeof(float), cudaMemcpyHostToDevice); cudaEventElapsedTime(&et_mem, start_mem_copy, stop_mem_copy); float *d_results_kernel; cudaMalloc((void **) &d_results_kernel, n * m * sizeof(float)); cudaFuncSetCacheConfig(EuclideanDistancesNaive, cudaFuncCachePreferL1); dim3 threads3(8, 32); // 1024 threads per block (maximum) dim3 blocks3(n / threads3.x, m / threads3.y); // assumes evenly divisible cudaEventRecord(start1); EuclideanDistancesNaive<<<blocks3, threads3>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); cudaEventRecord(stop1); cudaMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), cudaMemcpyDeviceToHost); // for (int i = 0; i < n * m; i++) { // if (results_kernel[i] != cpu_results_kernel[i]) { // printf("cpu/kernel3 mismatch at %d, cpu: %f, kernel3: %f\n", i, // cpu_results_kernel[i], results_kernel[i]); // return 1; // } // } cudaMemset(d_results_kernel, 0, n * m * sizeof(float)); cudaEventSynchronize(stop1); cudaEventElapsedTime(&et1, start1, stop1); // transpose matrix A float *matrixA_T = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n; i++) for (int j = 0; j < SIZE; j++) matrixA_T[(j * n) + i] = matrixA[(i * SIZE) + j]; cudaMemcpy(d_matrixA, matrixA_T, n * SIZE * sizeof(float), cudaMemcpyHostToDevice); cudaFuncSetCacheConfig(EuclideanDistancesFast, cudaFuncCachePreferL1); dim3 threads4(SIZE); // one thread per vector element dim3 blocks4(m / CHKSIZE); cudaEventRecord(start2); EuclideanDistancesFast<<<blocks4, threads4>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); cudaEventRecord(stop2); cudaMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), cudaMemcpyDeviceToHost); // test for correct transposed result C(m,n) for (int i = 0; i < n; i++) { for (int j = 0; j < m; j++) { printf("%f ", results_kernel[(j * n) + i]); if (results_kernel[(j * n) + i] != cpu_results_kernel[(i * m) + j]) { printf("cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n", i, j, cpu_results_kernel[(i * m) + j], results_kernel[(j * n) + i]); return 1; } } printf("\n"); } cudaEventSynchronize(stop2); cudaEventElapsedTime(&et2, start2, stop2); cudaFree(d_results_kernel); printf("Success!\n"); printf("CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n", cpu_time, et1, et2, et_mem); free(matrixA); free(matrixB); free(results_kernel); return 0; }
#include <hip/hip_runtime.h> #include <stdio.h> #include <stdlib.h> #include <iostream> // both M and N must be evenly divisible by SIZE, M must be evenly divisible by CHKSIZE #define SIZE 2 #define N 2 #define M 3 #define CHKSIZE 1 //naive kernel __global__ void EuclideanDistancesNaive(float *A, float *B, float *C, int n, int m) { int idx = threadIdx.x + blockDim.x * blockIdx.x; int idy = threadIdx.y + blockDim.y * blockIdx.y; float result = 0.0f; if ((idx < n) && (idy < m)) { for (int i = 0; i < SIZE; i++) { float temp = A[(idx * SIZE) + i] - B[(idy * SIZE) + i]; result += temp * temp; } C[(idx * m) + idy] = result; } } //optimized kernel __global__ void EuclideanDistancesFast(const float *A, const float *B, float *C, const int n, const int m) { // n, A, 4000 this kernel assumes A is column-major A(SIZE, n) // m, B, 20000 this kernel assumes B is row-major B(m, SIZE) // this kernel assumes C is column-major C(m,n) // this kernel assumes number of threads per threadblock == SIZE // CHKSIZE is the number of B vectors that will be compute per block __shared__ float my_sB[CHKSIZE * SIZE]; // enough shared storage for CHKSIZE vectors of B int bx = blockIdx.x; // one block per CHKSIZE rows of B (the larger input matrix) while ((bx * CHKSIZE) < m) { // not used, this while loop could be used to extend a block to multiple chunks int tx = threadIdx.x; for (int i = 0; i < CHKSIZE; i++) // load vectors of B into shared memory my_sB[(i * SIZE) + tx] = B[(((bx * CHKSIZE) + i) * SIZE) + tx]; __syncthreads(); while (tx < n) { //loop across all vectors in A float result[CHKSIZE]; for (int i = 0; i < CHKSIZE; i++) result[i] = 0.0f; for (int i = 0; i < SIZE; i++) { float Atemp = A[(n * i) + tx]; for (int j = 0; j < CHKSIZE; j++) { // compute all CHKSIZE B vectors with read of A float temp = Atemp - my_sB[i + (j * SIZE)]; result[j] += temp * temp; } } for (int i = 0; i < CHKSIZE; i++) // store CHKSIZE results C[((i + (bx * CHKSIZE)) * n) + tx] = result[i]; tx += blockDim.x; } // continue looping across vectors in A __syncthreads(); // necessary to prevent warps from racing ahead, if block looping is used bx += gridDim.x; } } float comp_euclid_sq(const float *rA, const float *rB, const int size) { float result = 0.0f; float temp; for (int i = 0; i < size; i++) { temp = (rA[i] - rB[i]); result += temp * temp; } return result; } int main_ed_gpu() { float cpu_time = 0.0f, et1 = 0.0f, et2 = 0.0f, et_mem = 0.0f; hipEvent_t start1, start2, stop1, stop2, start_mem_copy, stop_mem_copy; hipEventCreate(&start1); hipEventCreate(&start2); hipEventCreate(&start_mem_copy); hipEventCreate(&stop1); hipEventCreate(&stop2); hipEventCreate(&stop_mem_copy); int n = N; //MatrixA size : n * SIZE int m = M; //MatrixB size : m * SIZE srand((unsigned) time(0)); // Host Allocations float *matrixA = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n * SIZE; i++) matrixA[i] = (float) i + 1; float *matrixB = (float *) malloc(m * SIZE * sizeof(float)); for (int i = 0; i < m * SIZE; i++) matrixB[i] = (float) i + i; const clock_t begin_time = clock(); float *results_kernel = (float *) malloc(n * m * sizeof(float)); float *cpu_results_kernel = (float *) malloc(n * m * sizeof(float)); for (int i = 0; i < n * m; i++) cpu_results_kernel[i] = comp_euclid_sq(matrixA + ((i / m) * SIZE), matrixB + (i % m) * SIZE, SIZE); cpu_time = float( clock () - begin_time ) / 1000; //Device Allocation hipEventRecord(start_mem_copy); float *d_matrixA; float *d_matrixB; hipMalloc((void **) &d_matrixA, n * SIZE * sizeof(float)); hipMalloc((void **) &d_matrixB, m * SIZE * sizeof(float)); hipMemcpy(d_matrixA, matrixA, n * SIZE * sizeof(float), hipMemcpyHostToDevice); hipMemcpy(d_matrixB, matrixB, m * SIZE * sizeof(float), hipMemcpyHostToDevice); hipEventElapsedTime(&et_mem, start_mem_copy, stop_mem_copy); float *d_results_kernel; hipMalloc((void **) &d_results_kernel, n * m * sizeof(float)); hipFuncSetCacheConfig(reinterpret_cast<const void*>(EuclideanDistancesNaive), hipFuncCachePreferL1); dim3 threads3(8, 32); // 1024 threads per block (maximum) dim3 blocks3(n / threads3.x, m / threads3.y); // assumes evenly divisible hipEventRecord(start1); EuclideanDistancesNaive<<<blocks3, threads3>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); hipEventRecord(stop1); hipMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), hipMemcpyDeviceToHost); // for (int i = 0; i < n * m; i++) { // if (results_kernel[i] != cpu_results_kernel[i]) { // printf("cpu/kernel3 mismatch at %d, cpu: %f, kernel3: %f\n", i, // cpu_results_kernel[i], results_kernel[i]); // return 1; // } // } hipMemset(d_results_kernel, 0, n * m * sizeof(float)); hipEventSynchronize(stop1); hipEventElapsedTime(&et1, start1, stop1); // transpose matrix A float *matrixA_T = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n; i++) for (int j = 0; j < SIZE; j++) matrixA_T[(j * n) + i] = matrixA[(i * SIZE) + j]; hipMemcpy(d_matrixA, matrixA_T, n * SIZE * sizeof(float), hipMemcpyHostToDevice); hipFuncSetCacheConfig(reinterpret_cast<const void*>(EuclideanDistancesFast), hipFuncCachePreferL1); dim3 threads4(SIZE); // one thread per vector element dim3 blocks4(m / CHKSIZE); hipEventRecord(start2); EuclideanDistancesFast<<<blocks4, threads4>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); hipEventRecord(stop2); hipMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), hipMemcpyDeviceToHost); // test for correct transposed result C(m,n) for (int i = 0; i < n; i++) { for (int j = 0; j < m; j++) { printf("%f ", results_kernel[(j * n) + i]); if (results_kernel[(j * n) + i] != cpu_results_kernel[(i * m) + j]) { printf("cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n", i, j, cpu_results_kernel[(i * m) + j], results_kernel[(j * n) + i]); return 1; } } printf("\n"); } hipEventSynchronize(stop2); hipEventElapsedTime(&et2, start2, stop2); hipFree(d_results_kernel); printf("Success!\n"); printf("CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n", cpu_time, et1, et2, et_mem); free(matrixA); free(matrixB); free(results_kernel); 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> #include <stdlib.h> #include <iostream> // both M and N must be evenly divisible by SIZE, M must be evenly divisible by CHKSIZE #define SIZE 2 #define N 2 #define M 3 #define CHKSIZE 1 //naive kernel __global__ void EuclideanDistancesNaive(float *A, float *B, float *C, int n, int m) { int idx = threadIdx.x + blockDim.x * blockIdx.x; int idy = threadIdx.y + blockDim.y * blockIdx.y; float result = 0.0f; if ((idx < n) && (idy < m)) { for (int i = 0; i < SIZE; i++) { float temp = A[(idx * SIZE) + i] - B[(idy * SIZE) + i]; result += temp * temp; } C[(idx * m) + idy] = result; } } //optimized kernel __global__ void EuclideanDistancesFast(const float *A, const float *B, float *C, const int n, const int m) { // n, A, 4000 this kernel assumes A is column-major A(SIZE, n) // m, B, 20000 this kernel assumes B is row-major B(m, SIZE) // this kernel assumes C is column-major C(m,n) // this kernel assumes number of threads per threadblock == SIZE // CHKSIZE is the number of B vectors that will be compute per block __shared__ float my_sB[CHKSIZE * SIZE]; // enough shared storage for CHKSIZE vectors of B int bx = blockIdx.x; // one block per CHKSIZE rows of B (the larger input matrix) while ((bx * CHKSIZE) < m) { // not used, this while loop could be used to extend a block to multiple chunks int tx = threadIdx.x; for (int i = 0; i < CHKSIZE; i++) // load vectors of B into shared memory my_sB[(i * SIZE) + tx] = B[(((bx * CHKSIZE) + i) * SIZE) + tx]; __syncthreads(); while (tx < n) { //loop across all vectors in A float result[CHKSIZE]; for (int i = 0; i < CHKSIZE; i++) result[i] = 0.0f; for (int i = 0; i < SIZE; i++) { float Atemp = A[(n * i) + tx]; for (int j = 0; j < CHKSIZE; j++) { // compute all CHKSIZE B vectors with read of A float temp = Atemp - my_sB[i + (j * SIZE)]; result[j] += temp * temp; } } for (int i = 0; i < CHKSIZE; i++) // store CHKSIZE results C[((i + (bx * CHKSIZE)) * n) + tx] = result[i]; tx += blockDim.x; } // continue looping across vectors in A __syncthreads(); // necessary to prevent warps from racing ahead, if block looping is used bx += gridDim.x; } } float comp_euclid_sq(const float *rA, const float *rB, const int size) { float result = 0.0f; float temp; for (int i = 0; i < size; i++) { temp = (rA[i] - rB[i]); result += temp * temp; } return result; } int main_ed_gpu() { float cpu_time = 0.0f, et1 = 0.0f, et2 = 0.0f, et_mem = 0.0f; hipEvent_t start1, start2, stop1, stop2, start_mem_copy, stop_mem_copy; hipEventCreate(&start1); hipEventCreate(&start2); hipEventCreate(&start_mem_copy); hipEventCreate(&stop1); hipEventCreate(&stop2); hipEventCreate(&stop_mem_copy); int n = N; //MatrixA size : n * SIZE int m = M; //MatrixB size : m * SIZE srand((unsigned) time(0)); // Host Allocations float *matrixA = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n * SIZE; i++) matrixA[i] = (float) i + 1; float *matrixB = (float *) malloc(m * SIZE * sizeof(float)); for (int i = 0; i < m * SIZE; i++) matrixB[i] = (float) i + i; const clock_t begin_time = clock(); float *results_kernel = (float *) malloc(n * m * sizeof(float)); float *cpu_results_kernel = (float *) malloc(n * m * sizeof(float)); for (int i = 0; i < n * m; i++) cpu_results_kernel[i] = comp_euclid_sq(matrixA + ((i / m) * SIZE), matrixB + (i % m) * SIZE, SIZE); cpu_time = float( clock () - begin_time ) / 1000; //Device Allocation hipEventRecord(start_mem_copy); float *d_matrixA; float *d_matrixB; hipMalloc((void **) &d_matrixA, n * SIZE * sizeof(float)); hipMalloc((void **) &d_matrixB, m * SIZE * sizeof(float)); hipMemcpy(d_matrixA, matrixA, n * SIZE * sizeof(float), hipMemcpyHostToDevice); hipMemcpy(d_matrixB, matrixB, m * SIZE * sizeof(float), hipMemcpyHostToDevice); hipEventElapsedTime(&et_mem, start_mem_copy, stop_mem_copy); float *d_results_kernel; hipMalloc((void **) &d_results_kernel, n * m * sizeof(float)); hipFuncSetCacheConfig(reinterpret_cast<const void*>(EuclideanDistancesNaive), hipFuncCachePreferL1); dim3 threads3(8, 32); // 1024 threads per block (maximum) dim3 blocks3(n / threads3.x, m / threads3.y); // assumes evenly divisible hipEventRecord(start1); EuclideanDistancesNaive<<<blocks3, threads3>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); hipEventRecord(stop1); hipMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), hipMemcpyDeviceToHost); // for (int i = 0; i < n * m; i++) { // if (results_kernel[i] != cpu_results_kernel[i]) { // printf("cpu/kernel3 mismatch at %d, cpu: %f, kernel3: %f\n", i, // cpu_results_kernel[i], results_kernel[i]); // return 1; // } // } hipMemset(d_results_kernel, 0, n * m * sizeof(float)); hipEventSynchronize(stop1); hipEventElapsedTime(&et1, start1, stop1); // transpose matrix A float *matrixA_T = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n; i++) for (int j = 0; j < SIZE; j++) matrixA_T[(j * n) + i] = matrixA[(i * SIZE) + j]; hipMemcpy(d_matrixA, matrixA_T, n * SIZE * sizeof(float), hipMemcpyHostToDevice); hipFuncSetCacheConfig(reinterpret_cast<const void*>(EuclideanDistancesFast), hipFuncCachePreferL1); dim3 threads4(SIZE); // one thread per vector element dim3 blocks4(m / CHKSIZE); hipEventRecord(start2); EuclideanDistancesFast<<<blocks4, threads4>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); hipEventRecord(stop2); hipMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), hipMemcpyDeviceToHost); // test for correct transposed result C(m,n) for (int i = 0; i < n; i++) { for (int j = 0; j < m; j++) { printf("%f ", results_kernel[(j * n) + i]); if (results_kernel[(j * n) + i] != cpu_results_kernel[(i * m) + j]) { printf("cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n", i, j, cpu_results_kernel[(i * m) + j], results_kernel[(j * n) + i]); return 1; } } printf("\n"); } hipEventSynchronize(stop2); hipEventElapsedTime(&et2, start2, stop2); hipFree(d_results_kernel); printf("Success!\n"); printf("CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n", cpu_time, et1, et2, et_mem); free(matrixA); free(matrixB); free(results_kernel); return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z23EuclideanDistancesNaivePfS_S_ii .globl _Z23EuclideanDistancesNaivePfS_S_ii .p2align 8 .type _Z23EuclideanDistancesNaivePfS_S_ii,@function _Z23EuclideanDistancesNaivePfS_S_ii: s_clause 0x1 s_load_b32 s4, s[0:1], 0x2c s_load_b64 s[2:3], s[0:1], 0x18 v_and_b32_e32 v2, 0x3ff, v0 v_bfe_u32 v3, v0, 10, 10 s_waitcnt lgkmcnt(0) s_and_b32 s5, s4, 0xffff s_lshr_b32 s4, s4, 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, s4, v[3:4] v_cmp_gt_i32_e32 vcc_lo, s2, 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 s4, s2 s_cbranch_execz .LBB0_4 s_load_b128 s[4:7], s[0:1], 0x0 v_lshlrev_b32_e32 v2, 1, v0 v_lshlrev_b32_e32 v4, 1, v1 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_ashrrev_i32_e32 v5, 31, v4 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[6:7], 2, v[2:3] v_lshlrev_b64 v[8:9], 2, v[4:5] v_mov_b32_e32 v2, 0 s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_4) v_add_co_u32 v3, vcc_lo, s4, v6 v_add_co_ci_u32_e32 v4, vcc_lo, s5, v7, vcc_lo s_delay_alu instid0(VALU_DEP_4) v_add_co_u32 v5, vcc_lo, s6, v8 v_add_co_ci_u32_e32 v6, vcc_lo, s7, v9, vcc_lo s_mov_b64 s[4:5], 0 .LBB0_2: s_delay_alu instid0(SALU_CYCLE_1) v_add_co_u32 v7, vcc_lo, v3, s4 v_add_co_ci_u32_e32 v8, vcc_lo, s5, v4, vcc_lo v_add_co_u32 v9, vcc_lo, v5, s4 v_add_co_ci_u32_e32 v10, vcc_lo, s5, v6, vcc_lo s_add_u32 s4, s4, 4 global_load_b32 v7, v[7:8], off global_load_b32 v8, v[9:10], off s_addc_u32 s5, s5, 0 s_cmp_lg_u32 s4, 4 s_waitcnt vmcnt(0) v_sub_f32_e32 v7, v7, v8 s_delay_alu instid0(VALU_DEP_1) v_fmac_f32_e32 v2, v7, v7 s_cbranch_scc0 .LBB0_2 s_load_b64 s[0:1], s[0:1], 0x10 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u64_u32 v[3:4], null, v0, s3, v[1:2] v_ashrrev_i32_e32 v4, 31, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[0:1], 2, v[3:4] 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_4: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z23EuclideanDistancesNaivePfS_S_ii .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 288 .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 11 .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 _Z23EuclideanDistancesNaivePfS_S_ii, .Lfunc_end0-_Z23EuclideanDistancesNaivePfS_S_ii .section .AMDGPU.csdata,"",@progbits .text .protected _Z22EuclideanDistancesFastPKfS0_Pfii .globl _Z22EuclideanDistancesFastPKfS0_Pfii .p2align 8 .type _Z22EuclideanDistancesFastPKfS0_Pfii,@function _Z22EuclideanDistancesFastPKfS0_Pfii: s_load_b32 s14, s[0:1], 0x1c s_waitcnt lgkmcnt(0) s_cmp_ge_i32 s15, s14 s_cbranch_scc1 .LBB1_8 s_clause 0x3 s_load_b32 s2, s[0:1], 0x18 s_load_b128 s[4:7], s[0:1], 0x0 s_load_b64 s[8:9], s[0:1], 0x10 s_load_b32 s16, s[0:1], 0x20 v_lshlrev_b32_e32 v4, 2, v0 s_add_u32 s10, s0, 32 s_addc_u32 s11, s1, 0 s_waitcnt lgkmcnt(0) v_cmp_gt_i32_e32 vcc_lo, s2, v0 s_ashr_i32 s3, s2, 31 s_delay_alu instid0(SALU_CYCLE_1) s_lshl_b64 s[12:13], s[2:3], 2 s_branch .LBB1_3 .LBB1_2: s_set_inst_prefetch_distance 0x2 s_or_b32 exec_lo, exec_lo, s3 s_add_i32 s15, s16, s15 s_waitcnt_vscnt null, 0x0 s_cmp_lt_i32 s15, s14 s_barrier buffer_gl0_inv s_cbranch_scc0 .LBB1_8 .LBB1_3: v_lshl_add_u32 v1, s15, 1, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_ashrrev_i32_e32 v2, 31, v1 v_lshlrev_b64 v[1:2], 2, v[1:2] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_co_u32 v1, s0, s6, v1 v_add_co_ci_u32_e64 v2, s0, s7, v2, s0 global_load_b32 v1, v[1:2], off s_waitcnt vmcnt(0) ds_store_b32 v4, v1 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_and_saveexec_b32 s3, vcc_lo s_cbranch_execz .LBB1_2 s_load_b32 s0, s[10:11], 0xc v_mov_b32_e32 v1, v0 s_mul_i32 s17, s15, s2 s_mov_b32 s19, 0 s_waitcnt lgkmcnt(0) s_and_b32 s18, s0, 0xffff s_set_inst_prefetch_distance 0x1 .p2align 6 .LBB1_5: v_ashrrev_i32_e32 v2, 31, v1 v_mov_b32_e32 v5, 0 s_mov_b32 s1, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b64 v[2:3], 2, v[1:2] v_add_co_u32 v2, s0, s4, v2 s_delay_alu instid0(VALU_DEP_1) v_add_co_ci_u32_e64 v3, s0, s5, v3, s0 .LBB1_6: global_load_b32 v6, v[2:3], off v_mov_b32_e32 v7, s1 v_add_co_u32 v2, s0, v2, s12 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) v_add_co_ci_u32_e64 v3, s0, s13, v3, s0 ds_load_b32 v7, v7 s_add_i32 s1, s1, 4 s_cmp_eq_u32 s1, 4 s_waitcnt vmcnt(0) lgkmcnt(0) v_sub_f32_e32 v6, v6, v7 s_delay_alu instid0(VALU_DEP_1) v_fmac_f32_e32 v5, v6, v6 s_cbranch_scc1 .LBB1_6 v_add_nc_u32_e32 v2, s17, v1 v_add_nc_u32_e32 v1, s18, v1 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_cmp_le_i32_e64 s0, s2, v1 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[2:3], 2, v[2:3] s_or_b32 s19, s0, s19 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_co_u32 v2, s1, s8, v2 v_add_co_ci_u32_e64 v3, s1, s9, v3, s1 global_store_b32 v[2:3], v5, off s_and_not1_b32 exec_lo, exec_lo, s19 s_cbranch_execnz .LBB1_5 s_branch .LBB1_2 .LBB1_8: s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z22EuclideanDistancesFastPKfS0_Pfii .amdhsa_group_segment_fixed_size 8 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 288 .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 8 .amdhsa_next_free_sgpr 20 .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 _Z22EuclideanDistancesFastPKfS0_Pfii, .Lfunc_end1-_Z22EuclideanDistancesFastPKfS0_Pfii .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 - .offset: 28 .size: 4 .value_kind: by_value - .offset: 32 .size: 4 .value_kind: hidden_block_count_x - .offset: 36 .size: 4 .value_kind: hidden_block_count_y - .offset: 40 .size: 4 .value_kind: hidden_block_count_z - .offset: 44 .size: 2 .value_kind: hidden_group_size_x - .offset: 46 .size: 2 .value_kind: hidden_group_size_y - .offset: 48 .size: 2 .value_kind: hidden_group_size_z - .offset: 50 .size: 2 .value_kind: hidden_remainder_x - .offset: 52 .size: 2 .value_kind: hidden_remainder_y - .offset: 54 .size: 2 .value_kind: hidden_remainder_z - .offset: 72 .size: 8 .value_kind: hidden_global_offset_x - .offset: 80 .size: 8 .value_kind: hidden_global_offset_y - .offset: 88 .size: 8 .value_kind: hidden_global_offset_z - .offset: 96 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 288 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z23EuclideanDistancesNaivePfS_S_ii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z23EuclideanDistancesNaivePfS_S_ii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 11 .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 - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: by_value - .offset: 28 .size: 4 .value_kind: by_value - .offset: 32 .size: 4 .value_kind: hidden_block_count_x - .offset: 36 .size: 4 .value_kind: hidden_block_count_y - .offset: 40 .size: 4 .value_kind: hidden_block_count_z - .offset: 44 .size: 2 .value_kind: hidden_group_size_x - .offset: 46 .size: 2 .value_kind: hidden_group_size_y - .offset: 48 .size: 2 .value_kind: hidden_group_size_z - .offset: 50 .size: 2 .value_kind: hidden_remainder_x - .offset: 52 .size: 2 .value_kind: hidden_remainder_y - .offset: 54 .size: 2 .value_kind: hidden_remainder_z - .offset: 72 .size: 8 .value_kind: hidden_global_offset_x - .offset: 80 .size: 8 .value_kind: hidden_global_offset_y - .offset: 88 .size: 8 .value_kind: hidden_global_offset_z - .offset: 96 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 8 .kernarg_segment_align: 8 .kernarg_segment_size: 288 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z22EuclideanDistancesFastPKfS0_Pfii .private_segment_fixed_size: 0 .sgpr_count: 22 .sgpr_spill_count: 0 .symbol: _Z22EuclideanDistancesFastPKfS0_Pfii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 8 .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> #include <stdlib.h> #include <iostream> // both M and N must be evenly divisible by SIZE, M must be evenly divisible by CHKSIZE #define SIZE 2 #define N 2 #define M 3 #define CHKSIZE 1 //naive kernel __global__ void EuclideanDistancesNaive(float *A, float *B, float *C, int n, int m) { int idx = threadIdx.x + blockDim.x * blockIdx.x; int idy = threadIdx.y + blockDim.y * blockIdx.y; float result = 0.0f; if ((idx < n) && (idy < m)) { for (int i = 0; i < SIZE; i++) { float temp = A[(idx * SIZE) + i] - B[(idy * SIZE) + i]; result += temp * temp; } C[(idx * m) + idy] = result; } } //optimized kernel __global__ void EuclideanDistancesFast(const float *A, const float *B, float *C, const int n, const int m) { // n, A, 4000 this kernel assumes A is column-major A(SIZE, n) // m, B, 20000 this kernel assumes B is row-major B(m, SIZE) // this kernel assumes C is column-major C(m,n) // this kernel assumes number of threads per threadblock == SIZE // CHKSIZE is the number of B vectors that will be compute per block __shared__ float my_sB[CHKSIZE * SIZE]; // enough shared storage for CHKSIZE vectors of B int bx = blockIdx.x; // one block per CHKSIZE rows of B (the larger input matrix) while ((bx * CHKSIZE) < m) { // not used, this while loop could be used to extend a block to multiple chunks int tx = threadIdx.x; for (int i = 0; i < CHKSIZE; i++) // load vectors of B into shared memory my_sB[(i * SIZE) + tx] = B[(((bx * CHKSIZE) + i) * SIZE) + tx]; __syncthreads(); while (tx < n) { //loop across all vectors in A float result[CHKSIZE]; for (int i = 0; i < CHKSIZE; i++) result[i] = 0.0f; for (int i = 0; i < SIZE; i++) { float Atemp = A[(n * i) + tx]; for (int j = 0; j < CHKSIZE; j++) { // compute all CHKSIZE B vectors with read of A float temp = Atemp - my_sB[i + (j * SIZE)]; result[j] += temp * temp; } } for (int i = 0; i < CHKSIZE; i++) // store CHKSIZE results C[((i + (bx * CHKSIZE)) * n) + tx] = result[i]; tx += blockDim.x; } // continue looping across vectors in A __syncthreads(); // necessary to prevent warps from racing ahead, if block looping is used bx += gridDim.x; } } float comp_euclid_sq(const float *rA, const float *rB, const int size) { float result = 0.0f; float temp; for (int i = 0; i < size; i++) { temp = (rA[i] - rB[i]); result += temp * temp; } return result; } int main_ed_gpu() { float cpu_time = 0.0f, et1 = 0.0f, et2 = 0.0f, et_mem = 0.0f; hipEvent_t start1, start2, stop1, stop2, start_mem_copy, stop_mem_copy; hipEventCreate(&start1); hipEventCreate(&start2); hipEventCreate(&start_mem_copy); hipEventCreate(&stop1); hipEventCreate(&stop2); hipEventCreate(&stop_mem_copy); int n = N; //MatrixA size : n * SIZE int m = M; //MatrixB size : m * SIZE srand((unsigned) time(0)); // Host Allocations float *matrixA = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n * SIZE; i++) matrixA[i] = (float) i + 1; float *matrixB = (float *) malloc(m * SIZE * sizeof(float)); for (int i = 0; i < m * SIZE; i++) matrixB[i] = (float) i + i; const clock_t begin_time = clock(); float *results_kernel = (float *) malloc(n * m * sizeof(float)); float *cpu_results_kernel = (float *) malloc(n * m * sizeof(float)); for (int i = 0; i < n * m; i++) cpu_results_kernel[i] = comp_euclid_sq(matrixA + ((i / m) * SIZE), matrixB + (i % m) * SIZE, SIZE); cpu_time = float( clock () - begin_time ) / 1000; //Device Allocation hipEventRecord(start_mem_copy); float *d_matrixA; float *d_matrixB; hipMalloc((void **) &d_matrixA, n * SIZE * sizeof(float)); hipMalloc((void **) &d_matrixB, m * SIZE * sizeof(float)); hipMemcpy(d_matrixA, matrixA, n * SIZE * sizeof(float), hipMemcpyHostToDevice); hipMemcpy(d_matrixB, matrixB, m * SIZE * sizeof(float), hipMemcpyHostToDevice); hipEventElapsedTime(&et_mem, start_mem_copy, stop_mem_copy); float *d_results_kernel; hipMalloc((void **) &d_results_kernel, n * m * sizeof(float)); hipFuncSetCacheConfig(reinterpret_cast<const void*>(EuclideanDistancesNaive), hipFuncCachePreferL1); dim3 threads3(8, 32); // 1024 threads per block (maximum) dim3 blocks3(n / threads3.x, m / threads3.y); // assumes evenly divisible hipEventRecord(start1); EuclideanDistancesNaive<<<blocks3, threads3>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); hipEventRecord(stop1); hipMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), hipMemcpyDeviceToHost); // for (int i = 0; i < n * m; i++) { // if (results_kernel[i] != cpu_results_kernel[i]) { // printf("cpu/kernel3 mismatch at %d, cpu: %f, kernel3: %f\n", i, // cpu_results_kernel[i], results_kernel[i]); // return 1; // } // } hipMemset(d_results_kernel, 0, n * m * sizeof(float)); hipEventSynchronize(stop1); hipEventElapsedTime(&et1, start1, stop1); // transpose matrix A float *matrixA_T = (float *) malloc(n * SIZE * sizeof(float)); for (int i = 0; i < n; i++) for (int j = 0; j < SIZE; j++) matrixA_T[(j * n) + i] = matrixA[(i * SIZE) + j]; hipMemcpy(d_matrixA, matrixA_T, n * SIZE * sizeof(float), hipMemcpyHostToDevice); hipFuncSetCacheConfig(reinterpret_cast<const void*>(EuclideanDistancesFast), hipFuncCachePreferL1); dim3 threads4(SIZE); // one thread per vector element dim3 blocks4(m / CHKSIZE); hipEventRecord(start2); EuclideanDistancesFast<<<blocks4, threads4>>>(d_matrixA, d_matrixB, d_results_kernel, n, m); hipEventRecord(stop2); hipMemcpy(results_kernel, d_results_kernel, n * m * sizeof(float), hipMemcpyDeviceToHost); // test for correct transposed result C(m,n) for (int i = 0; i < n; i++) { for (int j = 0; j < m; j++) { printf("%f ", results_kernel[(j * n) + i]); if (results_kernel[(j * n) + i] != cpu_results_kernel[(i * m) + j]) { printf("cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n", i, j, cpu_results_kernel[(i * m) + j], results_kernel[(j * n) + i]); return 1; } } printf("\n"); } hipEventSynchronize(stop2); hipEventElapsedTime(&et2, start2, stop2); hipFree(d_results_kernel); printf("Success!\n"); printf("CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n", cpu_time, et1, et2, et_mem); free(matrixA); free(matrixB); free(results_kernel); return 0; }
.text .file "euclidean_distance_fast.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii # -- Begin function _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .p2align 4, 0x90 .type _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii,@function _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii: # @_Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .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) movl %r8d, (%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) movq %rsp, %rax movq %rax, 112(%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 $_Z23EuclideanDistancesNaivePfS_S_ii, %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 _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii, .Lfunc_end0-_Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .cfi_endproc # -- End function .globl _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii # -- Begin function _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .p2align 4, 0x90 .type _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii,@function _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii: # @_Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .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) movl %r8d, (%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) movq %rsp, %rax movq %rax, 112(%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 $_Z22EuclideanDistancesFastPKfS0_Pfii, %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_end1: .size _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii, .Lfunc_end1-_Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .cfi_endproc # -- End function .globl _Z14comp_euclid_sqPKfS0_i # -- Begin function _Z14comp_euclid_sqPKfS0_i .p2align 4, 0x90 .type _Z14comp_euclid_sqPKfS0_i,@function _Z14comp_euclid_sqPKfS0_i: # @_Z14comp_euclid_sqPKfS0_i .cfi_startproc # %bb.0: testl %edx, %edx jle .LBB2_1 # %bb.3: # %.lr.ph.preheader movl %edx, %eax xorps %xmm0, %xmm0 xorl %ecx, %ecx .p2align 4, 0x90 .LBB2_4: # %.lr.ph # =>This Inner Loop Header: Depth=1 movss (%rdi,%rcx,4), %xmm1 # xmm1 = mem[0],zero,zero,zero subss (%rsi,%rcx,4), %xmm1 mulss %xmm1, %xmm1 addss %xmm1, %xmm0 incq %rcx cmpq %rcx, %rax jne .LBB2_4 # %bb.2: # %._crit_edge retq .LBB2_1: xorps %xmm0, %xmm0 retq .Lfunc_end2: .size _Z14comp_euclid_sqPKfS0_i, .Lfunc_end2-_Z14comp_euclid_sqPKfS0_i .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function _Z11main_ed_gpuv .LCPI3_0: .long 0x3f800000 # float 1 .LCPI3_1: .long 0x447a0000 # float 1000 .text .globl _Z11main_ed_gpuv .p2align 4, 0x90 .type _Z11main_ed_gpuv,@function _Z11main_ed_gpuv: # @_Z11main_ed_gpuv .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 $248, %rsp .cfi_def_cfa_offset 304 .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 $0, 44(%rsp) movl $0, 40(%rsp) movl $0, 36(%rsp) leaq 168(%rsp), %rdi callq hipEventCreate leaq 160(%rsp), %rdi callq hipEventCreate leaq 152(%rsp), %rdi callq hipEventCreate leaq 72(%rsp), %rdi callq hipEventCreate leaq 64(%rsp), %rdi callq hipEventCreate leaq 240(%rsp), %rdi callq hipEventCreate xorl %r14d, %r14d xorl %edi, %edi callq time movl %eax, %edi callq srand movl $16, %edi callq malloc movq %rax, %rbx movss .LCPI3_0(%rip), %xmm0 # xmm0 = mem[0],zero,zero,zero .p2align 4, 0x90 .LBB3_1: # =>This Inner Loop Header: Depth=1 xorps %xmm1, %xmm1 cvtsi2ss %r14d, %xmm1 addss %xmm0, %xmm1 movss %xmm1, (%rbx,%r14,4) incq %r14 cmpq $4, %r14 jne .LBB3_1 # %bb.2: movl $24, %edi callq malloc movq %rax, %r14 xorl %eax, %eax .p2align 4, 0x90 .LBB3_3: # =>This Inner Loop Header: Depth=1 xorps %xmm0, %xmm0 cvtsi2ss %eax, %xmm0 addss %xmm0, %xmm0 movss %xmm0, (%r14,%rax,4) incq %rax cmpq $6, %rax jne .LBB3_3 # %bb.4: callq clock movq %rax, %r13 movl $24, %edi callq malloc movq %rax, %r15 movl $24, %edi callq malloc movq %rax, %r12 xorl %esi, %esi movl $2863311531, %edi # imm = 0xAAAAAAAB movabsq $-6148914691236517205, %r8 # imm = 0xAAAAAAAAAAAAAAAB xorl %ecx, %ecx .p2align 4, 0x90 .LBB3_5: # =>This Loop Header: Depth=1 # Child Loop BB3_6 Depth 2 movl %ecx, %eax imulq %rdi, %rax shrq $33, %rax addl %eax, %eax leal (%rax,%rax,2), %eax movl %esi, %edx subl %eax, %edx leaq (%r14,%rdx,4), %r9 movq %rcx, %rax mulq %r8 andq $-2, %rdx leaq (%rbx,%rdx,4), %rax xorps %xmm0, %xmm0 xorl %edx, %edx .p2align 4, 0x90 .LBB3_6: # %.lr.ph.i # Parent Loop BB3_5 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rax,%rdx,4), %xmm1 # xmm1 = mem[0],zero,zero,zero subss (%r9,%rdx,4), %xmm1 mulss %xmm1, %xmm1 addss %xmm1, %xmm0 incq %rdx cmpq $2, %rdx jne .LBB3_6 # %bb.7: # %_Z14comp_euclid_sqPKfS0_i.exit # in Loop: Header=BB3_5 Depth=1 movss %xmm0, (%r12,%rcx,4) incq %rcx addl $2, %esi cmpq $6, %rcx jne .LBB3_5 # %bb.8: callq clock subq %r13, %rax xorps %xmm0, %xmm0 cvtsi2ss %rax, %xmm0 movss %xmm0, 12(%rsp) # 4-byte Spill movq 152(%rsp), %rdi xorl %esi, %esi callq hipEventRecord leaq 48(%rsp), %rdi movl $16, %esi callq hipMalloc leaq 56(%rsp), %rdi movl $24, %esi callq hipMalloc movq 48(%rsp), %rdi movl $16, %edx movq %rbx, %rsi movl $1, %ecx callq hipMemcpy movq 56(%rsp), %rdi movl $24, %edx movq %r14, %rsi movl $1, %ecx callq hipMemcpy movq 152(%rsp), %rsi movq 240(%rsp), %rdx leaq 36(%rsp), %rdi callq hipEventElapsedTime leaq 24(%rsp), %rdi movl $24, %esi callq hipMalloc movl $_Z23EuclideanDistancesNaivePfS_S_ii, %edi movl $2, %esi callq hipFuncSetCacheConfig movq 168(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movabsq $137438953480, %rdx # imm = 0x2000000008 xorl %edi, %edi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB3_10 # %bb.9: movq 48(%rsp), %rax movq 56(%rsp), %rcx movq 24(%rsp), %rdx movq %rax, 144(%rsp) movq %rcx, 136(%rsp) movq %rdx, 128(%rsp) movl $2, 20(%rsp) movl $3, 16(%rsp) leaq 144(%rsp), %rax movq %rax, 176(%rsp) leaq 136(%rsp), %rax movq %rax, 184(%rsp) leaq 128(%rsp), %rax movq %rax, 192(%rsp) leaq 20(%rsp), %rax movq %rax, 200(%rsp) leaq 16(%rsp), %rax movq %rax, 208(%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 leaq 176(%rsp), %r9 movl $_Z23EuclideanDistancesNaivePfS_S_ii, %edi 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 .LBB3_10: movss 12(%rsp), %xmm0 # 4-byte Reload # xmm0 = mem[0],zero,zero,zero divss .LCPI3_1(%rip), %xmm0 movss %xmm0, 12(%rsp) # 4-byte Spill movq 72(%rsp), %rdi xorl %r13d, %r13d xorl %esi, %esi callq hipEventRecord movq 24(%rsp), %rsi movl $24, %edx movq %r15, %rdi movl $2, %ecx callq hipMemcpy movq 24(%rsp), %rdi movl $24, %edx xorl %esi, %esi callq hipMemset movq 72(%rsp), %rdi callq hipEventSynchronize movq 168(%rsp), %rsi movq 72(%rsp), %rdx leaq 44(%rsp), %rdi callq hipEventElapsedTime movl $16, %edi callq malloc movq %rax, %rcx movq %rbx, %rdx .p2align 4, 0x90 .LBB3_11: # %.preheader145 # =>This Loop Header: Depth=1 # Child Loop BB3_12 Depth 2 xorl %esi, %esi .p2align 4, 0x90 .LBB3_12: # Parent Loop BB3_11 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rdx,%rsi,4), %xmm0 # xmm0 = mem[0],zero,zero,zero movss %xmm0, (%rcx,%rsi,8) incq %rsi cmpq $1, %rsi je .LBB3_12 # %bb.13: # in Loop: Header=BB3_11 Depth=1 leaq 1(%r13), %rsi addq $8, %rdx addq $4, %rcx testq %r13, %r13 movq %rsi, %r13 je .LBB3_11 # %bb.14: movq 48(%rsp), %rdi movl $16, %edx movq %rax, %rsi movl $1, %ecx callq hipMemcpy movl $_Z22EuclideanDistancesFastPKfS0_Pfii, %edi movl $2, %esi callq hipFuncSetCacheConfig movq 160(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movabsq $4294967298, %rdx # imm = 0x100000002 leaq 1(%rdx), %rdi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB3_16 # %bb.15: movq 48(%rsp), %rax movq 56(%rsp), %rcx movq 24(%rsp), %rdx movq %rax, 144(%rsp) movq %rcx, 136(%rsp) movq %rdx, 128(%rsp) movl $2, 20(%rsp) movl $3, 16(%rsp) leaq 144(%rsp), %rax movq %rax, 176(%rsp) leaq 136(%rsp), %rax movq %rax, 184(%rsp) leaq 128(%rsp), %rax movq %rax, 192(%rsp) leaq 20(%rsp), %rax movq %rax, 200(%rsp) leaq 16(%rsp), %rax movq %rax, 208(%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 leaq 176(%rsp), %r9 movl $_Z22EuclideanDistancesFastPKfS0_Pfii, %edi 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 .LBB3_16: movq 64(%rsp), %rdi xorl %r13d, %r13d xorl %esi, %esi callq hipEventRecord movq 24(%rsp), %rsi movl $24, %edx movq %r15, %rdi movl $2, %ecx callq hipMemcpy movq %r15, 224(%rsp) # 8-byte Spill xorl %ecx, %ecx .LBB3_17: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB3_18 Depth 2 movq %rcx, 232(%rsp) # 8-byte Spill xorl %ebp, %ebp .p2align 4, 0x90 .LBB3_18: # Parent Loop BB3_17 Depth=1 # => This Inner Loop Header: Depth=2 movss (%r15,%rbp,8), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str, %edi movb $1, %al callq printf movss (%r15,%rbp,8), %xmm1 # xmm1 = mem[0],zero,zero,zero movss (%r12,%rbp,4), %xmm0 # xmm0 = mem[0],zero,zero,zero ucomiss %xmm0, %xmm1 jne .LBB3_19 jp .LBB3_19 # %bb.20: # in Loop: Header=BB3_18 Depth=2 incq %rbp cmpq $3, %rbp jne .LBB3_18 # %bb.21: # %.critedge # in Loop: Header=BB3_17 Depth=1 movl $10, %edi callq putchar@PLT leaq 1(%r13), %rax addq $12, %r12 addq $4, %r15 testq %r13, %r13 setne %cl movq %rax, %r13 je .LBB3_17 jmp .LBB3_22 .LBB3_19: cvtss2sd %xmm0, %xmm0 cvtss2sd %xmm1, %xmm1 movl $.L.str.1, %edi movl %r13d, %esi movl %ebp, %edx movb $2, %al callq printf movl $1, %eax testb $1, 232(%rsp) # 1-byte Folded Reload je .LBB3_23 .LBB3_22: # %.critedge132 movq 64(%rsp), %rdi callq hipEventSynchronize movq 160(%rsp), %rsi movq 64(%rsp), %rdx leaq 40(%rsp), %rdi callq hipEventElapsedTime movq 24(%rsp), %rdi callq hipFree movl $.Lstr, %edi callq puts@PLT movss 12(%rsp), %xmm0 # 4-byte Reload # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movss 44(%rsp), %xmm1 # xmm1 = mem[0],zero,zero,zero cvtss2sd %xmm1, %xmm1 movss 40(%rsp), %xmm2 # xmm2 = mem[0],zero,zero,zero cvtss2sd %xmm2, %xmm2 movss 36(%rsp), %xmm3 # xmm3 = mem[0],zero,zero,zero cvtss2sd %xmm3, %xmm3 movl $.L.str.4, %edi movb $4, %al callq printf movq %rbx, %rdi callq free movq %r14, %rdi callq free movq 224(%rsp), %rdi # 8-byte Reload callq free xorl %eax, %eax .LBB3_23: addq $248, %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 _Z11main_ed_gpuv, .Lfunc_end3-_Z11main_ed_gpuv .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 $_Z23EuclideanDistancesNaivePfS_S_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 $_Z22EuclideanDistancesFastPKfS0_Pfii, %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 _Z23EuclideanDistancesNaivePfS_S_ii,@object # @_Z23EuclideanDistancesNaivePfS_S_ii .section .rodata,"a",@progbits .globl _Z23EuclideanDistancesNaivePfS_S_ii .p2align 3, 0x0 _Z23EuclideanDistancesNaivePfS_S_ii: .quad _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .size _Z23EuclideanDistancesNaivePfS_S_ii, 8 .type _Z22EuclideanDistancesFastPKfS0_Pfii,@object # @_Z22EuclideanDistancesFastPKfS0_Pfii .globl _Z22EuclideanDistancesFastPKfS0_Pfii .p2align 3, 0x0 _Z22EuclideanDistancesFastPKfS0_Pfii: .quad _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .size _Z22EuclideanDistancesFastPKfS0_Pfii, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "%f " .size .L.str, 4 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz "cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n" .size .L.str.1, 53 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n" .size .L.str.4, 63 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z23EuclideanDistancesNaivePfS_S_ii" .size .L__unnamed_1, 36 .type .L__unnamed_2,@object # @1 .L__unnamed_2: .asciz "_Z22EuclideanDistancesFastPKfS0_Pfii" .size .L__unnamed_2, 37 .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 "Success!" .size .Lstr, 9 .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 _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .addrsig_sym _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z23EuclideanDistancesNaivePfS_S_ii .addrsig_sym _Z22EuclideanDistancesFastPKfS0_Pfii .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 : _Z22EuclideanDistancesFastPKfS0_Pfii .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.X ; /* 0x0000000000007919 */ /* 0x000e240000002500 */ /*0020*/ ISETP.GE.AND P0, PT, R0, c[0x0][0x17c], PT ; /* 0x00005f0000007a0c */ /* 0x001fda0003f06270 */ /*0030*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0040*/ S2R R15, SR_TID.X ; /* 0x00000000000f7919 */ /* 0x000e220000002100 */ /*0050*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fc60000000a00 */ /*0060*/ HFMA2.MMA R17, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff117435 */ /* 0x000fe200000001ff */ /*0070*/ LEA R2, R0, R15, 0x1 ; /* 0x0000000f00027211 */ /* 0x001fd200078e08ff */ /*0080*/ IMAD.WIDE R2, R2, R17, c[0x0][0x168] ; /* 0x00005a0002027625 */ /* 0x000fcc00078e0211 */ /*0090*/ LDG.E R2, [R2.64] ; /* 0x0000000402027981 */ /* 0x000ea2000c1e1900 */ /*00a0*/ ISETP.GE.AND P0, PT, R15, c[0x0][0x178], PT ; /* 0x00005e000f007a0c */ /* 0x000fe20003f06270 */ /*00b0*/ BSSY B0, 0x200 ; /* 0x0000014000007945 */ /* 0x000fe40003800000 */ /*00c0*/ STS [R15.X4], R2 ; /* 0x000000020f007388 */ /* 0x0041e80000004800 */ /*00d0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fec0000010000 */ /*00e0*/ @P0 BRA 0x1f0 ; /* 0x0000010000000947 */ /* 0x000fea0003800000 */ /*00f0*/ LDS.64 R2, [RZ] ; /* 0x00000000ff027984 */ /* 0x001e220000000a00 */ /*0100*/ MOV R11, R15 ; /* 0x0000000f000b7202 */ /* 0x000fca0000000f00 */ /*0110*/ IMAD.WIDE R4, R11, R17, c[0x0][0x160] ; /* 0x000058000b047625 */ /* 0x000fca00078e0211 */ /*0120*/ LDG.E R9, [R4.64] ; /* 0x0000000404097981 */ /* 0x001ea2000c1e1900 */ /*0130*/ IMAD.WIDE R6, R17, c[0x0][0x178], R4 ; /* 0x00005e0011067a25 */ /* 0x000fcc00078e0204 */ /*0140*/ LDG.E R6, [R6.64] ; /* 0x0000000406067981 */ /* 0x000ee2000c1e1900 */ /*0150*/ IMAD R8, R0, c[0x0][0x178], R11 ; /* 0x00005e0000087a24 */ /* 0x000fe200078e020b */ /*0160*/ IADD3 R11, R11, c[0x0][0x0], RZ ; /* 0x000000000b0b7a10 */ /* 0x000fc80007ffe0ff */ /*0170*/ ISETP.GE.AND P0, PT, R11, c[0x0][0x178], PT ; /* 0x00005e000b007a0c */ /* 0x000fe20003f06270 */ /*0180*/ FADD R9, -R2, R9 ; /* 0x0000000902097221 */ /* 0x005fc80000000100 */ /*0190*/ FFMA R10, R9, R9, RZ ; /* 0x00000009090a7223 */ /* 0x000fe400000000ff */ /*01a0*/ IMAD.WIDE R8, R8, R17, c[0x0][0x170] ; /* 0x00005c0008087625 */ /* 0x000fc800078e0211 */ /*01b0*/ FADD R13, R6, -R3 ; /* 0x80000003060d7221 */ /* 0x008fc80000000000 */ /*01c0*/ FFMA R13, R13, R13, R10 ; /* 0x0000000d0d0d7223 */ /* 0x000fca000000000a */ /*01d0*/ STG.E [R8.64], R13 ; /* 0x0000000d08007986 */ /* 0x0001e2000c101904 */ /*01e0*/ @!P0 BRA 0x110 ; /* 0xffffff2000008947 */ /* 0x000fea000383ffff */ /*01f0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x001fea0003800000 */ /*0200*/ IADD3 R0, R0, c[0x0][0xc], RZ ; /* 0x0000030000007a10 */ /* 0x000fe20007ffe0ff */ /*0210*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe60000010000 */ /*0220*/ ISETP.GE.AND P0, PT, R0, c[0x0][0x17c], PT ; /* 0x00005f0000007a0c */ /* 0x000fda0003f06270 */ /*0230*/ @!P0 BRA 0x60 ; /* 0xfffffe2000008947 */ /* 0x000fea000383ffff */ /*0240*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0250*/ BRA 0x250; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0260*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0270*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0280*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0290*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ .......... Function : _Z23EuclideanDistancesNaivePfS_S_ii .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 R7, SR_CTAID.Y ; /* 0x0000000000077919 */ /* 0x000e280000002600 */ /*0020*/ S2R R2, SR_TID.Y ; /* 0x0000000000027919 */ /* 0x000e280000002200 */ /*0030*/ S2R R0, SR_CTAID.X ; /* 0x0000000000007919 */ /* 0x000e680000002500 */ /*0040*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e620000002100 */ /*0050*/ IMAD R7, R7, c[0x0][0x4], R2 ; /* 0x0000010007077a24 */ /* 0x001fca00078e0202 */ /*0060*/ ISETP.GE.AND P0, PT, R7, c[0x0][0x17c], PT ; /* 0x00005f0007007a0c */ /* 0x000fe20003f06270 */ /*0070*/ IMAD R0, R0, c[0x0][0x0], R3 ; /* 0x0000000000007a24 */ /* 0x002fca00078e0203 */ /*0080*/ ISETP.GE.OR P0, PT, R0, c[0x0][0x178], P0 ; /* 0x00005e0000007a0c */ /* 0x000fda0000706670 */ /*0090*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*00a0*/ HFMA2.MMA R13, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff0d7435 */ /* 0x000fe200000001ff */ /*00b0*/ SHF.L.U32 R4, R7, 0x1, RZ ; /* 0x0000000107047819 */ /* 0x000fe200000006ff */ /*00c0*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe20000000a00 */ /*00d0*/ SHF.L.U32 R2, R0, 0x1, RZ ; /* 0x0000000100027819 */ /* 0x000fce00000006ff */ /*00e0*/ IMAD.WIDE R4, R4, R13, c[0x0][0x168] ; /* 0x00005a0004047625 */ /* 0x000fc800078e020d */ /*00f0*/ IMAD.WIDE R2, R2, R13, c[0x0][0x160] ; /* 0x0000580002027625 */ /* 0x000fe200078e020d */ /*0100*/ LDG.E R6, [R4.64] ; /* 0x0000000404067981 */ /* 0x000ea8000c1e1900 */ /*0110*/ LDG.E R9, [R2.64] ; /* 0x0000000402097981 */ /* 0x000ea8000c1e1900 */ /*0120*/ LDG.E R8, [R4.64+0x4] ; /* 0x0000040404087981 */ /* 0x000ee8000c1e1900 */ /*0130*/ LDG.E R11, [R2.64+0x4] ; /* 0x00000404020b7981 */ /* 0x000ee2000c1e1900 */ /*0140*/ IMAD R7, R0, c[0x0][0x17c], R7 ; /* 0x00005f0000077a24 */ /* 0x000fc400078e0207 */ /*0150*/ FADD R6, -R6, R9 ; /* 0x0000000906067221 */ /* 0x004fc80000000100 */ /*0160*/ FFMA R0, R6, R6, RZ ; /* 0x0000000606007223 */ /* 0x000fe400000000ff */ /*0170*/ IMAD.WIDE R6, R7, R13, c[0x0][0x170] ; /* 0x00005c0007067625 */ /* 0x000fc800078e020d */ /*0180*/ FADD R11, -R8, R11 ; /* 0x0000000b080b7221 */ /* 0x008fc80000000100 */ /*0190*/ FFMA R11, R11, R11, R0 ; /* 0x0000000b0b0b7223 */ /* 0x000fca0000000000 */ /*01a0*/ STG.E [R6.64], R11 ; /* 0x0000000b06007986 */ /* 0x000fe2000c101904 */ /*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 _Z23EuclideanDistancesNaivePfS_S_ii .globl _Z23EuclideanDistancesNaivePfS_S_ii .p2align 8 .type _Z23EuclideanDistancesNaivePfS_S_ii,@function _Z23EuclideanDistancesNaivePfS_S_ii: s_clause 0x1 s_load_b32 s4, s[0:1], 0x2c s_load_b64 s[2:3], s[0:1], 0x18 v_and_b32_e32 v2, 0x3ff, v0 v_bfe_u32 v3, v0, 10, 10 s_waitcnt lgkmcnt(0) s_and_b32 s5, s4, 0xffff s_lshr_b32 s4, s4, 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, s4, v[3:4] v_cmp_gt_i32_e32 vcc_lo, s2, 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 s4, s2 s_cbranch_execz .LBB0_4 s_load_b128 s[4:7], s[0:1], 0x0 v_lshlrev_b32_e32 v2, 1, v0 v_lshlrev_b32_e32 v4, 1, v1 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_ashrrev_i32_e32 v5, 31, v4 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[6:7], 2, v[2:3] v_lshlrev_b64 v[8:9], 2, v[4:5] v_mov_b32_e32 v2, 0 s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_4) v_add_co_u32 v3, vcc_lo, s4, v6 v_add_co_ci_u32_e32 v4, vcc_lo, s5, v7, vcc_lo s_delay_alu instid0(VALU_DEP_4) v_add_co_u32 v5, vcc_lo, s6, v8 v_add_co_ci_u32_e32 v6, vcc_lo, s7, v9, vcc_lo s_mov_b64 s[4:5], 0 .LBB0_2: s_delay_alu instid0(SALU_CYCLE_1) v_add_co_u32 v7, vcc_lo, v3, s4 v_add_co_ci_u32_e32 v8, vcc_lo, s5, v4, vcc_lo v_add_co_u32 v9, vcc_lo, v5, s4 v_add_co_ci_u32_e32 v10, vcc_lo, s5, v6, vcc_lo s_add_u32 s4, s4, 4 global_load_b32 v7, v[7:8], off global_load_b32 v8, v[9:10], off s_addc_u32 s5, s5, 0 s_cmp_lg_u32 s4, 4 s_waitcnt vmcnt(0) v_sub_f32_e32 v7, v7, v8 s_delay_alu instid0(VALU_DEP_1) v_fmac_f32_e32 v2, v7, v7 s_cbranch_scc0 .LBB0_2 s_load_b64 s[0:1], s[0:1], 0x10 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u64_u32 v[3:4], null, v0, s3, v[1:2] v_ashrrev_i32_e32 v4, 31, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[0:1], 2, v[3:4] 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_4: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z23EuclideanDistancesNaivePfS_S_ii .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 288 .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 11 .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 _Z23EuclideanDistancesNaivePfS_S_ii, .Lfunc_end0-_Z23EuclideanDistancesNaivePfS_S_ii .section .AMDGPU.csdata,"",@progbits .text .protected _Z22EuclideanDistancesFastPKfS0_Pfii .globl _Z22EuclideanDistancesFastPKfS0_Pfii .p2align 8 .type _Z22EuclideanDistancesFastPKfS0_Pfii,@function _Z22EuclideanDistancesFastPKfS0_Pfii: s_load_b32 s14, s[0:1], 0x1c s_waitcnt lgkmcnt(0) s_cmp_ge_i32 s15, s14 s_cbranch_scc1 .LBB1_8 s_clause 0x3 s_load_b32 s2, s[0:1], 0x18 s_load_b128 s[4:7], s[0:1], 0x0 s_load_b64 s[8:9], s[0:1], 0x10 s_load_b32 s16, s[0:1], 0x20 v_lshlrev_b32_e32 v4, 2, v0 s_add_u32 s10, s0, 32 s_addc_u32 s11, s1, 0 s_waitcnt lgkmcnt(0) v_cmp_gt_i32_e32 vcc_lo, s2, v0 s_ashr_i32 s3, s2, 31 s_delay_alu instid0(SALU_CYCLE_1) s_lshl_b64 s[12:13], s[2:3], 2 s_branch .LBB1_3 .LBB1_2: s_set_inst_prefetch_distance 0x2 s_or_b32 exec_lo, exec_lo, s3 s_add_i32 s15, s16, s15 s_waitcnt_vscnt null, 0x0 s_cmp_lt_i32 s15, s14 s_barrier buffer_gl0_inv s_cbranch_scc0 .LBB1_8 .LBB1_3: v_lshl_add_u32 v1, s15, 1, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_ashrrev_i32_e32 v2, 31, v1 v_lshlrev_b64 v[1:2], 2, v[1:2] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_co_u32 v1, s0, s6, v1 v_add_co_ci_u32_e64 v2, s0, s7, v2, s0 global_load_b32 v1, v[1:2], off s_waitcnt vmcnt(0) ds_store_b32 v4, v1 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_and_saveexec_b32 s3, vcc_lo s_cbranch_execz .LBB1_2 s_load_b32 s0, s[10:11], 0xc v_mov_b32_e32 v1, v0 s_mul_i32 s17, s15, s2 s_mov_b32 s19, 0 s_waitcnt lgkmcnt(0) s_and_b32 s18, s0, 0xffff s_set_inst_prefetch_distance 0x1 .p2align 6 .LBB1_5: v_ashrrev_i32_e32 v2, 31, v1 v_mov_b32_e32 v5, 0 s_mov_b32 s1, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b64 v[2:3], 2, v[1:2] v_add_co_u32 v2, s0, s4, v2 s_delay_alu instid0(VALU_DEP_1) v_add_co_ci_u32_e64 v3, s0, s5, v3, s0 .LBB1_6: global_load_b32 v6, v[2:3], off v_mov_b32_e32 v7, s1 v_add_co_u32 v2, s0, v2, s12 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) v_add_co_ci_u32_e64 v3, s0, s13, v3, s0 ds_load_b32 v7, v7 s_add_i32 s1, s1, 4 s_cmp_eq_u32 s1, 4 s_waitcnt vmcnt(0) lgkmcnt(0) v_sub_f32_e32 v6, v6, v7 s_delay_alu instid0(VALU_DEP_1) v_fmac_f32_e32 v5, v6, v6 s_cbranch_scc1 .LBB1_6 v_add_nc_u32_e32 v2, s17, v1 v_add_nc_u32_e32 v1, s18, v1 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_cmp_le_i32_e64 s0, s2, v1 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[2:3], 2, v[2:3] s_or_b32 s19, s0, s19 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_co_u32 v2, s1, s8, v2 v_add_co_ci_u32_e64 v3, s1, s9, v3, s1 global_store_b32 v[2:3], v5, off s_and_not1_b32 exec_lo, exec_lo, s19 s_cbranch_execnz .LBB1_5 s_branch .LBB1_2 .LBB1_8: s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z22EuclideanDistancesFastPKfS0_Pfii .amdhsa_group_segment_fixed_size 8 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 288 .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 8 .amdhsa_next_free_sgpr 20 .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 _Z22EuclideanDistancesFastPKfS0_Pfii, .Lfunc_end1-_Z22EuclideanDistancesFastPKfS0_Pfii .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 - .offset: 28 .size: 4 .value_kind: by_value - .offset: 32 .size: 4 .value_kind: hidden_block_count_x - .offset: 36 .size: 4 .value_kind: hidden_block_count_y - .offset: 40 .size: 4 .value_kind: hidden_block_count_z - .offset: 44 .size: 2 .value_kind: hidden_group_size_x - .offset: 46 .size: 2 .value_kind: hidden_group_size_y - .offset: 48 .size: 2 .value_kind: hidden_group_size_z - .offset: 50 .size: 2 .value_kind: hidden_remainder_x - .offset: 52 .size: 2 .value_kind: hidden_remainder_y - .offset: 54 .size: 2 .value_kind: hidden_remainder_z - .offset: 72 .size: 8 .value_kind: hidden_global_offset_x - .offset: 80 .size: 8 .value_kind: hidden_global_offset_y - .offset: 88 .size: 8 .value_kind: hidden_global_offset_z - .offset: 96 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 288 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z23EuclideanDistancesNaivePfS_S_ii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z23EuclideanDistancesNaivePfS_S_ii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 11 .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 - .address_space: global .offset: 16 .size: 8 .value_kind: global_buffer - .offset: 24 .size: 4 .value_kind: by_value - .offset: 28 .size: 4 .value_kind: by_value - .offset: 32 .size: 4 .value_kind: hidden_block_count_x - .offset: 36 .size: 4 .value_kind: hidden_block_count_y - .offset: 40 .size: 4 .value_kind: hidden_block_count_z - .offset: 44 .size: 2 .value_kind: hidden_group_size_x - .offset: 46 .size: 2 .value_kind: hidden_group_size_y - .offset: 48 .size: 2 .value_kind: hidden_group_size_z - .offset: 50 .size: 2 .value_kind: hidden_remainder_x - .offset: 52 .size: 2 .value_kind: hidden_remainder_y - .offset: 54 .size: 2 .value_kind: hidden_remainder_z - .offset: 72 .size: 8 .value_kind: hidden_global_offset_x - .offset: 80 .size: 8 .value_kind: hidden_global_offset_y - .offset: 88 .size: 8 .value_kind: hidden_global_offset_z - .offset: 96 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 8 .kernarg_segment_align: 8 .kernarg_segment_size: 288 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z22EuclideanDistancesFastPKfS0_Pfii .private_segment_fixed_size: 0 .sgpr_count: 22 .sgpr_spill_count: 0 .symbol: _Z22EuclideanDistancesFastPKfS0_Pfii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 8 .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_000df931_00000000-6_euclidean_distance_fast.cudafe1.cpp" .text #APP .globl _ZSt21ios_base_library_initv #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB3673: .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 .LFE3673: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z14comp_euclid_sqPKfS0_i .type _Z14comp_euclid_sqPKfS0_i, @function _Z14comp_euclid_sqPKfS0_i: .LFB3669: .cfi_startproc endbr64 testl %edx, %edx jle .L6 movslq %edx, %rdx salq $2, %rdx movl $0, %eax pxor %xmm1, %xmm1 .L5: movss (%rdi,%rax), %xmm0 subss (%rsi,%rax), %xmm0 mulss %xmm0, %xmm0 addss %xmm0, %xmm1 addq $4, %rax cmpq %rdx, %rax jne .L5 .L3: movaps %xmm1, %xmm0 ret .L6: pxor %xmm1, %xmm1 jmp .L3 .cfi_endproc .LFE3669: .size _Z14comp_euclid_sqPKfS0_i, .-_Z14comp_euclid_sqPKfS0_i .globl _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii .type _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii, @function _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii: .LFB3695: .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) movl %r8d, (%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) movq %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 .L12 .L8: movq 136(%rsp), %rax subq %fs:40, %rax jne .L13 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L12: .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 _Z23EuclideanDistancesNaivePfS_S_ii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L8 .L13: call __stack_chk_fail@PLT .cfi_endproc .LFE3695: .size _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii, .-_Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii .globl _Z23EuclideanDistancesNaivePfS_S_ii .type _Z23EuclideanDistancesNaivePfS_S_ii, @function _Z23EuclideanDistancesNaivePfS_S_ii: .LFB3696: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3696: .size _Z23EuclideanDistancesNaivePfS_S_ii, .-_Z23EuclideanDistancesNaivePfS_S_ii .globl _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii .type _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii, @function _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii: .LFB3697: .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) movl %r8d, (%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) movq %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 .L20 .L16: movq 136(%rsp), %rax subq %fs:40, %rax jne .L21 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L20: .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 _Z22EuclideanDistancesFastPKfS0_Pfii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L16 .L21: call __stack_chk_fail@PLT .cfi_endproc .LFE3697: .size _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii, .-_Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii .globl _Z22EuclideanDistancesFastPKfS0_Pfii .type _Z22EuclideanDistancesFastPKfS0_Pfii, @function _Z22EuclideanDistancesFastPKfS0_Pfii: .LFB3698: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3698: .size _Z22EuclideanDistancesFastPKfS0_Pfii, .-_Z22EuclideanDistancesFastPKfS0_Pfii .section .rodata.str1.1,"aMS",@progbits,1 .LC6: .string "%f " .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC7: .string "cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n" .section .rodata.str1.1 .LC8: .string "\n" .LC9: .string "Success!\n" .section .rodata.str1.8 .align 8 .LC10: .string "CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n" .text .globl _Z11main_ed_gpuv .type _Z11main_ed_gpuv, @function _Z11main_ed_gpuv: .LFB3670: .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 $168, %rsp .cfi_def_cfa_offset 224 movq %fs:40, %rax movq %rax, 152(%rsp) xorl %eax, %eax movl $0x00000000, 20(%rsp) movl $0x00000000, 24(%rsp) movl $0x00000000, 28(%rsp) leaq 32(%rsp), %rdi call cudaEventCreate@PLT leaq 40(%rsp), %rdi call cudaEventCreate@PLT leaq 64(%rsp), %rdi call cudaEventCreate@PLT leaq 48(%rsp), %rdi call cudaEventCreate@PLT leaq 56(%rsp), %rdi call cudaEventCreate@PLT leaq 72(%rsp), %rdi call cudaEventCreate@PLT movl $0, %edi call time@PLT movl %eax, %edi call srand@PLT movl $16, %edi call malloc@PLT movq %rax, %rbp movl $0x3f800000, (%rax) movl $0x40000000, 4(%rax) movl $0x40400000, 8(%rax) movl $0x40800000, 12(%rax) movl $24, %edi call malloc@PLT movq %rax, %r13 movl $0, %eax .L25: pxor %xmm0, %xmm0 cvtsi2ssl %eax, %xmm0 addss %xmm0, %xmm0 movss %xmm0, 0(%r13,%rax,4) addq $1, %rax cmpq $6, %rax jne .L25 call clock@PLT movq %rax, %r14 movl $24, %edi call malloc@PLT movq %rax, (%rsp) movl $24, %edi call malloc@PLT movq %rax, %r12 movl $0, %ebx .L26: movslq %ebx, %rax imulq $1431655766, %rax, %rax shrq $32, %rax movl %ebx, %edx sarl $31, %edx subl %edx, %eax leal (%rax,%rax), %ecx addl %ecx, %eax movl %ebx, %edx subl %eax, %edx leal (%rdx,%rdx), %eax cltq leaq 0(%r13,%rax,4), %rsi movslq %ecx, %rcx leaq 0(%rbp,%rcx,4), %rdi movl $2, %edx call _Z14comp_euclid_sqPKfS0_i movss %xmm0, (%r12,%rbx,4) addq $1, %rbx cmpq $6, %rbx jne .L26 call clock@PLT subq %r14, %rax pxor %xmm0, %xmm0 cvtsi2ssq %rax, %xmm0 divss .LC5(%rip), %xmm0 movss %xmm0, 12(%rsp) movl $0, %esi movq 64(%rsp), %rdi call cudaEventRecord@PLT leaq 80(%rsp), %rdi movl $16, %esi call cudaMalloc@PLT leaq 88(%rsp), %rdi movl $24, %esi call cudaMalloc@PLT movl $1, %ecx movl $16, %edx movq %rbp, %rsi movq 80(%rsp), %rdi call cudaMemcpy@PLT movl $1, %ecx movl $24, %edx movq %r13, %rsi movq 88(%rsp), %rdi call cudaMemcpy@PLT leaq 28(%rsp), %rdi movq 72(%rsp), %rdx movq 64(%rsp), %rsi call cudaEventElapsedTime@PLT leaq 96(%rsp), %rdi movl $24, %esi call cudaMalloc@PLT movl $2, %esi leaq _Z23EuclideanDistancesNaivePfS_S_ii(%rip), %rdi call cudaFuncSetCacheConfig@PLT movl $1, 112(%rsp) movl $0, 116(%rsp) movl $0, 120(%rsp) movl $1, 124(%rsp) movl $0, %esi movq 32(%rsp), %rdi call cudaEventRecord@PLT movl $8, 104(%rsp) movl $32, 108(%rsp) movl 112(%rsp), %ecx movl $0, %r9d movl $0, %r8d movq 104(%rsp), %rdx movq 116(%rsp), %rdi movl 124(%rsp), %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L41 .L27: movl $0, %esi movq 48(%rsp), %rdi call cudaEventRecord@PLT movl $2, %ecx movl $24, %edx movq 96(%rsp), %rsi movq (%rsp), %rdi call cudaMemcpy@PLT movl $24, %edx movl $0, %esi movq 96(%rsp), %rdi call cudaMemset@PLT movq 48(%rsp), %rdi call cudaEventSynchronize@PLT leaq 20(%rsp), %rdi movq 48(%rsp), %rdx movq 32(%rsp), %rsi call cudaEventElapsedTime@PLT movl $16, %edi call malloc@PLT movq %rax, %rsi movss 0(%rbp), %xmm0 movss %xmm0, (%rax) movss 4(%rbp), %xmm0 movss %xmm0, 8(%rax) movss 8(%rbp), %xmm0 movss %xmm0, 4(%rax) movss 12(%rbp), %xmm0 movss %xmm0, 12(%rax) movl $1, %ecx movl $16, %edx movq 80(%rsp), %rdi call cudaMemcpy@PLT movl $2, %esi leaq _Z22EuclideanDistancesFastPKfS0_Pfii(%rip), %rdi call cudaFuncSetCacheConfig@PLT movl $2, 128(%rsp) movl $1, 132(%rsp) movl $1, 136(%rsp) movl $3, 140(%rsp) movl $1, 144(%rsp) movl $1, 148(%rsp) movl $0, %esi movq 40(%rsp), %rdi call cudaEventRecord@PLT movl 136(%rsp), %ecx movl $0, %r9d movl $0, %r8d movq 128(%rsp), %rdx movq 140(%rsp), %rdi movl 148(%rsp), %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L42 .L28: movl $0, %esi movq 56(%rsp), %rdi call cudaEventRecord@PLT movl $2, %ecx movl $24, %edx movq 96(%rsp), %rsi movq (%rsp), %r14 movq %r14, %rdi call cudaMemcpy@PLT movl $0, %r15d .L29: movl $0, %ebx .L33: pxor %xmm0, %xmm0 cvtss2sd (%r14,%rbx,8), %xmm0 leaq .LC6(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movss (%r14,%rbx,8), %xmm1 movss (%r12,%rbx,4), %xmm0 ucomiss %xmm0, %xmm1 jp .L36 jne .L36 addq $1, %rbx cmpq $3, %rbx jne .L33 leaq .LC8(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addl $1, %r15d addq $4, %r14 addq $12, %r12 cmpl $2, %r15d jne .L29 movq 56(%rsp), %rdi call cudaEventSynchronize@PLT leaq 24(%rsp), %rdi movq 56(%rsp), %rdx movq 40(%rsp), %rsi call cudaEventElapsedTime@PLT movq 96(%rsp), %rdi call cudaFree@PLT leaq .LC9(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT pxor %xmm0, %xmm0 cvtss2sd 12(%rsp), %xmm0 pxor %xmm3, %xmm3 cvtss2sd 28(%rsp), %xmm3 pxor %xmm2, %xmm2 cvtss2sd 24(%rsp), %xmm2 pxor %xmm1, %xmm1 cvtss2sd 20(%rsp), %xmm1 leaq .LC10(%rip), %rsi movl $2, %edi movl $4, %eax call __printf_chk@PLT movq %rbp, %rdi call free@PLT movq %r13, %rdi call free@PLT movq (%rsp), %rdi call free@PLT movl $0, %eax .L24: movq 152(%rsp), %rdx subq %fs:40, %rdx jne .L43 addq $168, %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 .L41: .cfi_restore_state movl $3, %r8d movl $2, %ecx movq 96(%rsp), %rdx movq 88(%rsp), %rsi movq 80(%rsp), %rdi call _Z49__device_stub__Z23EuclideanDistancesNaivePfS_S_iiPfS_S_ii jmp .L27 .L42: movl $3, %r8d movl $2, %ecx movq 96(%rsp), %rdx movq 88(%rsp), %rsi movq 80(%rsp), %rdi call _Z50__device_stub__Z22EuclideanDistancesFastPKfS0_PfiiPKfS0_Pfii jmp .L28 .L36: cvtss2sd %xmm0, %xmm0 cvtss2sd %xmm1, %xmm1 movl %ebx, %ecx movl %r15d, %edx leaq .LC7(%rip), %rsi movl $2, %edi movl $2, %eax call __printf_chk@PLT movl $1, %eax jmp .L24 .L43: call __stack_chk_fail@PLT .cfi_endproc .LFE3670: .size _Z11main_ed_gpuv, .-_Z11main_ed_gpuv .section .rodata.str1.8 .align 8 .LC11: .string "_Z22EuclideanDistancesFastPKfS0_Pfii" .align 8 .LC12: .string "_Z23EuclideanDistancesNaivePfS_S_ii" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB3700: .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 _Z22EuclideanDistancesFastPKfS0_Pfii(%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 _Z23EuclideanDistancesNaivePfS_S_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 .LFE3700: .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 .LC5: .long 1148846080 .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 "euclidean_distance_fast.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii # -- Begin function _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .p2align 4, 0x90 .type _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii,@function _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii: # @_Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .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) movl %r8d, (%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) movq %rsp, %rax movq %rax, 112(%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 $_Z23EuclideanDistancesNaivePfS_S_ii, %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 _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii, .Lfunc_end0-_Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .cfi_endproc # -- End function .globl _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii # -- Begin function _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .p2align 4, 0x90 .type _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii,@function _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii: # @_Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .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) movl %r8d, (%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) movq %rsp, %rax movq %rax, 112(%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 $_Z22EuclideanDistancesFastPKfS0_Pfii, %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_end1: .size _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii, .Lfunc_end1-_Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .cfi_endproc # -- End function .globl _Z14comp_euclid_sqPKfS0_i # -- Begin function _Z14comp_euclid_sqPKfS0_i .p2align 4, 0x90 .type _Z14comp_euclid_sqPKfS0_i,@function _Z14comp_euclid_sqPKfS0_i: # @_Z14comp_euclid_sqPKfS0_i .cfi_startproc # %bb.0: testl %edx, %edx jle .LBB2_1 # %bb.3: # %.lr.ph.preheader movl %edx, %eax xorps %xmm0, %xmm0 xorl %ecx, %ecx .p2align 4, 0x90 .LBB2_4: # %.lr.ph # =>This Inner Loop Header: Depth=1 movss (%rdi,%rcx,4), %xmm1 # xmm1 = mem[0],zero,zero,zero subss (%rsi,%rcx,4), %xmm1 mulss %xmm1, %xmm1 addss %xmm1, %xmm0 incq %rcx cmpq %rcx, %rax jne .LBB2_4 # %bb.2: # %._crit_edge retq .LBB2_1: xorps %xmm0, %xmm0 retq .Lfunc_end2: .size _Z14comp_euclid_sqPKfS0_i, .Lfunc_end2-_Z14comp_euclid_sqPKfS0_i .cfi_endproc # -- End function .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 # -- Begin function _Z11main_ed_gpuv .LCPI3_0: .long 0x3f800000 # float 1 .LCPI3_1: .long 0x447a0000 # float 1000 .text .globl _Z11main_ed_gpuv .p2align 4, 0x90 .type _Z11main_ed_gpuv,@function _Z11main_ed_gpuv: # @_Z11main_ed_gpuv .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 $248, %rsp .cfi_def_cfa_offset 304 .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 $0, 44(%rsp) movl $0, 40(%rsp) movl $0, 36(%rsp) leaq 168(%rsp), %rdi callq hipEventCreate leaq 160(%rsp), %rdi callq hipEventCreate leaq 152(%rsp), %rdi callq hipEventCreate leaq 72(%rsp), %rdi callq hipEventCreate leaq 64(%rsp), %rdi callq hipEventCreate leaq 240(%rsp), %rdi callq hipEventCreate xorl %r14d, %r14d xorl %edi, %edi callq time movl %eax, %edi callq srand movl $16, %edi callq malloc movq %rax, %rbx movss .LCPI3_0(%rip), %xmm0 # xmm0 = mem[0],zero,zero,zero .p2align 4, 0x90 .LBB3_1: # =>This Inner Loop Header: Depth=1 xorps %xmm1, %xmm1 cvtsi2ss %r14d, %xmm1 addss %xmm0, %xmm1 movss %xmm1, (%rbx,%r14,4) incq %r14 cmpq $4, %r14 jne .LBB3_1 # %bb.2: movl $24, %edi callq malloc movq %rax, %r14 xorl %eax, %eax .p2align 4, 0x90 .LBB3_3: # =>This Inner Loop Header: Depth=1 xorps %xmm0, %xmm0 cvtsi2ss %eax, %xmm0 addss %xmm0, %xmm0 movss %xmm0, (%r14,%rax,4) incq %rax cmpq $6, %rax jne .LBB3_3 # %bb.4: callq clock movq %rax, %r13 movl $24, %edi callq malloc movq %rax, %r15 movl $24, %edi callq malloc movq %rax, %r12 xorl %esi, %esi movl $2863311531, %edi # imm = 0xAAAAAAAB movabsq $-6148914691236517205, %r8 # imm = 0xAAAAAAAAAAAAAAAB xorl %ecx, %ecx .p2align 4, 0x90 .LBB3_5: # =>This Loop Header: Depth=1 # Child Loop BB3_6 Depth 2 movl %ecx, %eax imulq %rdi, %rax shrq $33, %rax addl %eax, %eax leal (%rax,%rax,2), %eax movl %esi, %edx subl %eax, %edx leaq (%r14,%rdx,4), %r9 movq %rcx, %rax mulq %r8 andq $-2, %rdx leaq (%rbx,%rdx,4), %rax xorps %xmm0, %xmm0 xorl %edx, %edx .p2align 4, 0x90 .LBB3_6: # %.lr.ph.i # Parent Loop BB3_5 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rax,%rdx,4), %xmm1 # xmm1 = mem[0],zero,zero,zero subss (%r9,%rdx,4), %xmm1 mulss %xmm1, %xmm1 addss %xmm1, %xmm0 incq %rdx cmpq $2, %rdx jne .LBB3_6 # %bb.7: # %_Z14comp_euclid_sqPKfS0_i.exit # in Loop: Header=BB3_5 Depth=1 movss %xmm0, (%r12,%rcx,4) incq %rcx addl $2, %esi cmpq $6, %rcx jne .LBB3_5 # %bb.8: callq clock subq %r13, %rax xorps %xmm0, %xmm0 cvtsi2ss %rax, %xmm0 movss %xmm0, 12(%rsp) # 4-byte Spill movq 152(%rsp), %rdi xorl %esi, %esi callq hipEventRecord leaq 48(%rsp), %rdi movl $16, %esi callq hipMalloc leaq 56(%rsp), %rdi movl $24, %esi callq hipMalloc movq 48(%rsp), %rdi movl $16, %edx movq %rbx, %rsi movl $1, %ecx callq hipMemcpy movq 56(%rsp), %rdi movl $24, %edx movq %r14, %rsi movl $1, %ecx callq hipMemcpy movq 152(%rsp), %rsi movq 240(%rsp), %rdx leaq 36(%rsp), %rdi callq hipEventElapsedTime leaq 24(%rsp), %rdi movl $24, %esi callq hipMalloc movl $_Z23EuclideanDistancesNaivePfS_S_ii, %edi movl $2, %esi callq hipFuncSetCacheConfig movq 168(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movabsq $137438953480, %rdx # imm = 0x2000000008 xorl %edi, %edi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB3_10 # %bb.9: movq 48(%rsp), %rax movq 56(%rsp), %rcx movq 24(%rsp), %rdx movq %rax, 144(%rsp) movq %rcx, 136(%rsp) movq %rdx, 128(%rsp) movl $2, 20(%rsp) movl $3, 16(%rsp) leaq 144(%rsp), %rax movq %rax, 176(%rsp) leaq 136(%rsp), %rax movq %rax, 184(%rsp) leaq 128(%rsp), %rax movq %rax, 192(%rsp) leaq 20(%rsp), %rax movq %rax, 200(%rsp) leaq 16(%rsp), %rax movq %rax, 208(%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 leaq 176(%rsp), %r9 movl $_Z23EuclideanDistancesNaivePfS_S_ii, %edi 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 .LBB3_10: movss 12(%rsp), %xmm0 # 4-byte Reload # xmm0 = mem[0],zero,zero,zero divss .LCPI3_1(%rip), %xmm0 movss %xmm0, 12(%rsp) # 4-byte Spill movq 72(%rsp), %rdi xorl %r13d, %r13d xorl %esi, %esi callq hipEventRecord movq 24(%rsp), %rsi movl $24, %edx movq %r15, %rdi movl $2, %ecx callq hipMemcpy movq 24(%rsp), %rdi movl $24, %edx xorl %esi, %esi callq hipMemset movq 72(%rsp), %rdi callq hipEventSynchronize movq 168(%rsp), %rsi movq 72(%rsp), %rdx leaq 44(%rsp), %rdi callq hipEventElapsedTime movl $16, %edi callq malloc movq %rax, %rcx movq %rbx, %rdx .p2align 4, 0x90 .LBB3_11: # %.preheader145 # =>This Loop Header: Depth=1 # Child Loop BB3_12 Depth 2 xorl %esi, %esi .p2align 4, 0x90 .LBB3_12: # Parent Loop BB3_11 Depth=1 # => This Inner Loop Header: Depth=2 movss (%rdx,%rsi,4), %xmm0 # xmm0 = mem[0],zero,zero,zero movss %xmm0, (%rcx,%rsi,8) incq %rsi cmpq $1, %rsi je .LBB3_12 # %bb.13: # in Loop: Header=BB3_11 Depth=1 leaq 1(%r13), %rsi addq $8, %rdx addq $4, %rcx testq %r13, %r13 movq %rsi, %r13 je .LBB3_11 # %bb.14: movq 48(%rsp), %rdi movl $16, %edx movq %rax, %rsi movl $1, %ecx callq hipMemcpy movl $_Z22EuclideanDistancesFastPKfS0_Pfii, %edi movl $2, %esi callq hipFuncSetCacheConfig movq 160(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movabsq $4294967298, %rdx # imm = 0x100000002 leaq 1(%rdx), %rdi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB3_16 # %bb.15: movq 48(%rsp), %rax movq 56(%rsp), %rcx movq 24(%rsp), %rdx movq %rax, 144(%rsp) movq %rcx, 136(%rsp) movq %rdx, 128(%rsp) movl $2, 20(%rsp) movl $3, 16(%rsp) leaq 144(%rsp), %rax movq %rax, 176(%rsp) leaq 136(%rsp), %rax movq %rax, 184(%rsp) leaq 128(%rsp), %rax movq %rax, 192(%rsp) leaq 20(%rsp), %rax movq %rax, 200(%rsp) leaq 16(%rsp), %rax movq %rax, 208(%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 leaq 176(%rsp), %r9 movl $_Z22EuclideanDistancesFastPKfS0_Pfii, %edi 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 .LBB3_16: movq 64(%rsp), %rdi xorl %r13d, %r13d xorl %esi, %esi callq hipEventRecord movq 24(%rsp), %rsi movl $24, %edx movq %r15, %rdi movl $2, %ecx callq hipMemcpy movq %r15, 224(%rsp) # 8-byte Spill xorl %ecx, %ecx .LBB3_17: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB3_18 Depth 2 movq %rcx, 232(%rsp) # 8-byte Spill xorl %ebp, %ebp .p2align 4, 0x90 .LBB3_18: # Parent Loop BB3_17 Depth=1 # => This Inner Loop Header: Depth=2 movss (%r15,%rbp,8), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str, %edi movb $1, %al callq printf movss (%r15,%rbp,8), %xmm1 # xmm1 = mem[0],zero,zero,zero movss (%r12,%rbp,4), %xmm0 # xmm0 = mem[0],zero,zero,zero ucomiss %xmm0, %xmm1 jne .LBB3_19 jp .LBB3_19 # %bb.20: # in Loop: Header=BB3_18 Depth=2 incq %rbp cmpq $3, %rbp jne .LBB3_18 # %bb.21: # %.critedge # in Loop: Header=BB3_17 Depth=1 movl $10, %edi callq putchar@PLT leaq 1(%r13), %rax addq $12, %r12 addq $4, %r15 testq %r13, %r13 setne %cl movq %rax, %r13 je .LBB3_17 jmp .LBB3_22 .LBB3_19: cvtss2sd %xmm0, %xmm0 cvtss2sd %xmm1, %xmm1 movl $.L.str.1, %edi movl %r13d, %esi movl %ebp, %edx movb $2, %al callq printf movl $1, %eax testb $1, 232(%rsp) # 1-byte Folded Reload je .LBB3_23 .LBB3_22: # %.critedge132 movq 64(%rsp), %rdi callq hipEventSynchronize movq 160(%rsp), %rsi movq 64(%rsp), %rdx leaq 40(%rsp), %rdi callq hipEventElapsedTime movq 24(%rsp), %rdi callq hipFree movl $.Lstr, %edi callq puts@PLT movss 12(%rsp), %xmm0 # 4-byte Reload # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movss 44(%rsp), %xmm1 # xmm1 = mem[0],zero,zero,zero cvtss2sd %xmm1, %xmm1 movss 40(%rsp), %xmm2 # xmm2 = mem[0],zero,zero,zero cvtss2sd %xmm2, %xmm2 movss 36(%rsp), %xmm3 # xmm3 = mem[0],zero,zero,zero cvtss2sd %xmm3, %xmm3 movl $.L.str.4, %edi movb $4, %al callq printf movq %rbx, %rdi callq free movq %r14, %rdi callq free movq 224(%rsp), %rdi # 8-byte Reload callq free xorl %eax, %eax .LBB3_23: addq $248, %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 _Z11main_ed_gpuv, .Lfunc_end3-_Z11main_ed_gpuv .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 $_Z23EuclideanDistancesNaivePfS_S_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 $_Z22EuclideanDistancesFastPKfS0_Pfii, %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 _Z23EuclideanDistancesNaivePfS_S_ii,@object # @_Z23EuclideanDistancesNaivePfS_S_ii .section .rodata,"a",@progbits .globl _Z23EuclideanDistancesNaivePfS_S_ii .p2align 3, 0x0 _Z23EuclideanDistancesNaivePfS_S_ii: .quad _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .size _Z23EuclideanDistancesNaivePfS_S_ii, 8 .type _Z22EuclideanDistancesFastPKfS0_Pfii,@object # @_Z22EuclideanDistancesFastPKfS0_Pfii .globl _Z22EuclideanDistancesFastPKfS0_Pfii .p2align 3, 0x0 _Z22EuclideanDistancesFastPKfS0_Pfii: .quad _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .size _Z22EuclideanDistancesFastPKfS0_Pfii, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "%f " .size .L.str, 4 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz "cpu/kernel4 mismatch at %d,%d, cpu: %f, kernel4: %f\n" .size .L.str.1, 53 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "CPU: %.fms, kernel1 : %.fms, kernel2 : %.fms, Mem copy: %.fms\n" .size .L.str.4, 63 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z23EuclideanDistancesNaivePfS_S_ii" .size .L__unnamed_1, 36 .type .L__unnamed_2,@object # @1 .L__unnamed_2: .asciz "_Z22EuclideanDistancesFastPKfS0_Pfii" .size .L__unnamed_2, 37 .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 "Success!" .size .Lstr, 9 .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 _Z38__device_stub__EuclideanDistancesNaivePfS_S_ii .addrsig_sym _Z37__device_stub__EuclideanDistancesFastPKfS0_Pfii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z23EuclideanDistancesNaivePfS_S_ii .addrsig_sym _Z22EuclideanDistancesFastPKfS0_Pfii .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.
__device__ int evalDirt() { return 4; }
code for sm_80
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
__device__ int evalDirt() { return 4; }
.file "tmpxft_000af1d1_00000000-6_dirt.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2030: .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 .LFE2030: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z8evalDirtv .type _Z8evalDirtv, @function _Z8evalDirtv: .LFB2027: .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 .LFE2027: .size _Z8evalDirtv, .-_Z8evalDirtv .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2053: .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 .LFE2053: .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.
__device__ int evalDirt() { return 4; }
#include <hip/hip_runtime.h> __device__ int evalDirt() { return 4; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> __device__ int evalDirt() { return 4; }
.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 <hip/hip_runtime.h> __device__ int evalDirt() { return 4; }
.text .file "dirt.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_000af1d1_00000000-6_dirt.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2030: .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 .LFE2030: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z8evalDirtv .type _Z8evalDirtv, @function _Z8evalDirtv: .LFB2027: .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 .LFE2027: .size _Z8evalDirtv, .-_Z8evalDirtv .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2053: .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 .LFE2053: .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 "dirt.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 host assembly.
#include<cuda.h> #include<cuda_runtime.h> #include<stdio.h> #include<stdlib.h> #include<cmath> #define TILE_SIZE 2 __device__ void store_full(float*,float*,int); __device__ void load_full(float*,float*,int); __device__ void potrf_tile(float*,int,int); __device__ void trsm_tile(float*,int,int,int); __device__ void syrk_tile(float*,int,int,int,int); __global__ void right_looking_launch_kernel(float*,int); __device__ void store_zeros(float*,int); __device__ void store_zeros_diagonal(float*,int,int); __device__ void store_zeros_last(float*,int); __device__ void store_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID + N*N*blockIdx.x] = read_data[ID]; } } __syncthreads(); } __device__ void load_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID] = read_data[ID + N*N*blockIdx.x]; } } __syncthreads(); } __device__ void potrf_tile(float* t_A,int i,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int k=0;k<TILE_SIZE;k++) { if(t_x==t_y && t_x==k) { t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x] = sqrtf(t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]); } __syncthreads(); if(t_x<t_y && t_x == k) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]/= t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]; } __syncthreads(); if(k<t_y && k<t_x && t_x<=t_y) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]-= t_A[i*TILE_SIZE*(1+N) + t_x*N + k]*t_A[i*TILE_SIZE*(1+N) + t_y*N + k]; } __syncthreads(); } } __device__ void trsm_tile(float *row_data,int i,int j,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int s=0;s<TILE_SIZE;s++) { if(t_x==s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]/= row_data[i*TILE_SIZE*(1+N) + t_x*(1+N)]; } __syncthreads(); if(t_x > s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]-= row_data[(t_x + i*TILE_SIZE)*N + s]*row_data[(t_y + j*TILE_SIZE)*N + s]; } __syncthreads(); } } __device__ void syrk_tile(float* row_data,int i,int j,int k,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; float valueToSubtract = 0.0; for(int r=0;r<TILE_SIZE;r++) { valueToSubtract+= row_data[(t_x + k*TILE_SIZE)*N + i*TILE_SIZE + r]*row_data[(t_y + j*TILE_SIZE)*N + i*TILE_SIZE + r]; } row_data[(t_y + j*TILE_SIZE)*N + t_x + k*TILE_SIZE]-= valueToSubtract; __syncthreads(); } __device__ void store_zeros(float* A,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; int i,j; for(i=0;i<N/TILE_SIZE-1;i++) { for(j=i+1;j<N/TILE_SIZE;j++) A[j*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_diagonal(float* A,int N,int b) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; int i; for(i=0;i<N/TILE_SIZE-b;i+=2) { if(t_x>t_y) A[i*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; if(t_x<t_y) A[(i+1)*blockDim.x*(1+N) + (blockDim.x-t_x-1) + (blockDim.y-t_y-1)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_last(float* A,int N) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; if(t_x>t_y) A[(N/TILE_SIZE-1)*blockDim.x*(1+N) + t_x + t_y*N] = 0.0; __syncthreads(); } __global__ void right_looking_launch_kernel(float* read_data,int N) { extern __shared__ float data[]; int i,j,k; load_full(read_data,data,N); for(i=0;i<N/TILE_SIZE;i++) { potrf_tile(data,i,N); for(j=i+1;j<N/TILE_SIZE;j++) { trsm_tile(data,i,j,N); for(k=i+1;k<=j;k++) { syrk_tile(data,i,j,k,N); } } } store_zeros(data,N); if((N/TILE_SIZE)%2==0) store_zeros_diagonal(data,N,0); else { store_zeros_diagonal(data,N,1); store_zeros_last(data,N); } store_full(data,read_data,N); }
.file "tmpxft_0015b61b_00000000-6_right_looking.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2067: .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 .LFE2067: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z10store_fullPfS_i .type _Z10store_fullPfS_i, @function _Z10store_fullPfS_i: .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 _Z10store_fullPfS_i, .-_Z10store_fullPfS_i .globl _Z9load_fullPfS_i .type _Z9load_fullPfS_i, @function _Z9load_fullPfS_i: .LFB2058: .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 .LFE2058: .size _Z9load_fullPfS_i, .-_Z9load_fullPfS_i .globl _Z10potrf_tilePfii .type _Z10potrf_tilePfii, @function _Z10potrf_tilePfii: .LFB2059: .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 .LFE2059: .size _Z10potrf_tilePfii, .-_Z10potrf_tilePfii .globl _Z9trsm_tilePfiii .type _Z9trsm_tilePfiii, @function _Z9trsm_tilePfiii: .LFB2060: .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 .LFE2060: .size _Z9trsm_tilePfiii, .-_Z9trsm_tilePfiii .globl _Z9syrk_tilePfiiii .type _Z9syrk_tilePfiiii, @function _Z9syrk_tilePfiiii: .LFB2061: .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 .LFE2061: .size _Z9syrk_tilePfiiii, .-_Z9syrk_tilePfiiii .globl _Z11store_zerosPfi .type _Z11store_zerosPfi, @function _Z11store_zerosPfi: .LFB2062: .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 .LFE2062: .size _Z11store_zerosPfi, .-_Z11store_zerosPfi .globl _Z20store_zeros_diagonalPfii .type _Z20store_zeros_diagonalPfii, @function _Z20store_zeros_diagonalPfii: .LFB2063: .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 .LFE2063: .size _Z20store_zeros_diagonalPfii, .-_Z20store_zeros_diagonalPfii .globl _Z16store_zeros_lastPfi .type _Z16store_zeros_lastPfi, @function _Z16store_zeros_lastPfi: .LFB2064: .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 .LFE2064: .size _Z16store_zeros_lastPfi, .-_Z16store_zeros_lastPfi .globl _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi .type _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi, @function _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi: .LFB2089: .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 .L23 .L19: movq 104(%rsp), %rax subq %fs:40, %rax jne .L24 addq $120, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L23: .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 _Z27right_looking_launch_kernelPfi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 128 jmp .L19 .L24: call __stack_chk_fail@PLT .cfi_endproc .LFE2089: .size _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi, .-_Z48__device_stub__Z27right_looking_launch_kernelPfiPfi .globl _Z27right_looking_launch_kernelPfi .type _Z27right_looking_launch_kernelPfi, @function _Z27right_looking_launch_kernelPfi: .LFB2090: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2090: .size _Z27right_looking_launch_kernelPfi, .-_Z27right_looking_launch_kernelPfi .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC0: .string "_Z27right_looking_launch_kernelPfi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2092: .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 _Z27right_looking_launch_kernelPfi(%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 .LFE2092: .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<cuda.h> #include<cuda_runtime.h> #include<stdio.h> #include<stdlib.h> #include<cmath> #define TILE_SIZE 2 __device__ void store_full(float*,float*,int); __device__ void load_full(float*,float*,int); __device__ void potrf_tile(float*,int,int); __device__ void trsm_tile(float*,int,int,int); __device__ void syrk_tile(float*,int,int,int,int); __global__ void right_looking_launch_kernel(float*,int); __device__ void store_zeros(float*,int); __device__ void store_zeros_diagonal(float*,int,int); __device__ void store_zeros_last(float*,int); __device__ void store_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID + N*N*blockIdx.x] = read_data[ID]; } } __syncthreads(); } __device__ void load_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID] = read_data[ID + N*N*blockIdx.x]; } } __syncthreads(); } __device__ void potrf_tile(float* t_A,int i,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int k=0;k<TILE_SIZE;k++) { if(t_x==t_y && t_x==k) { t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x] = sqrtf(t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]); } __syncthreads(); if(t_x<t_y && t_x == k) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]/= t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]; } __syncthreads(); if(k<t_y && k<t_x && t_x<=t_y) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]-= t_A[i*TILE_SIZE*(1+N) + t_x*N + k]*t_A[i*TILE_SIZE*(1+N) + t_y*N + k]; } __syncthreads(); } } __device__ void trsm_tile(float *row_data,int i,int j,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int s=0;s<TILE_SIZE;s++) { if(t_x==s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]/= row_data[i*TILE_SIZE*(1+N) + t_x*(1+N)]; } __syncthreads(); if(t_x > s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]-= row_data[(t_x + i*TILE_SIZE)*N + s]*row_data[(t_y + j*TILE_SIZE)*N + s]; } __syncthreads(); } } __device__ void syrk_tile(float* row_data,int i,int j,int k,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; float valueToSubtract = 0.0; for(int r=0;r<TILE_SIZE;r++) { valueToSubtract+= row_data[(t_x + k*TILE_SIZE)*N + i*TILE_SIZE + r]*row_data[(t_y + j*TILE_SIZE)*N + i*TILE_SIZE + r]; } row_data[(t_y + j*TILE_SIZE)*N + t_x + k*TILE_SIZE]-= valueToSubtract; __syncthreads(); } __device__ void store_zeros(float* A,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; int i,j; for(i=0;i<N/TILE_SIZE-1;i++) { for(j=i+1;j<N/TILE_SIZE;j++) A[j*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_diagonal(float* A,int N,int b) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; int i; for(i=0;i<N/TILE_SIZE-b;i+=2) { if(t_x>t_y) A[i*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; if(t_x<t_y) A[(i+1)*blockDim.x*(1+N) + (blockDim.x-t_x-1) + (blockDim.y-t_y-1)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_last(float* A,int N) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; if(t_x>t_y) A[(N/TILE_SIZE-1)*blockDim.x*(1+N) + t_x + t_y*N] = 0.0; __syncthreads(); } __global__ void right_looking_launch_kernel(float* read_data,int N) { extern __shared__ float data[]; int i,j,k; load_full(read_data,data,N); for(i=0;i<N/TILE_SIZE;i++) { potrf_tile(data,i,N); for(j=i+1;j<N/TILE_SIZE;j++) { trsm_tile(data,i,j,N); for(k=i+1;k<=j;k++) { syrk_tile(data,i,j,k,N); } } } store_zeros(data,N); if((N/TILE_SIZE)%2==0) store_zeros_diagonal(data,N,0); else { store_zeros_diagonal(data,N,1); store_zeros_last(data,N); } store_full(data,read_data,N); }
#include<hip/hip_runtime.h> #include<stdio.h> #include<stdlib.h> #include<cmath> #define TILE_SIZE 2 __device__ void store_full(float*,float*,int); __device__ void load_full(float*,float*,int); __device__ void potrf_tile(float*,int,int); __device__ void trsm_tile(float*,int,int,int); __device__ void syrk_tile(float*,int,int,int,int); __global__ void right_looking_launch_kernel(float*,int); __device__ void store_zeros(float*,int); __device__ void store_zeros_diagonal(float*,int,int); __device__ void store_zeros_last(float*,int); __device__ void store_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID + N*N*blockIdx.x] = read_data[ID]; } } __syncthreads(); } __device__ void load_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID] = read_data[ID + N*N*blockIdx.x]; } } __syncthreads(); } __device__ void potrf_tile(float* t_A,int i,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int k=0;k<TILE_SIZE;k++) { if(t_x==t_y && t_x==k) { t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x] = sqrtf(t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]); } __syncthreads(); if(t_x<t_y && t_x == k) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]/= t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]; } __syncthreads(); if(k<t_y && k<t_x && t_x<=t_y) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]-= t_A[i*TILE_SIZE*(1+N) + t_x*N + k]*t_A[i*TILE_SIZE*(1+N) + t_y*N + k]; } __syncthreads(); } } __device__ void trsm_tile(float *row_data,int i,int j,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int s=0;s<TILE_SIZE;s++) { if(t_x==s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]/= row_data[i*TILE_SIZE*(1+N) + t_x*(1+N)]; } __syncthreads(); if(t_x > s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]-= row_data[(t_x + i*TILE_SIZE)*N + s]*row_data[(t_y + j*TILE_SIZE)*N + s]; } __syncthreads(); } } __device__ void syrk_tile(float* row_data,int i,int j,int k,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; float valueToSubtract = 0.0; for(int r=0;r<TILE_SIZE;r++) { valueToSubtract+= row_data[(t_x + k*TILE_SIZE)*N + i*TILE_SIZE + r]*row_data[(t_y + j*TILE_SIZE)*N + i*TILE_SIZE + r]; } row_data[(t_y + j*TILE_SIZE)*N + t_x + k*TILE_SIZE]-= valueToSubtract; __syncthreads(); } __device__ void store_zeros(float* A,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; int i,j; for(i=0;i<N/TILE_SIZE-1;i++) { for(j=i+1;j<N/TILE_SIZE;j++) A[j*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_diagonal(float* A,int N,int b) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; int i; for(i=0;i<N/TILE_SIZE-b;i+=2) { if(t_x>t_y) A[i*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; if(t_x<t_y) A[(i+1)*blockDim.x*(1+N) + (blockDim.x-t_x-1) + (blockDim.y-t_y-1)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_last(float* A,int N) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; if(t_x>t_y) A[(N/TILE_SIZE-1)*blockDim.x*(1+N) + t_x + t_y*N] = 0.0; __syncthreads(); } __global__ void right_looking_launch_kernel(float* read_data,int N) { extern __shared__ float data[]; int i,j,k; load_full(read_data,data,N); for(i=0;i<N/TILE_SIZE;i++) { potrf_tile(data,i,N); for(j=i+1;j<N/TILE_SIZE;j++) { trsm_tile(data,i,j,N); for(k=i+1;k<=j;k++) { syrk_tile(data,i,j,k,N); } } } store_zeros(data,N); if((N/TILE_SIZE)%2==0) store_zeros_diagonal(data,N,0); else { store_zeros_diagonal(data,N,1); store_zeros_last(data,N); } store_full(data,read_data,N); }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include<hip/hip_runtime.h> #include<stdio.h> #include<stdlib.h> #include<cmath> #define TILE_SIZE 2 __device__ void store_full(float*,float*,int); __device__ void load_full(float*,float*,int); __device__ void potrf_tile(float*,int,int); __device__ void trsm_tile(float*,int,int,int); __device__ void syrk_tile(float*,int,int,int,int); __global__ void right_looking_launch_kernel(float*,int); __device__ void store_zeros(float*,int); __device__ void store_zeros_diagonal(float*,int,int); __device__ void store_zeros_last(float*,int); __device__ void store_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID + N*N*blockIdx.x] = read_data[ID]; } } __syncthreads(); } __device__ void load_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID] = read_data[ID + N*N*blockIdx.x]; } } __syncthreads(); } __device__ void potrf_tile(float* t_A,int i,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int k=0;k<TILE_SIZE;k++) { if(t_x==t_y && t_x==k) { t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x] = sqrtf(t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]); } __syncthreads(); if(t_x<t_y && t_x == k) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]/= t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]; } __syncthreads(); if(k<t_y && k<t_x && t_x<=t_y) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]-= t_A[i*TILE_SIZE*(1+N) + t_x*N + k]*t_A[i*TILE_SIZE*(1+N) + t_y*N + k]; } __syncthreads(); } } __device__ void trsm_tile(float *row_data,int i,int j,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int s=0;s<TILE_SIZE;s++) { if(t_x==s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]/= row_data[i*TILE_SIZE*(1+N) + t_x*(1+N)]; } __syncthreads(); if(t_x > s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]-= row_data[(t_x + i*TILE_SIZE)*N + s]*row_data[(t_y + j*TILE_SIZE)*N + s]; } __syncthreads(); } } __device__ void syrk_tile(float* row_data,int i,int j,int k,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; float valueToSubtract = 0.0; for(int r=0;r<TILE_SIZE;r++) { valueToSubtract+= row_data[(t_x + k*TILE_SIZE)*N + i*TILE_SIZE + r]*row_data[(t_y + j*TILE_SIZE)*N + i*TILE_SIZE + r]; } row_data[(t_y + j*TILE_SIZE)*N + t_x + k*TILE_SIZE]-= valueToSubtract; __syncthreads(); } __device__ void store_zeros(float* A,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; int i,j; for(i=0;i<N/TILE_SIZE-1;i++) { for(j=i+1;j<N/TILE_SIZE;j++) A[j*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_diagonal(float* A,int N,int b) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; int i; for(i=0;i<N/TILE_SIZE-b;i+=2) { if(t_x>t_y) A[i*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; if(t_x<t_y) A[(i+1)*blockDim.x*(1+N) + (blockDim.x-t_x-1) + (blockDim.y-t_y-1)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_last(float* A,int N) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; if(t_x>t_y) A[(N/TILE_SIZE-1)*blockDim.x*(1+N) + t_x + t_y*N] = 0.0; __syncthreads(); } __global__ void right_looking_launch_kernel(float* read_data,int N) { extern __shared__ float data[]; int i,j,k; load_full(read_data,data,N); for(i=0;i<N/TILE_SIZE;i++) { potrf_tile(data,i,N); for(j=i+1;j<N/TILE_SIZE;j++) { trsm_tile(data,i,j,N); for(k=i+1;k<=j;k++) { syrk_tile(data,i,j,k,N); } } } store_zeros(data,N); if((N/TILE_SIZE)%2==0) store_zeros_diagonal(data,N,0); else { store_zeros_diagonal(data,N,1); store_zeros_last(data,N); } store_full(data,read_data,N); }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z27right_looking_launch_kernelPfi .globl _Z27right_looking_launch_kernelPfi .p2align 8 .type _Z27right_looking_launch_kernelPfi,@function _Z27right_looking_launch_kernelPfi: s_clause 0x1 s_load_b32 s9, s[0:1], 0x8 s_load_b64 s[6:7], s[0:1], 0x0 v_bfe_u32 v1, v0, 10, 10 s_waitcnt lgkmcnt(0) s_lshr_b32 s2, s9, 31 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_add_i32 s2, s9, s2 s_ashr_i32 s8, s2, 1 s_cmp_gt_i32 s9, 1 s_cselect_b32 s10, -1, 0 s_cmp_lt_i32 s9, 2 s_cbranch_scc1 .LBB0_5 v_mad_u64_u32 v[5:6], null, s15, s9, v[1:2] v_and_b32_e32 v4, 0x3ff, v0 s_mov_b32 s2, 0 s_lshl_b32 s3, s9, 3 s_lshl_b32 s4, s9, 1 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_3) v_mad_u64_u32 v[6:7], null, v1, s9, v[4:5] v_mad_u64_u32 v[2:3], null, s9, v5, v[4:5] v_mov_b32_e32 v4, 0 v_lshl_add_u32 v5, v6, 2, 0 .p2align 6 .LBB0_2: s_delay_alu instid0(VALU_DEP_1) v_dual_mov_b32 v3, v2 :: v_dual_mov_b32 v6, v5 s_mov_b32 s5, s8 .LBB0_3: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) v_lshlrev_b64 v[7:8], 2, v[3:4] v_add_nc_u32_e32 v3, 2, v3 s_add_i32 s5, s5, -1 s_cmp_lg_u32 s5, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v7, vcc_lo, s6, v7 v_add_co_ci_u32_e32 v8, vcc_lo, s7, v8, vcc_lo global_load_b32 v7, v[7:8], off s_waitcnt vmcnt(0) ds_store_b32 v6, v7 v_add_nc_u32_e32 v6, 8, v6 s_cbranch_scc1 .LBB0_3 v_add_nc_u32_e32 v5, s3, v5 v_add_nc_u32_e32 v2, s4, v2 s_add_i32 s2, s2, 1 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_lg_u32 s2, s8 s_cbranch_scc1 .LBB0_2 .LBB0_5: s_cmp_lt_i32 s9, 2 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_cbranch_scc1 .LBB0_30 v_add_nc_u32_e32 v4, 2, v1 v_mul_lo_u32 v3, v1, s9 s_lshl_b32 s12, s9, 3 s_mov_b32 s4, 0 s_mov_b32 s11, 2 v_mul_lo_u32 v8, s9, v4 s_add_i32 s13, s9, 1 s_add_i32 s14, s12, 8 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_lshl_add_u32 v6, v3, 2, 0 v_lshlrev_b32_e32 v12, 2, v3 v_lshl_add_u32 v8, v8, 2, 0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_dual_mov_b32 v13, v8 :: v_dual_and_b32 v2, 0x3ff, v0 v_mul_lo_u32 v10, v2, s9 v_cmp_gt_u32_e32 vcc_lo, v2, v1 v_cmp_eq_u32_e64 s2, v2, v1 v_cmp_lt_u32_e64 s3, v2, v1 v_min_u32_e32 v4, v1, v2 v_lshlrev_b32_e32 v11, 2, v2 s_xor_b32 s16, vcc_lo, -1 v_lshl_add_u32 v7, v10, 2, 0 v_add_nc_u32_e32 v5, 2, v2 v_lshlrev_b32_e32 v10, 2, v10 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_mov_b32_e32 v14, v7 v_mul_lo_u32 v9, s9, v5 v_lshl_add_u32 v5, v2, 2, 0 s_delay_alu instid0(VALU_DEP_2) v_lshl_add_u32 v9, v9, 2, 0 s_branch .LBB0_8 .LBB0_7: v_add_nc_u32_e32 v6, s14, v6 v_add_nc_u32_e32 v14, s14, v14 v_add_nc_u32_e32 v13, s12, v13 v_add_nc_u32_e32 v7, s12, v7 v_add_nc_u32_e32 v8, s14, v8 v_add_nc_u32_e32 v9, s14, v9 s_add_i32 s11, s11, 1 s_cmp_eq_u32 s4, s8 s_cbranch_scc1 .LBB0_30 .LBB0_8: s_lshl_b32 s18, s4, 1 s_mov_b32 s17, s4 s_mul_i32 s4, s18, s13 s_delay_alu instid0(SALU_CYCLE_1) v_dual_mov_b32 v19, v6 :: v_dual_add_nc_u32 v18, s4, v3 v_mov_b32_e32 v15, v14 s_lshl_b32 s5, s4, 2 s_mov_b32 s19, 0 s_add_i32 s4, s5, 0 v_lshl_add_u32 v18, v18, 2, v5 v_add3_u32 v16, s4, v10, v11 v_add3_u32 v17, s4, v12, v11 s_branch .LBB0_10 .LBB0_9: s_or_b32 exec_lo, exec_lo, s4 v_add_nc_u32_e32 v19, 4, v19 v_add_nc_u32_e32 v15, 4, v15 s_add_i32 s19, s19, 1 s_waitcnt lgkmcnt(0) s_cmp_eq_u32 s19, 1 s_barrier buffer_gl0_inv s_cbranch_scc0 .LBB0_16 .LBB0_10: v_cmp_eq_u32_e32 vcc_lo, s19, v2 s_and_b32 s4, s2, vcc_lo s_delay_alu instid0(SALU_CYCLE_1) s_and_saveexec_b32 s20, s4 s_cbranch_execz .LBB0_12 ds_load_b32 v20, v16 s_waitcnt lgkmcnt(0) v_mul_f32_e32 v21, 0x4f800000, v20 v_cmp_gt_f32_e64 s4, 0xf800000, v20 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cndmask_b32_e64 v20, v20, v21, s4 v_sqrt_f32_e32 v21, v20 s_waitcnt_depctr 0xfff v_add_nc_u32_e32 v22, -1, v21 v_add_nc_u32_e32 v23, 1, v21 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_fma_f32 v24, -v22, v21, v20 v_fma_f32 v25, -v23, v21, v20 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_ge_f32_e64 s5, 0, v24 v_cndmask_b32_e64 v21, v21, v22, s5 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_lt_f32_e64 s5, 0, v25 v_cndmask_b32_e64 v21, v21, v23, s5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f32_e32 v22, 0x37800000, v21 v_cndmask_b32_e64 v21, v21, v22, s4 v_cmp_class_f32_e64 s4, v20, 0x260 s_delay_alu instid0(VALU_DEP_1) v_cndmask_b32_e64 v20, v21, v20, s4 ds_store_b32 v16, v20 .LBB0_12: s_or_b32 exec_lo, exec_lo, s20 s_and_b32 s5, s3, vcc_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_and_saveexec_b32 s4, s5 s_cbranch_execz .LBB0_14 ds_load_b32 v20, v16 ds_load_b32 v21, v17 s_waitcnt lgkmcnt(0) v_div_scale_f32 v22, null, v20, v20, v21 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_rcp_f32_e32 v23, v22 s_waitcnt_depctr 0xfff v_fma_f32 v24, -v22, v23, 1.0 v_fmac_f32_e32 v23, v24, v23 v_div_scale_f32 v24, vcc_lo, v21, v20, v21 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f32_e32 v25, v24, v23 v_fma_f32 v26, -v22, v25, v24 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fmac_f32_e32 v25, v26, v23 v_fma_f32 v22, -v22, v25, v24 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_div_fmas_f32 v22, v22, v23, v25 v_div_fixup_f32 v20, v22, v20, v21 ds_store_b32 v17, v20 .LBB0_14: s_or_b32 exec_lo, exec_lo, s4 v_cmp_lt_u32_e32 vcc_lo, s19, v4 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_and_b32 s5, s16, vcc_lo s_delay_alu instid0(SALU_CYCLE_1) s_and_saveexec_b32 s4, s5 s_cbranch_execz .LBB0_9 ds_load_b32 v20, v15 ds_load_b32 v21, v19 ds_load_b32 v22, v18 s_waitcnt lgkmcnt(0) v_fma_f32 v20, -v20, v21, v22 ds_store_b32 v18, v20 s_branch .LBB0_9 .LBB0_16: s_add_i32 s4, s17, 1 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_ge_i32 s4, s8 s_cbranch_scc1 .LBB0_7 v_dual_mov_b32 v16, v8 :: v_dual_add_nc_u32 v15, s18, v2 v_mov_b32_e32 v18, v13 s_mov_b32 s5, s11 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(VALU_DEP_2) v_mul_lo_u32 v17, v15, s13 v_lshl_add_u32 v15, s18, 2, v5 s_mov_b32 s18, s4 v_lshl_add_u32 v17, v17, 2, 0 s_branch .LBB0_19 .LBB0_18: v_add_nc_u32_e32 v18, s12, v18 v_add_nc_u32_e32 v16, s12, v16 s_add_i32 s18, s18, 1 s_add_i32 s5, s5, 1 s_cmp_ge_i32 s18, s8 s_cbranch_scc1 .LBB0_7 .LBB0_19: v_lshl_add_u32 v19, s18, 1, v1 v_mov_b32_e32 v20, v7 v_mov_b32_e32 v22, v18 s_mov_b32 s19, 0 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_lo_u32 v19, v19, s9 v_lshl_add_u32 v21, v19, 2, v15 s_branch .LBB0_21 .LBB0_20: s_or_b32 exec_lo, exec_lo, s20 v_add_nc_u32_e32 v22, 4, v22 v_add_nc_u32_e32 v20, 4, v20 s_add_i32 s19, s19, 1 s_waitcnt lgkmcnt(0) s_cmp_eq_u32 s19, 1 s_barrier buffer_gl0_inv s_cbranch_scc0 .LBB0_25 .LBB0_21: s_mov_b32 s20, exec_lo v_cmpx_eq_u32_e64 s19, v2 s_cbranch_execz .LBB0_23 ds_load_b32 v23, v17 ds_load_b32 v24, v21 s_waitcnt lgkmcnt(0) v_div_scale_f32 v25, null, v23, v23, v24 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_rcp_f32_e32 v26, v25 s_waitcnt_depctr 0xfff v_fma_f32 v27, -v25, v26, 1.0 v_fmac_f32_e32 v26, v27, v26 v_div_scale_f32 v27, vcc_lo, v24, v23, v24 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f32_e32 v28, v27, v26 v_fma_f32 v29, -v25, v28, v27 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fmac_f32_e32 v28, v29, v26 v_fma_f32 v25, -v25, v28, v27 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_div_fmas_f32 v25, v25, v26, v28 v_div_fixup_f32 v23, v25, v23, v24 ds_store_b32 v21, v23 .LBB0_23: s_or_b32 exec_lo, exec_lo, s20 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s20, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_lt_u32_e64 s19, v2 s_cbranch_execz .LBB0_20 ds_load_b32 v23, v20 ds_load_b32 v24, v22 ds_load_b32 v25, v21 s_waitcnt lgkmcnt(0) v_fma_f32 v23, -v23, v24, v25 ds_store_b32 v21, v23 s_branch .LBB0_20 .LBB0_25: s_cmp_ge_u32 s17, s18 s_cbranch_scc1 .LBB0_18 v_dual_mov_b32 v20, v9 :: v_dual_lshlrev_b32 v19, 2, v19 s_mov_b32 s19, s4 s_delay_alu instid0(VALU_DEP_1) v_add3_u32 v19, 0, v19, v11 .p2align 6 .LBB0_27: v_mov_b32_e32 v21, 0 s_mov_b32 s20, 0 .LBB0_28: s_delay_alu instid0(SALU_CYCLE_1) v_add_nc_u32_e32 v22, s20, v20 v_add_nc_u32_e32 v23, s20, v16 s_add_i32 s20, s20, 4 ds_load_b32 v22, v22 ds_load_b32 v23, v23 s_cmp_eq_u32 s20, 4 s_waitcnt lgkmcnt(0) v_fmac_f32_e32 v21, v22, v23 s_cbranch_scc1 .LBB0_28 v_lshl_add_u32 v22, s19, 3, v19 s_add_i32 s19, s19, 1 v_add_nc_u32_e32 v20, s12, v20 s_cmp_lg_u32 s19, s5 ds_load_b32 v23, v22 s_waitcnt lgkmcnt(0) v_sub_f32_e32 v21, v23, v21 ds_store_b32 v22, v21 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_cbranch_scc1 .LBB0_27 s_branch .LBB0_18 .LBB0_30: s_cmp_lt_i32 s9, 4 s_mov_b32 s4, 4 s_cbranch_scc1 .LBB0_36 v_dual_mov_b32 v3, 0 :: v_dual_and_b32 v2, 0x3ff, v0 s_add_u32 s2, s0, 16 s_addc_u32 s3, s1, 0 s_max_i32 s5, s8, 2 s_delay_alu instid0(VALU_DEP_1) v_lshl_add_u32 v2, v2, 2, 0 s_add_i32 s5, s5, -2 s_mov_b32 s12, 0 s_lshl_b32 s11, s9, 2 s_set_inst_prefetch_distance 0x1 s_branch .LBB0_33 .p2align 6 .LBB0_32: s_add_i32 s4, s4, 4 s_cmp_eq_u32 s13, s5 s_cbranch_scc1 .LBB0_36 .LBB0_33: s_mov_b32 s13, s12 s_add_i32 s12, s12, 1 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_ge_i32 s12, s8 s_cbranch_scc1 .LBB0_32 s_load_b32 s14, s[2:3], 0xc s_waitcnt lgkmcnt(0) s_lshr_b32 s16, s14, 16 s_and_b32 s14, s14, 0xffff v_mad_u64_u32 v[4:5], null, s13, s16, v[1:2] s_mul_i32 s16, s4, s14 s_lshl_b32 s14, s14, 2 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_lo_u32 v4, s11, v4 v_add3_u32 v4, v4, s16, v2 s_mov_b32 s16, s12 .LBB0_35: ds_store_b32 v4, v3 v_add_nc_u32_e32 v4, s14, v4 s_add_i32 s16, s16, 1 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_ge_i32 s16, s8 s_cbranch_scc0 .LBB0_35 s_branch .LBB0_32 .LBB0_36: s_set_inst_prefetch_distance 0x2 s_bitcmp1_b32 s8, 0 v_and_b32_e32 v2, 0x3ff, v0 s_cselect_b32 s2, -1, 0 s_waitcnt lgkmcnt(0) s_and_b32 vcc_lo, exec_lo, s2 s_mov_b32 s2, -1 s_barrier buffer_gl0_inv s_cbranch_vccz .LBB0_47 s_cmp_lt_i32 s9, 4 s_cbranch_scc1 .LBB0_44 v_not_b32_e32 v6, v2 s_add_i32 s3, s8, -1 v_lshl_add_u32 v3, v2, 2, 0 v_cmp_gt_u32_e32 vcc_lo, v2, v1 s_add_u32 s4, s0, 16 v_cmp_lt_u32_e64 s2, v2, v1 v_not_b32_e32 v4, v1 v_dual_mov_b32 v5, 0 :: v_dual_lshlrev_b32 v6, 2, v6 s_addc_u32 s5, s1, 0 s_lshl_b32 s13, s9, 1 s_mov_b32 s11, 0 s_add_i32 s12, s9, 1 s_add_i32 s13, s13, 2 s_set_inst_prefetch_distance 0x1 s_branch .LBB0_40 .p2align 6 .LBB0_39: s_or_b32 exec_lo, exec_lo, s14 s_add_i32 s11, s11, 2 s_add_i32 s12, s12, s13 s_cmp_lt_i32 s11, s3 s_cbranch_scc0 .LBB0_44 .LBB0_40: s_and_saveexec_b32 s14, vcc_lo s_cbranch_execz .LBB0_42 s_load_b32 s16, s[4:5], 0xc s_waitcnt lgkmcnt(0) s_lshr_b32 s17, s16, 16 s_and_b32 s16, s16, 0xffff v_mad_u64_u32 v[7:8], null, s11, s17, v[1:2] s_mul_i32 s16, s11, s16 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) s_lshl_b32 s16, s16, 2 v_mul_lo_u32 v7, v7, s9 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b32_e32 v7, 2, v7 v_add3_u32 v7, v3, s16, v7 ds_store_b32 v7, v5 .LBB0_42: s_or_b32 exec_lo, exec_lo, s14 s_and_saveexec_b32 s14, s2 s_cbranch_execz .LBB0_39 s_load_b32 s16, s[4:5], 0xc s_waitcnt lgkmcnt(0) s_lshr_b32 s17, s16, 16 s_and_b32 s16, s16, 0xffff v_add_nc_u32_e32 v7, s17, v4 s_mul_i32 s17, s12, s16 s_lshl_b32 s16, s16, 2 s_lshl_b32 s17, s17, 2 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) s_add_i32 s17, s17, 0 v_mul_lo_u32 v7, v7, s9 s_add_i32 s17, s17, s16 v_lshlrev_b32_e32 v7, 2, v7 s_delay_alu instid0(VALU_DEP_1) v_add3_u32 v7, s17, v6, v7 ds_store_b32 v7, v5 s_branch .LBB0_39 .LBB0_44: s_set_inst_prefetch_distance 0x2 s_mov_b32 s2, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e64 v2, v1 s_cbranch_execz .LBB0_46 s_load_b32 s3, s[0:1], 0x1c v_mul_lo_u32 v3, v1, s9 s_add_i32 s4, s8, -1 s_add_i32 s5, s9, 1 v_lshlrev_b32_e32 v4, 2, v2 s_mul_i32 s4, s4, s5 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) v_lshlrev_b32_e32 v3, 2, v3 s_waitcnt lgkmcnt(0) s_and_b32 s3, s3, 0xffff s_mul_i32 s4, s4, s3 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_lshl_b32 s3, s4, 2 s_add_i32 s3, s3, 0 s_delay_alu instid0(SALU_CYCLE_1) v_add3_u32 v3, s3, v4, v3 v_mov_b32_e32 v4, 0 ds_store_b32 v3, v4 .LBB0_46: s_or_b32 exec_lo, exec_lo, s2 s_waitcnt lgkmcnt(0) s_mov_b32 s2, 0 s_barrier .LBB0_47: s_and_b32 vcc_lo, exec_lo, s2 s_cbranch_vccz .LBB0_56 s_cmp_lt_i32 s9, 2 s_cbranch_scc1 .LBB0_55 v_not_b32_e32 v5, v2 v_lshl_add_u32 v3, v2, 2, 0 v_cmp_gt_u32_e32 vcc_lo, v2, v1 s_add_u32 s2, s0, 16 v_cmp_lt_u32_e64 s0, v2, v1 v_not_b32_e32 v2, v1 v_dual_mov_b32 v4, 0 :: v_dual_lshlrev_b32 v5, 2, v5 s_addc_u32 s3, s1, 0 s_lshl_b32 s5, s9, 1 s_mov_b32 s4, 0 s_add_i32 s1, s9, 1 s_add_i32 s5, s5, 2 s_set_inst_prefetch_distance 0x1 s_branch .LBB0_51 .p2align 6 .LBB0_50: s_or_b32 exec_lo, exec_lo, s11 s_add_i32 s4, s4, 2 s_add_i32 s1, s1, s5 s_cmp_lt_i32 s4, s8 s_cbranch_scc0 .LBB0_55 .LBB0_51: s_and_saveexec_b32 s11, vcc_lo s_cbranch_execz .LBB0_53 s_load_b32 s12, s[2:3], 0xc s_waitcnt lgkmcnt(0) s_lshr_b32 s13, s12, 16 s_and_b32 s12, s12, 0xffff v_mad_u64_u32 v[6:7], null, s4, s13, v[1:2] s_mul_i32 s12, s4, s12 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_1) s_lshl_b32 s12, s12, 2 v_mul_lo_u32 v6, v6, s9 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b32_e32 v6, 2, v6 v_add3_u32 v6, v3, s12, v6 ds_store_b32 v6, v4 .LBB0_53: s_or_b32 exec_lo, exec_lo, s11 s_and_saveexec_b32 s11, s0 s_cbranch_execz .LBB0_50 s_load_b32 s12, s[2:3], 0xc s_waitcnt lgkmcnt(0) s_lshr_b32 s13, s12, 16 s_and_b32 s12, s12, 0xffff v_add_nc_u32_e32 v6, s13, v2 s_mul_i32 s13, s1, s12 s_lshl_b32 s12, s12, 2 s_lshl_b32 s13, s13, 2 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) s_add_i32 s13, s13, 0 v_mul_lo_u32 v6, v6, s9 s_add_i32 s13, s13, s12 v_lshlrev_b32_e32 v6, 2, v6 s_delay_alu instid0(VALU_DEP_1) v_add3_u32 v6, s13, v5, v6 ds_store_b32 v6, v4 s_branch .LBB0_50 .LBB0_55: s_set_inst_prefetch_distance 0x2 s_waitcnt lgkmcnt(0) s_barrier .LBB0_56: s_and_not1_b32 vcc_lo, exec_lo, s10 buffer_gl0_inv s_cbranch_vccnz .LBB0_61 v_and_b32_e32 v2, 0x3ff, v0 s_lshl_b32 s0, s9, 1 s_mov_b32 s1, 0 s_lshl_b32 s2, s9, 3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u64_u32 v[3:4], null, s15, s9, v[1:2] v_mad_u64_u32 v[4:5], null, v1, s9, v[2:3] v_mad_u64_u32 v[0:1], null, s9, v3, v[2:3] v_mov_b32_e32 v2, 0 s_delay_alu instid0(VALU_DEP_3) v_lshl_add_u32 v3, v4, 2, 0 .p2align 6 .LBB0_58: s_delay_alu instid0(VALU_DEP_1) v_dual_mov_b32 v4, v3 :: v_dual_mov_b32 v1, v0 s_mov_b32 s3, s8 .LBB0_59: ds_load_b32 v7, v4 v_lshlrev_b64 v[5:6], 2, v[1:2] v_add_nc_u32_e32 v1, 2, v1 v_add_nc_u32_e32 v4, 8, v4 s_add_i32 s3, s3, -1 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(VALU_DEP_3) s_cmp_lg_u32 s3, 0 v_add_co_u32 v5, vcc_lo, s6, v5 v_add_co_ci_u32_e32 v6, vcc_lo, s7, v6, vcc_lo s_waitcnt lgkmcnt(0) global_store_b32 v[5:6], v7, off s_cbranch_scc1 .LBB0_59 v_add_nc_u32_e32 v0, s0, v0 v_add_nc_u32_e32 v3, s2, v3 s_add_i32 s1, s1, 1 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_lg_u32 s1, s8 s_cbranch_scc1 .LBB0_58 .LBB0_61: s_waitcnt_vscnt null, 0x0 s_barrier buffer_gl0_inv s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z27right_looking_launch_kernelPfi .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 1 .amdhsa_next_free_vgpr 30 .amdhsa_next_free_sgpr 21 .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 _Z27right_looking_launch_kernelPfi, .Lfunc_end0-_Z27right_looking_launch_kernelPfi .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 - .offset: 136 .size: 4 .value_kind: hidden_dynamic_lds_size .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: _Z27right_looking_launch_kernelPfi .private_segment_fixed_size: 0 .sgpr_count: 23 .sgpr_spill_count: 0 .symbol: _Z27right_looking_launch_kernelPfi.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 30 .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> #include<stdlib.h> #include<cmath> #define TILE_SIZE 2 __device__ void store_full(float*,float*,int); __device__ void load_full(float*,float*,int); __device__ void potrf_tile(float*,int,int); __device__ void trsm_tile(float*,int,int,int); __device__ void syrk_tile(float*,int,int,int,int); __global__ void right_looking_launch_kernel(float*,int); __device__ void store_zeros(float*,int); __device__ void store_zeros_diagonal(float*,int,int); __device__ void store_zeros_last(float*,int); __device__ void store_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID + N*N*blockIdx.x] = read_data[ID]; } } __syncthreads(); } __device__ void load_full(float* read_data,float* write_data,int N) { int i,j,ID; for(i=0;i<N/TILE_SIZE;i++) { for(j=0;j<N/TILE_SIZE;j++) { ID = (i*TILE_SIZE + threadIdx.y)*N + j*TILE_SIZE + threadIdx.x; write_data[ID] = read_data[ID + N*N*blockIdx.x]; } } __syncthreads(); } __device__ void potrf_tile(float* t_A,int i,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int k=0;k<TILE_SIZE;k++) { if(t_x==t_y && t_x==k) { t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x] = sqrtf(t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]); } __syncthreads(); if(t_x<t_y && t_x == k) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]/= t_A[i*TILE_SIZE*(1+N) + t_x*N + t_x]; } __syncthreads(); if(k<t_y && k<t_x && t_x<=t_y) { t_A[i*TILE_SIZE*(1+N) + t_y*N + t_x]-= t_A[i*TILE_SIZE*(1+N) + t_x*N + k]*t_A[i*TILE_SIZE*(1+N) + t_y*N + k]; } __syncthreads(); } } __device__ void trsm_tile(float *row_data,int i,int j,int N) { int t_x = threadIdx.x; int t_y = threadIdx.y; for(int s=0;s<TILE_SIZE;s++) { if(t_x==s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]/= row_data[i*TILE_SIZE*(1+N) + t_x*(1+N)]; } __syncthreads(); if(t_x > s) { row_data[(t_y + j*TILE_SIZE)*N + t_x + i*TILE_SIZE]-= row_data[(t_x + i*TILE_SIZE)*N + s]*row_data[(t_y + j*TILE_SIZE)*N + s]; } __syncthreads(); } } __device__ void syrk_tile(float* row_data,int i,int j,int k,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; float valueToSubtract = 0.0; for(int r=0;r<TILE_SIZE;r++) { valueToSubtract+= row_data[(t_x + k*TILE_SIZE)*N + i*TILE_SIZE + r]*row_data[(t_y + j*TILE_SIZE)*N + i*TILE_SIZE + r]; } row_data[(t_y + j*TILE_SIZE)*N + t_x + k*TILE_SIZE]-= valueToSubtract; __syncthreads(); } __device__ void store_zeros(float* A,int N) { int t_y = threadIdx.y; int t_x = threadIdx.x; int i,j; for(i=0;i<N/TILE_SIZE-1;i++) { for(j=i+1;j<N/TILE_SIZE;j++) A[j*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_diagonal(float* A,int N,int b) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; int i; for(i=0;i<N/TILE_SIZE-b;i+=2) { if(t_x>t_y) A[i*blockDim.x + t_x + (i*blockDim.y + t_y)*N] = 0.0; if(t_x<t_y) A[(i+1)*blockDim.x*(1+N) + (blockDim.x-t_x-1) + (blockDim.y-t_y-1)*N] = 0.0; } __syncthreads(); } __device__ void store_zeros_last(float* A,int N) // Will only work if (N/TILE_SIZE) is even { int t_y = threadIdx.y; int t_x = threadIdx.x; if(t_x>t_y) A[(N/TILE_SIZE-1)*blockDim.x*(1+N) + t_x + t_y*N] = 0.0; __syncthreads(); } __global__ void right_looking_launch_kernel(float* read_data,int N) { extern __shared__ float data[]; int i,j,k; load_full(read_data,data,N); for(i=0;i<N/TILE_SIZE;i++) { potrf_tile(data,i,N); for(j=i+1;j<N/TILE_SIZE;j++) { trsm_tile(data,i,j,N); for(k=i+1;k<=j;k++) { syrk_tile(data,i,j,k,N); } } } store_zeros(data,N); if((N/TILE_SIZE)%2==0) store_zeros_diagonal(data,N,0); else { store_zeros_diagonal(data,N,1); store_zeros_last(data,N); } store_full(data,read_data,N); }
.text .file "right_looking.hip" .globl _Z42__device_stub__right_looking_launch_kernelPfi # -- Begin function _Z42__device_stub__right_looking_launch_kernelPfi .p2align 4, 0x90 .type _Z42__device_stub__right_looking_launch_kernelPfi,@function _Z42__device_stub__right_looking_launch_kernelPfi: # @_Z42__device_stub__right_looking_launch_kernelPfi .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 $_Z27right_looking_launch_kernelPfi, %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 _Z42__device_stub__right_looking_launch_kernelPfi, .Lfunc_end0-_Z42__device_stub__right_looking_launch_kernelPfi .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 $_Z27right_looking_launch_kernelPfi, %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 _Z27right_looking_launch_kernelPfi,@object # @_Z27right_looking_launch_kernelPfi .section .rodata,"a",@progbits .globl _Z27right_looking_launch_kernelPfi .p2align 3, 0x0 _Z27right_looking_launch_kernelPfi: .quad _Z42__device_stub__right_looking_launch_kernelPfi .size _Z27right_looking_launch_kernelPfi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z27right_looking_launch_kernelPfi" .size .L__unnamed_1, 35 .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 _Z42__device_stub__right_looking_launch_kernelPfi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z27right_looking_launch_kernelPfi .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_0015b61b_00000000-6_right_looking.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2067: .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 .LFE2067: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z10store_fullPfS_i .type _Z10store_fullPfS_i, @function _Z10store_fullPfS_i: .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 _Z10store_fullPfS_i, .-_Z10store_fullPfS_i .globl _Z9load_fullPfS_i .type _Z9load_fullPfS_i, @function _Z9load_fullPfS_i: .LFB2058: .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 .LFE2058: .size _Z9load_fullPfS_i, .-_Z9load_fullPfS_i .globl _Z10potrf_tilePfii .type _Z10potrf_tilePfii, @function _Z10potrf_tilePfii: .LFB2059: .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 .LFE2059: .size _Z10potrf_tilePfii, .-_Z10potrf_tilePfii .globl _Z9trsm_tilePfiii .type _Z9trsm_tilePfiii, @function _Z9trsm_tilePfiii: .LFB2060: .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 .LFE2060: .size _Z9trsm_tilePfiii, .-_Z9trsm_tilePfiii .globl _Z9syrk_tilePfiiii .type _Z9syrk_tilePfiiii, @function _Z9syrk_tilePfiiii: .LFB2061: .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 .LFE2061: .size _Z9syrk_tilePfiiii, .-_Z9syrk_tilePfiiii .globl _Z11store_zerosPfi .type _Z11store_zerosPfi, @function _Z11store_zerosPfi: .LFB2062: .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 .LFE2062: .size _Z11store_zerosPfi, .-_Z11store_zerosPfi .globl _Z20store_zeros_diagonalPfii .type _Z20store_zeros_diagonalPfii, @function _Z20store_zeros_diagonalPfii: .LFB2063: .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 .LFE2063: .size _Z20store_zeros_diagonalPfii, .-_Z20store_zeros_diagonalPfii .globl _Z16store_zeros_lastPfi .type _Z16store_zeros_lastPfi, @function _Z16store_zeros_lastPfi: .LFB2064: .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 .LFE2064: .size _Z16store_zeros_lastPfi, .-_Z16store_zeros_lastPfi .globl _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi .type _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi, @function _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi: .LFB2089: .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 .L23 .L19: movq 104(%rsp), %rax subq %fs:40, %rax jne .L24 addq $120, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L23: .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 _Z27right_looking_launch_kernelPfi(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 128 jmp .L19 .L24: call __stack_chk_fail@PLT .cfi_endproc .LFE2089: .size _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi, .-_Z48__device_stub__Z27right_looking_launch_kernelPfiPfi .globl _Z27right_looking_launch_kernelPfi .type _Z27right_looking_launch_kernelPfi, @function _Z27right_looking_launch_kernelPfi: .LFB2090: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z48__device_stub__Z27right_looking_launch_kernelPfiPfi addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2090: .size _Z27right_looking_launch_kernelPfi, .-_Z27right_looking_launch_kernelPfi .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC0: .string "_Z27right_looking_launch_kernelPfi" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2092: .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 _Z27right_looking_launch_kernelPfi(%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 .LFE2092: .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 "right_looking.hip" .globl _Z42__device_stub__right_looking_launch_kernelPfi # -- Begin function _Z42__device_stub__right_looking_launch_kernelPfi .p2align 4, 0x90 .type _Z42__device_stub__right_looking_launch_kernelPfi,@function _Z42__device_stub__right_looking_launch_kernelPfi: # @_Z42__device_stub__right_looking_launch_kernelPfi .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 $_Z27right_looking_launch_kernelPfi, %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 _Z42__device_stub__right_looking_launch_kernelPfi, .Lfunc_end0-_Z42__device_stub__right_looking_launch_kernelPfi .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 $_Z27right_looking_launch_kernelPfi, %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 _Z27right_looking_launch_kernelPfi,@object # @_Z27right_looking_launch_kernelPfi .section .rodata,"a",@progbits .globl _Z27right_looking_launch_kernelPfi .p2align 3, 0x0 _Z27right_looking_launch_kernelPfi: .quad _Z42__device_stub__right_looking_launch_kernelPfi .size _Z27right_looking_launch_kernelPfi, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z27right_looking_launch_kernelPfi" .size .L__unnamed_1, 35 .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 _Z42__device_stub__right_looking_launch_kernelPfi .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z27right_looking_launch_kernelPfi .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 <stdlib.h> #define N 256 __global__ void bitreverse(unsigned int *data){ unsigned int *idata = data; unsigned int x = idata[threadIdx.x]; x = ((0xf0f0f0f0 & x) >> 4) | ((0x0f0f0f0f & x) << 4); x = ((0xcccccccc & x) >> 2) | ((0x33333333 & x) << 2); x = ((0xaaaaaaaa & x) >> 1) | ((0x55555555 & x) << 1); idata[threadIdx.x] = x; } int main(void){ unsigned int *d = NULL; int i; unsigned int idata[N], odata[N]; for (i = 0; i < N; i++) idata[i] = (unsigned int)i; cudaMalloc((void**)&d, sizeof(int)*N); cudaMemcpy(d, idata, sizeof(int)*N,cudaMemcpyHostToDevice); bitreverse<<<1, N>>>(d); cudaMemcpy(odata, d, sizeof(int)*N,cudaMemcpyHostToDevice); for (i = 0; i < N; i++) printf("%u -> %u\n", idata[i], odata[i]); cudaFree((void*)d); return 0; }
code for sm_80 Function : _Z10bitreversePj .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 R2, SR_TID.X ; /* 0x0000000000027919 */ /* 0x000e220000002100 */ /*0020*/ IMAD.MOV.U32 R3, RZ, RZ, 0x4 ; /* 0x00000004ff037424 */ /* 0x000fe200078e00ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fc60000000a00 */ /*0040*/ IMAD.WIDE.U32 R2, R2, R3, c[0x0][0x160] ; /* 0x0000580002027625 */ /* 0x001fca00078e0003 */ /*0050*/ LDG.E R0, [R2.64] ; /* 0x0000000402007981 */ /* 0x000ea4000c1e1900 */ /*0060*/ IMAD.SHL.U32 R4, R0.reuse, 0x10, RZ ; /* 0x0000001000047824 */ /* 0x044fe200078e00ff */ /*0070*/ LOP3.LUT R0, R0, 0xf0f0f0f0, RZ, 0xc0, !PT ; /* 0xf0f0f0f000007812 */ /* 0x000fc800078ec0ff */ /*0080*/ LOP3.LUT R5, R4, 0xf0f0f0f0, RZ, 0xc0, !PT ; /* 0xf0f0f0f004057812 */ /* 0x000fc800078ec0ff */ /*0090*/ LEA.HI R0, R0, R5, RZ, 0x1c ; /* 0x0000000500007211 */ /* 0x000fca00078fe0ff */ /*00a0*/ IMAD.SHL.U32 R4, R0.reuse, 0x4, RZ ; /* 0x0000000400047824 */ /* 0x040fe200078e00ff */ /*00b0*/ LOP3.LUT R0, R0, 0xcccccccc, RZ, 0xc0, !PT ; /* 0xcccccccc00007812 */ /* 0x000fc800078ec0ff */ /*00c0*/ LOP3.LUT R5, R4, 0xcccccccc, RZ, 0xc0, !PT ; /* 0xcccccccc04057812 */ /* 0x000fc800078ec0ff */ /*00d0*/ LEA.HI R0, R0, R5, RZ, 0x1e ; /* 0x0000000500007211 */ /* 0x000fca00078ff0ff */ /*00e0*/ IMAD.SHL.U32 R4, R0.reuse, 0x2, RZ ; /* 0x0000000200047824 */ /* 0x040fe200078e00ff */ /*00f0*/ LOP3.LUT R0, R0, 0xaaaaaaaa, RZ, 0xc0, !PT ; /* 0xaaaaaaaa00007812 */ /* 0x000fc800078ec0ff */ /*0100*/ LOP3.LUT R5, R4, 0xaaaaaaaa, RZ, 0xc0, !PT ; /* 0xaaaaaaaa04057812 */ /* 0x000fc800078ec0ff */ /*0110*/ LEA.HI R5, R0, R5, RZ, 0x1f ; /* 0x0000000500057211 */ /* 0x000fca00078ff8ff */ /*0120*/ STG.E [R2.64], R5 ; /* 0x0000000502007986 */ /* 0x000fe2000c101904 */ /*0130*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0140*/ BRA 0x140; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*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 <stdlib.h> #define N 256 __global__ void bitreverse(unsigned int *data){ unsigned int *idata = data; unsigned int x = idata[threadIdx.x]; x = ((0xf0f0f0f0 & x) >> 4) | ((0x0f0f0f0f & x) << 4); x = ((0xcccccccc & x) >> 2) | ((0x33333333 & x) << 2); x = ((0xaaaaaaaa & x) >> 1) | ((0x55555555 & x) << 1); idata[threadIdx.x] = x; } int main(void){ unsigned int *d = NULL; int i; unsigned int idata[N], odata[N]; for (i = 0; i < N; i++) idata[i] = (unsigned int)i; cudaMalloc((void**)&d, sizeof(int)*N); cudaMemcpy(d, idata, sizeof(int)*N,cudaMemcpyHostToDevice); bitreverse<<<1, N>>>(d); cudaMemcpy(odata, d, sizeof(int)*N,cudaMemcpyHostToDevice); for (i = 0; i < N; i++) printf("%u -> %u\n", idata[i], odata[i]); cudaFree((void*)d); return 0; }
.file "tmpxft_000bcc46_00000000-6_bitreverse.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 _Z30__device_stub__Z10bitreversePjPj .type _Z30__device_stub__Z10bitreversePjPj, @function _Z30__device_stub__Z10bitreversePjPj: .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 _Z10bitreversePj(%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 _Z30__device_stub__Z10bitreversePjPj, .-_Z30__device_stub__Z10bitreversePjPj .globl _Z10bitreversePj .type _Z10bitreversePj, @function _Z10bitreversePj: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z30__device_stub__Z10bitreversePjPj addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z10bitreversePj, .-_Z10bitreversePj .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "%u -> %u\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 $2104, %rsp .cfi_def_cfa_offset 2128 movq %fs:40, %rax movq %rax, 2088(%rsp) xorl %eax, %eax movq $0, (%rsp) .L12: movl %eax, 32(%rsp,%rax,4) addq $1, %rax cmpq $256, %rax jne .L12 movq %rsp, %rdi movl $1024, %esi call cudaMalloc@PLT leaq 32(%rsp), %rsi movl $1, %ecx movl $1024, %edx movq (%rsp), %rdi call cudaMemcpy@PLT movl $256, 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 .L19 .L13: leaq 1056(%rsp), %rdi movl $1, %ecx movl $1024, %edx movq (%rsp), %rsi call cudaMemcpy@PLT movl $0, %ebx leaq .LC0(%rip), %rbp .L14: movl 1056(%rsp,%rbx), %ecx movl 32(%rsp,%rbx), %edx movq %rbp, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addq $4, %rbx cmpq $1024, %rbx jne .L14 movq (%rsp), %rdi call cudaFree@PLT movq 2088(%rsp), %rax subq %fs:40, %rax jne .L20 movl $0, %eax addq $2104, %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 (%rsp), %rdi call _Z30__device_stub__Z10bitreversePjPj jmp .L13 .L20: call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1 .LC1: .string "_Z10bitreversePj" .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 .LC1(%rip), %rdx movq %rdx, %rcx leaq _Z10bitreversePj(%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> #include <stdlib.h> #define N 256 __global__ void bitreverse(unsigned int *data){ unsigned int *idata = data; unsigned int x = idata[threadIdx.x]; x = ((0xf0f0f0f0 & x) >> 4) | ((0x0f0f0f0f & x) << 4); x = ((0xcccccccc & x) >> 2) | ((0x33333333 & x) << 2); x = ((0xaaaaaaaa & x) >> 1) | ((0x55555555 & x) << 1); idata[threadIdx.x] = x; } int main(void){ unsigned int *d = NULL; int i; unsigned int idata[N], odata[N]; for (i = 0; i < N; i++) idata[i] = (unsigned int)i; cudaMalloc((void**)&d, sizeof(int)*N); cudaMemcpy(d, idata, sizeof(int)*N,cudaMemcpyHostToDevice); bitreverse<<<1, N>>>(d); cudaMemcpy(odata, d, sizeof(int)*N,cudaMemcpyHostToDevice); for (i = 0; i < N; i++) printf("%u -> %u\n", idata[i], odata[i]); cudaFree((void*)d); return 0; }
#include <hip/hip_runtime.h> #include <stdio.h> #include <stdlib.h> #define N 256 __global__ void bitreverse(unsigned int *data){ unsigned int *idata = data; unsigned int x = idata[threadIdx.x]; x = ((0xf0f0f0f0 & x) >> 4) | ((0x0f0f0f0f & x) << 4); x = ((0xcccccccc & x) >> 2) | ((0x33333333 & x) << 2); x = ((0xaaaaaaaa & x) >> 1) | ((0x55555555 & x) << 1); idata[threadIdx.x] = x; } int main(void){ unsigned int *d = NULL; int i; unsigned int idata[N], odata[N]; for (i = 0; i < N; i++) idata[i] = (unsigned int)i; hipMalloc((void**)&d, sizeof(int)*N); hipMemcpy(d, idata, sizeof(int)*N,hipMemcpyHostToDevice); bitreverse<<<1, N>>>(d); hipMemcpy(odata, d, sizeof(int)*N,hipMemcpyHostToDevice); for (i = 0; i < N; i++) printf("%u -> %u\n", idata[i], odata[i]); hipFree((void*)d); 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> #include <stdlib.h> #define N 256 __global__ void bitreverse(unsigned int *data){ unsigned int *idata = data; unsigned int x = idata[threadIdx.x]; x = ((0xf0f0f0f0 & x) >> 4) | ((0x0f0f0f0f & x) << 4); x = ((0xcccccccc & x) >> 2) | ((0x33333333 & x) << 2); x = ((0xaaaaaaaa & x) >> 1) | ((0x55555555 & x) << 1); idata[threadIdx.x] = x; } int main(void){ unsigned int *d = NULL; int i; unsigned int idata[N], odata[N]; for (i = 0; i < N; i++) idata[i] = (unsigned int)i; hipMalloc((void**)&d, sizeof(int)*N); hipMemcpy(d, idata, sizeof(int)*N,hipMemcpyHostToDevice); bitreverse<<<1, N>>>(d); hipMemcpy(odata, d, sizeof(int)*N,hipMemcpyHostToDevice); for (i = 0; i < N; i++) printf("%u -> %u\n", idata[i], odata[i]); hipFree((void*)d); return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z10bitreversePj .globl _Z10bitreversePj .p2align 8 .type _Z10bitreversePj,@function _Z10bitreversePj: s_load_b64 s[0:1], s[0:1], 0x0 v_lshlrev_b32_e32 v0, 2, v0 s_waitcnt lgkmcnt(0) global_load_b32 v1, v0, s[0:1] s_waitcnt vmcnt(0) v_lshrrev_b32_e32 v2, 4, v1 v_lshlrev_b32_e32 v1, 4, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_bfi_b32 v1, 0xf0f0f0f, v2, v1 v_lshrrev_b32_e32 v2, 2, v1 v_lshlrev_b32_e32 v1, 2, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_bfi_b32 v1, 0x33333333, v2, v1 v_lshrrev_b32_e32 v2, 1, v1 v_lshlrev_b32_e32 v1, 1, v1 s_delay_alu instid0(VALU_DEP_1) v_bfi_b32 v1, 0x55555555, v2, 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 _Z10bitreversePj .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 8 .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 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 _Z10bitreversePj, .Lfunc_end0-_Z10bitreversePj .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: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 8 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z10bitreversePj .private_segment_fixed_size: 0 .sgpr_count: 2 .sgpr_spill_count: 0 .symbol: _Z10bitreversePj.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 <stdio.h> #include <stdlib.h> #define N 256 __global__ void bitreverse(unsigned int *data){ unsigned int *idata = data; unsigned int x = idata[threadIdx.x]; x = ((0xf0f0f0f0 & x) >> 4) | ((0x0f0f0f0f & x) << 4); x = ((0xcccccccc & x) >> 2) | ((0x33333333 & x) << 2); x = ((0xaaaaaaaa & x) >> 1) | ((0x55555555 & x) << 1); idata[threadIdx.x] = x; } int main(void){ unsigned int *d = NULL; int i; unsigned int idata[N], odata[N]; for (i = 0; i < N; i++) idata[i] = (unsigned int)i; hipMalloc((void**)&d, sizeof(int)*N); hipMemcpy(d, idata, sizeof(int)*N,hipMemcpyHostToDevice); bitreverse<<<1, N>>>(d); hipMemcpy(odata, d, sizeof(int)*N,hipMemcpyHostToDevice); for (i = 0; i < N; i++) printf("%u -> %u\n", idata[i], odata[i]); hipFree((void*)d); return 0; }
.text .file "bitreverse.hip" .globl _Z25__device_stub__bitreversePj # -- Begin function _Z25__device_stub__bitreversePj .p2align 4, 0x90 .type _Z25__device_stub__bitreversePj,@function _Z25__device_stub__bitreversePj: # @_Z25__device_stub__bitreversePj .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 $_Z10bitreversePj, %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 _Z25__device_stub__bitreversePj, .Lfunc_end0-_Z25__device_stub__bitreversePj .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 $2112, %rsp # imm = 0x840 .cfi_def_cfa_offset 2128 .cfi_offset %rbx, -16 movq $0, 8(%rsp) xorl %eax, %eax .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 movl %eax, 1088(%rsp,%rax,4) incq %rax cmpq $256, %rax # imm = 0x100 jne .LBB1_1 # %bb.2: leaq 8(%rsp), %rdi movl $1024, %esi # imm = 0x400 callq hipMalloc movq 8(%rsp), %rdi leaq 1088(%rsp), %rsi movl $1024, %edx # imm = 0x400 movl $1, %ecx callq hipMemcpy movabsq $4294967297, %rdi # imm = 0x100000001 leaq 255(%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, 56(%rsp) leaq 56(%rsp), %rax movq %rax, 16(%rsp) leaq 64(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 64(%rsp), %rsi movl 72(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 16(%rsp), %r9 movl $_Z10bitreversePj, %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), %rsi leaq 64(%rsp), %rdi movl $1024, %edx # imm = 0x400 movl $1, %ecx callq hipMemcpy xorl %ebx, %ebx .p2align 4, 0x90 .LBB1_5: # =>This Inner Loop Header: Depth=1 movl 1088(%rsp,%rbx,4), %esi movl 64(%rsp,%rbx,4), %edx movl $.L.str, %edi xorl %eax, %eax callq printf incq %rbx cmpq $256, %rbx # imm = 0x100 jne .LBB1_5 # %bb.6: movq 8(%rsp), %rdi callq hipFree xorl %eax, %eax addq $2112, %rsp # imm = 0x840 .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 $_Z10bitreversePj, %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 _Z10bitreversePj,@object # @_Z10bitreversePj .section .rodata,"a",@progbits .globl _Z10bitreversePj .p2align 3, 0x0 _Z10bitreversePj: .quad _Z25__device_stub__bitreversePj .size _Z10bitreversePj, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "%u -> %u\n" .size .L.str, 10 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z10bitreversePj" .size .L__unnamed_1, 17 .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__bitreversePj .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z10bitreversePj .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 : _Z10bitreversePj .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 R2, SR_TID.X ; /* 0x0000000000027919 */ /* 0x000e220000002100 */ /*0020*/ IMAD.MOV.U32 R3, RZ, RZ, 0x4 ; /* 0x00000004ff037424 */ /* 0x000fe200078e00ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fc60000000a00 */ /*0040*/ IMAD.WIDE.U32 R2, R2, R3, c[0x0][0x160] ; /* 0x0000580002027625 */ /* 0x001fca00078e0003 */ /*0050*/ LDG.E R0, [R2.64] ; /* 0x0000000402007981 */ /* 0x000ea4000c1e1900 */ /*0060*/ IMAD.SHL.U32 R4, R0.reuse, 0x10, RZ ; /* 0x0000001000047824 */ /* 0x044fe200078e00ff */ /*0070*/ LOP3.LUT R0, R0, 0xf0f0f0f0, RZ, 0xc0, !PT ; /* 0xf0f0f0f000007812 */ /* 0x000fc800078ec0ff */ /*0080*/ LOP3.LUT R5, R4, 0xf0f0f0f0, RZ, 0xc0, !PT ; /* 0xf0f0f0f004057812 */ /* 0x000fc800078ec0ff */ /*0090*/ LEA.HI R0, R0, R5, RZ, 0x1c ; /* 0x0000000500007211 */ /* 0x000fca00078fe0ff */ /*00a0*/ IMAD.SHL.U32 R4, R0.reuse, 0x4, RZ ; /* 0x0000000400047824 */ /* 0x040fe200078e00ff */ /*00b0*/ LOP3.LUT R0, R0, 0xcccccccc, RZ, 0xc0, !PT ; /* 0xcccccccc00007812 */ /* 0x000fc800078ec0ff */ /*00c0*/ LOP3.LUT R5, R4, 0xcccccccc, RZ, 0xc0, !PT ; /* 0xcccccccc04057812 */ /* 0x000fc800078ec0ff */ /*00d0*/ LEA.HI R0, R0, R5, RZ, 0x1e ; /* 0x0000000500007211 */ /* 0x000fca00078ff0ff */ /*00e0*/ IMAD.SHL.U32 R4, R0.reuse, 0x2, RZ ; /* 0x0000000200047824 */ /* 0x040fe200078e00ff */ /*00f0*/ LOP3.LUT R0, R0, 0xaaaaaaaa, RZ, 0xc0, !PT ; /* 0xaaaaaaaa00007812 */ /* 0x000fc800078ec0ff */ /*0100*/ LOP3.LUT R5, R4, 0xaaaaaaaa, RZ, 0xc0, !PT ; /* 0xaaaaaaaa04057812 */ /* 0x000fc800078ec0ff */ /*0110*/ LEA.HI R5, R0, R5, RZ, 0x1f ; /* 0x0000000500057211 */ /* 0x000fca00078ff8ff */ /*0120*/ STG.E [R2.64], R5 ; /* 0x0000000502007986 */ /* 0x000fe2000c101904 */ /*0130*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0140*/ BRA 0x140; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*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 _Z10bitreversePj .globl _Z10bitreversePj .p2align 8 .type _Z10bitreversePj,@function _Z10bitreversePj: s_load_b64 s[0:1], s[0:1], 0x0 v_lshlrev_b32_e32 v0, 2, v0 s_waitcnt lgkmcnt(0) global_load_b32 v1, v0, s[0:1] s_waitcnt vmcnt(0) v_lshrrev_b32_e32 v2, 4, v1 v_lshlrev_b32_e32 v1, 4, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_bfi_b32 v1, 0xf0f0f0f, v2, v1 v_lshrrev_b32_e32 v2, 2, v1 v_lshlrev_b32_e32 v1, 2, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_bfi_b32 v1, 0x33333333, v2, v1 v_lshrrev_b32_e32 v2, 1, v1 v_lshlrev_b32_e32 v1, 1, v1 s_delay_alu instid0(VALU_DEP_1) v_bfi_b32 v1, 0x55555555, v2, 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 _Z10bitreversePj .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 8 .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 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 _Z10bitreversePj, .Lfunc_end0-_Z10bitreversePj .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: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 8 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z10bitreversePj .private_segment_fixed_size: 0 .sgpr_count: 2 .sgpr_spill_count: 0 .symbol: _Z10bitreversePj.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_000bcc46_00000000-6_bitreverse.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 _Z30__device_stub__Z10bitreversePjPj .type _Z30__device_stub__Z10bitreversePjPj, @function _Z30__device_stub__Z10bitreversePjPj: .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 _Z10bitreversePj(%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 _Z30__device_stub__Z10bitreversePjPj, .-_Z30__device_stub__Z10bitreversePjPj .globl _Z10bitreversePj .type _Z10bitreversePj, @function _Z10bitreversePj: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z30__device_stub__Z10bitreversePjPj addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z10bitreversePj, .-_Z10bitreversePj .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "%u -> %u\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 $2104, %rsp .cfi_def_cfa_offset 2128 movq %fs:40, %rax movq %rax, 2088(%rsp) xorl %eax, %eax movq $0, (%rsp) .L12: movl %eax, 32(%rsp,%rax,4) addq $1, %rax cmpq $256, %rax jne .L12 movq %rsp, %rdi movl $1024, %esi call cudaMalloc@PLT leaq 32(%rsp), %rsi movl $1, %ecx movl $1024, %edx movq (%rsp), %rdi call cudaMemcpy@PLT movl $256, 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 .L19 .L13: leaq 1056(%rsp), %rdi movl $1, %ecx movl $1024, %edx movq (%rsp), %rsi call cudaMemcpy@PLT movl $0, %ebx leaq .LC0(%rip), %rbp .L14: movl 1056(%rsp,%rbx), %ecx movl 32(%rsp,%rbx), %edx movq %rbp, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addq $4, %rbx cmpq $1024, %rbx jne .L14 movq (%rsp), %rdi call cudaFree@PLT movq 2088(%rsp), %rax subq %fs:40, %rax jne .L20 movl $0, %eax addq $2104, %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 (%rsp), %rdi call _Z30__device_stub__Z10bitreversePjPj jmp .L13 .L20: call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1 .LC1: .string "_Z10bitreversePj" .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 .LC1(%rip), %rdx movq %rdx, %rcx leaq _Z10bitreversePj(%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 "bitreverse.hip" .globl _Z25__device_stub__bitreversePj # -- Begin function _Z25__device_stub__bitreversePj .p2align 4, 0x90 .type _Z25__device_stub__bitreversePj,@function _Z25__device_stub__bitreversePj: # @_Z25__device_stub__bitreversePj .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 $_Z10bitreversePj, %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 _Z25__device_stub__bitreversePj, .Lfunc_end0-_Z25__device_stub__bitreversePj .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 $2112, %rsp # imm = 0x840 .cfi_def_cfa_offset 2128 .cfi_offset %rbx, -16 movq $0, 8(%rsp) xorl %eax, %eax .p2align 4, 0x90 .LBB1_1: # =>This Inner Loop Header: Depth=1 movl %eax, 1088(%rsp,%rax,4) incq %rax cmpq $256, %rax # imm = 0x100 jne .LBB1_1 # %bb.2: leaq 8(%rsp), %rdi movl $1024, %esi # imm = 0x400 callq hipMalloc movq 8(%rsp), %rdi leaq 1088(%rsp), %rsi movl $1024, %edx # imm = 0x400 movl $1, %ecx callq hipMemcpy movabsq $4294967297, %rdi # imm = 0x100000001 leaq 255(%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, 56(%rsp) leaq 56(%rsp), %rax movq %rax, 16(%rsp) leaq 64(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration movq 64(%rsp), %rsi movl 72(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d leaq 16(%rsp), %r9 movl $_Z10bitreversePj, %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), %rsi leaq 64(%rsp), %rdi movl $1024, %edx # imm = 0x400 movl $1, %ecx callq hipMemcpy xorl %ebx, %ebx .p2align 4, 0x90 .LBB1_5: # =>This Inner Loop Header: Depth=1 movl 1088(%rsp,%rbx,4), %esi movl 64(%rsp,%rbx,4), %edx movl $.L.str, %edi xorl %eax, %eax callq printf incq %rbx cmpq $256, %rbx # imm = 0x100 jne .LBB1_5 # %bb.6: movq 8(%rsp), %rdi callq hipFree xorl %eax, %eax addq $2112, %rsp # imm = 0x840 .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 $_Z10bitreversePj, %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 _Z10bitreversePj,@object # @_Z10bitreversePj .section .rodata,"a",@progbits .globl _Z10bitreversePj .p2align 3, 0x0 _Z10bitreversePj: .quad _Z25__device_stub__bitreversePj .size _Z10bitreversePj, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "%u -> %u\n" .size .L.str, 10 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z10bitreversePj" .size .L__unnamed_1, 17 .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__bitreversePj .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z10bitreversePj .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 <device_launch_parameters.h> #include <cuda_runtime_api.h> #include <cstdio> // Device input vectors int *d_a; //Device output vector int *d_b; __device__ int mod(int a, int b) { return a >= 0 ? a%b : ( b - abs ( a%b ) ) % b; } __global__ void update(int *A, int *B, int height, int width) { const int index = blockIdx.x * blockDim.x + threadIdx.x; if (index < height*width) { int aliveNeighbours = 0; const int h = index / width; const int w = index % width; /*The neighbours of the cell are checked*/ for (int i = -1; i < 2; ++i) { for (int j = -1; j < 2; ++j) { aliveNeighbours += ( (i | j) && A[mod(h+i,height)*width + mod(w+j,width)]); } } B[index] = (A[index] == 0 && (aliveNeighbours == 3 || aliveNeighbours == 6)) || A[index] == 1 && (aliveNeighbours == 2 || aliveNeighbours == 3); } } void setUp(int height, int width) { // Allocate memory for each vector on GPU cudaMalloc((void **) &d_a, height*width*sizeof(int)); cudaMalloc((void **) &d_b, height*width*sizeof(int)); } void destroy() { // Release device memory cudaFree(d_a); cudaFree(d_b); } void updateCuda(int *A, int *B, int height, int width) { // Size, in bytes, of each vector size_t bytes = height*width*sizeof(int); // Copy host vectors to device cudaMemcpy(d_a, A, bytes, cudaMemcpyHostToDevice); int blockSize, gridSize, n; // // Tamaño de la matriz. // n = height*width; // // // Tamaño del bloque. Elegir entre 32 y 31. // //blockSize = 32; // blockSize = 32; // // // Number of thread blocks in grid // gridSize = (int)ceil((float)n/blockSize); // Execute the kernel update<<< width*height, 1 >>>(d_a, d_b, height, width); // Copy array back to host cudaMemcpy( B, d_b, bytes, cudaMemcpyDeviceToHost ); }
.file "tmpxft_000207cd_00000000-6_kernel.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2063: .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 .LFE2063: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z3modii .type _Z3modii, @function _Z3modii: .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 _Z3modii, .-_Z3modii .globl _Z5setUpii .type _Z5setUpii, @function _Z5setUpii: .LFB2058: .cfi_startproc endbr64 pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset 3, -16 imull %esi, %edi movslq %edi, %rbx salq $2, %rbx movq %rbx, %rsi leaq d_a(%rip), %rdi call cudaMalloc@PLT movq %rbx, %rsi leaq d_b(%rip), %rdi call cudaMalloc@PLT popq %rbx .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2058: .size _Z5setUpii, .-_Z5setUpii .globl _Z7destroyv .type _Z7destroyv, @function _Z7destroyv: .LFB2059: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq d_a(%rip), %rdi call cudaFree@PLT movq d_b(%rip), %rdi call cudaFree@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2059: .size _Z7destroyv, .-_Z7destroyv .globl _Z29__device_stub__Z6updatePiS_iiPiS_ii .type _Z29__device_stub__Z6updatePiS_iiPiS_ii, @function _Z29__device_stub__Z6updatePiS_iiPiS_ii: .LFB2085: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%rsp) movl %ecx, 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 .L13 .L9: movq 136(%rsp), %rax subq %fs:40, %rax jne .L14 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L13: .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 _Z6updatePiS_ii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L9 .L14: call __stack_chk_fail@PLT .cfi_endproc .LFE2085: .size _Z29__device_stub__Z6updatePiS_iiPiS_ii, .-_Z29__device_stub__Z6updatePiS_iiPiS_ii .globl _Z6updatePiS_ii .type _Z6updatePiS_ii, @function _Z6updatePiS_ii: .LFB2086: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z6updatePiS_iiPiS_ii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2086: .size _Z6updatePiS_ii, .-_Z6updatePiS_ii .globl _Z10updateCudaPiS_ii .type _Z10updateCudaPiS_ii, @function _Z10updateCudaPiS_ii: .LFB2060: .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 $32, %rsp .cfi_def_cfa_offset 80 movq %rsi, %r13 movl %edx, %ebp movl %ecx, %r12d movl %edx, %r14d imull %ecx, %r14d movslq %r14d, %rbx salq $2, %rbx movl $1, %ecx movq %rbx, %rdx movq %rdi, %rsi movq d_a(%rip), %rdi call cudaMemcpy@PLT movl $1, 20(%rsp) movl $1, 24(%rsp) movl %r14d, 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 .L20 .L18: movl $2, %ecx movq %rbx, %rdx movq d_b(%rip), %rsi movq %r13, %rdi call cudaMemcpy@PLT addq $32, %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 .L20: .cfi_restore_state movl %r12d, %ecx movl %ebp, %edx movq d_b(%rip), %rsi movq d_a(%rip), %rdi call _Z29__device_stub__Z6updatePiS_iiPiS_ii jmp .L18 .cfi_endproc .LFE2060: .size _Z10updateCudaPiS_ii, .-_Z10updateCudaPiS_ii .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z6updatePiS_ii" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2088: .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 _Z6updatePiS_ii(%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 .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 .globl d_b .bss .align 8 .type d_b, @object .size d_b, 8 d_b: .zero 8 .globl d_a .align 8 .type d_a, @object .size d_a, 8 d_a: .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 <device_launch_parameters.h> #include <cuda_runtime_api.h> #include <cstdio> // Device input vectors int *d_a; //Device output vector int *d_b; __device__ int mod(int a, int b) { return a >= 0 ? a%b : ( b - abs ( a%b ) ) % b; } __global__ void update(int *A, int *B, int height, int width) { const int index = blockIdx.x * blockDim.x + threadIdx.x; if (index < height*width) { int aliveNeighbours = 0; const int h = index / width; const int w = index % width; /*The neighbours of the cell are checked*/ for (int i = -1; i < 2; ++i) { for (int j = -1; j < 2; ++j) { aliveNeighbours += ( (i | j) && A[mod(h+i,height)*width + mod(w+j,width)]); } } B[index] = (A[index] == 0 && (aliveNeighbours == 3 || aliveNeighbours == 6)) || A[index] == 1 && (aliveNeighbours == 2 || aliveNeighbours == 3); } } void setUp(int height, int width) { // Allocate memory for each vector on GPU cudaMalloc((void **) &d_a, height*width*sizeof(int)); cudaMalloc((void **) &d_b, height*width*sizeof(int)); } void destroy() { // Release device memory cudaFree(d_a); cudaFree(d_b); } void updateCuda(int *A, int *B, int height, int width) { // Size, in bytes, of each vector size_t bytes = height*width*sizeof(int); // Copy host vectors to device cudaMemcpy(d_a, A, bytes, cudaMemcpyHostToDevice); int blockSize, gridSize, n; // // Tamaño de la matriz. // n = height*width; // // // Tamaño del bloque. Elegir entre 32 y 31. // //blockSize = 32; // blockSize = 32; // // // Number of thread blocks in grid // gridSize = (int)ceil((float)n/blockSize); // Execute the kernel update<<< width*height, 1 >>>(d_a, d_b, height, width); // Copy array back to host cudaMemcpy( B, d_b, bytes, cudaMemcpyDeviceToHost ); }
#include <hip/hip_runtime.h> #include <hip/hip_runtime_api.h> #include <cstdio> // Device input vectors int *d_a; //Device output vector int *d_b; __device__ int mod(int a, int b) { return a >= 0 ? a%b : ( b - abs ( a%b ) ) % b; } __global__ void update(int *A, int *B, int height, int width) { const int index = blockIdx.x * blockDim.x + threadIdx.x; if (index < height*width) { int aliveNeighbours = 0; const int h = index / width; const int w = index % width; /*The neighbours of the cell are checked*/ for (int i = -1; i < 2; ++i) { for (int j = -1; j < 2; ++j) { aliveNeighbours += ( (i | j) && A[mod(h+i,height)*width + mod(w+j,width)]); } } B[index] = (A[index] == 0 && (aliveNeighbours == 3 || aliveNeighbours == 6)) || A[index] == 1 && (aliveNeighbours == 2 || aliveNeighbours == 3); } } void setUp(int height, int width) { // Allocate memory for each vector on GPU hipMalloc((void **) &d_a, height*width*sizeof(int)); hipMalloc((void **) &d_b, height*width*sizeof(int)); } void destroy() { // Release device memory hipFree(d_a); hipFree(d_b); } void updateCuda(int *A, int *B, int height, int width) { // Size, in bytes, of each vector size_t bytes = height*width*sizeof(int); // Copy host vectors to device hipMemcpy(d_a, A, bytes, hipMemcpyHostToDevice); int blockSize, gridSize, n; // // Tamaño de la matriz. // n = height*width; // // // Tamaño del bloque. Elegir entre 32 y 31. // //blockSize = 32; // blockSize = 32; // // // Number of thread blocks in grid // gridSize = (int)ceil((float)n/blockSize); // Execute the kernel update<<< width*height, 1 >>>(d_a, d_b, height, width); // Copy array back to host hipMemcpy( B, d_b, bytes, hipMemcpyDeviceToHost ); }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> #include <hip/hip_runtime_api.h> #include <cstdio> // Device input vectors int *d_a; //Device output vector int *d_b; __device__ int mod(int a, int b) { return a >= 0 ? a%b : ( b - abs ( a%b ) ) % b; } __global__ void update(int *A, int *B, int height, int width) { const int index = blockIdx.x * blockDim.x + threadIdx.x; if (index < height*width) { int aliveNeighbours = 0; const int h = index / width; const int w = index % width; /*The neighbours of the cell are checked*/ for (int i = -1; i < 2; ++i) { for (int j = -1; j < 2; ++j) { aliveNeighbours += ( (i | j) && A[mod(h+i,height)*width + mod(w+j,width)]); } } B[index] = (A[index] == 0 && (aliveNeighbours == 3 || aliveNeighbours == 6)) || A[index] == 1 && (aliveNeighbours == 2 || aliveNeighbours == 3); } } void setUp(int height, int width) { // Allocate memory for each vector on GPU hipMalloc((void **) &d_a, height*width*sizeof(int)); hipMalloc((void **) &d_b, height*width*sizeof(int)); } void destroy() { // Release device memory hipFree(d_a); hipFree(d_b); } void updateCuda(int *A, int *B, int height, int width) { // Size, in bytes, of each vector size_t bytes = height*width*sizeof(int); // Copy host vectors to device hipMemcpy(d_a, A, bytes, hipMemcpyHostToDevice); int blockSize, gridSize, n; // // Tamaño de la matriz. // n = height*width; // // // Tamaño del bloque. Elegir entre 32 y 31. // //blockSize = 32; // blockSize = 32; // // // Number of thread blocks in grid // gridSize = (int)ceil((float)n/blockSize); // Execute the kernel update<<< width*height, 1 >>>(d_a, d_b, height, width); // Copy array back to host hipMemcpy( B, d_b, bytes, hipMemcpyDeviceToHost ); }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z6updatePiS_ii .globl _Z6updatePiS_ii .p2align 8 .type _Z6updatePiS_ii,@function _Z6updatePiS_ii: s_clause 0x1 s_load_b32 s2, s[0:1], 0x24 s_load_b64 s[4:5], s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_and_b32 s2, s2, 0xffff s_delay_alu instid0(SALU_CYCLE_1) v_mad_u64_u32 v[1:2], null, s15, s2, v[0:1] s_mul_i32 s2, s5, s4 s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) v_cmp_gt_i32_e32 vcc_lo, s2, v1 s_and_saveexec_b32 s2, vcc_lo s_cbranch_execz .LBB0_27 s_ashr_i32 s2, s5, 31 v_ashrrev_i32_e32 v3, 31, v1 s_add_i32 s3, s5, s2 s_mov_b32 s9, -1 s_xor_b32 s3, s3, s2 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v4, v1, v3 v_cvt_f32_u32_e32 v0, s3 s_sub_i32 s6, 0, s3 v_xor_b32_e32 v4, v4, v3 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_rcp_iflag_f32_e32 v0, v0 v_xor_b32_e32 v3, s2, v3 s_waitcnt_depctr 0xfff v_mul_f32_e32 v0, 0x4f7ffffe, v0 v_cvt_u32_f32_e32 v0, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_mul_lo_u32 v2, s6, v0 s_ashr_i32 s6, s4, 31 s_add_i32 s7, s4, s6 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_2) s_xor_b32 s8, s7, s6 s_load_b64 s[6:7], s[0:1], 0x0 v_cvt_f32_u32_e32 v6, s8 v_mul_hi_u32 v2, v0, v2 s_sub_i32 s2, 0, s8 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_rcp_iflag_f32_e32 v6, v6 v_add_nc_u32_e32 v2, v0, v2 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_1) v_mul_hi_u32 v0, v4, v2 s_waitcnt_depctr 0xfff v_mul_f32_e32 v6, 0x4f7ffffe, v6 v_mul_lo_u32 v5, v0, s3 v_sub_nc_u32_e32 v4, v4, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_subrev_nc_u32_e32 v7, s3, v4 v_cmp_le_u32_e32 vcc_lo, s3, v4 v_dual_cndmask_b32 v4, v4, v7 :: v_dual_add_nc_u32 v5, 1, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_cndmask_b32_e32 v0, v0, v5, vcc_lo v_cmp_le_u32_e32 vcc_lo, s3, v4 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_nc_u32_e32 v5, 1, v0 v_cndmask_b32_e32 v0, v0, v5, vcc_lo v_cvt_u32_f32_e32 v5, v6 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_xor_b32_e32 v0, v0, v3 v_mul_lo_u32 v4, s2, v5 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_3) v_sub_nc_u32_e32 v3, v0, v3 v_mov_b32_e32 v0, 0 v_mul_hi_u32 v7, v5, v4 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_lo_u32 v6, v3, s5 v_add_nc_u32_e32 v5, v5, v7 s_delay_alu instid0(VALU_DEP_2) v_sub_nc_u32_e32 v4, v1, v6 s_branch .LBB0_3 .LBB0_2: s_add_i32 s9, s9, 1 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_eq_u32 s9, 2 s_cbranch_scc1 .LBB0_12 .LBB0_3: v_add_nc_u32_e32 v6, s9, v3 s_mov_b32 s10, -1 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v7, 31, v6 v_cmp_gt_i32_e32 vcc_lo, 0, v6 v_add_nc_u32_e32 v8, v6, v7 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_xor_b32_e32 v8, v8, v7 v_mul_hi_u32 v9, v8, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_lo_u32 v9, v9, s8 v_sub_nc_u32_e32 v8, v8, v9 s_delay_alu instid0(VALU_DEP_1) v_subrev_nc_u32_e32 v9, s8, v8 s_branch .LBB0_6 .LBB0_4: s_or_b32 exec_lo, exec_lo, s11 v_ashrrev_i32_e32 v12, 31, v10 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v13, 31, v11 v_add_nc_u32_e32 v10, v10, v12 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v11, v11, v13 v_xor_b32_e32 v14, v10, v12 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_xor_b32_e32 v15, v11, v13 v_mad_u64_u32 v[10:11], null, v14, v5, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_hi_u32 v10, v15, v2 v_mul_lo_u32 v11, v11, s8 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_mul_lo_u32 v10, v10, s3 v_sub_nc_u32_e32 v11, v14, v11 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_sub_nc_u32_e32 v10, v15, v10 v_subrev_nc_u32_e32 v14, s8, v11 v_cmp_le_u32_e64 s2, s8, v11 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2) v_subrev_nc_u32_e32 v15, s3, v10 v_cndmask_b32_e64 v11, v11, v14, s2 v_cmp_le_u32_e64 s2, s3, v10 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_subrev_nc_u32_e32 v14, s8, v11 v_cndmask_b32_e64 v10, v10, v15, s2 v_cmp_le_u32_e64 s2, s8, v11 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_subrev_nc_u32_e32 v15, s3, v10 v_cndmask_b32_e64 v11, v11, v14, s2 v_cmp_le_u32_e64 s2, s3, v10 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_xor_b32_e32 v11, v11, v12 v_cndmask_b32_e64 v10, v10, v15, s2 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_sub_nc_u32_e32 v14, v11, v12 v_xor_b32_e32 v10, v10, v13 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_sub_nc_u32_e32 v10, v10, v13 v_mad_u64_u32 v[11:12], null, v14, s5, v[10:11] s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_ashrrev_i32_e32 v12, 31, v11 v_lshlrev_b64 v[10:11], 2, v[11:12] s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add_co_u32 v10, s2, s6, v10 v_add_co_ci_u32_e64 v11, s2, s7, v11, s2 global_load_b32 v10, v[10:11], off s_waitcnt vmcnt(0) v_cmp_ne_u32_e64 s2, 0, v10 .LBB0_5: s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_cndmask_b32_e64 v10, 0, 1, s2 s_add_i32 s10, s10, 1 s_cmp_eq_u32 s10, 2 s_delay_alu instid0(VALU_DEP_1) v_add_nc_u32_e32 v0, v0, v10 s_cbranch_scc1 .LBB0_2 .LBB0_6: s_or_b32 s2, s10, s9 s_delay_alu instid0(SALU_CYCLE_1) s_cmp_eq_u32 s2, 0 s_cbranch_scc1 .LBB0_11 v_mov_b32_e32 v10, v6 s_and_saveexec_b32 s11, vcc_lo v_cmp_le_u32_e64 s2, s8, v8 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cndmask_b32_e64 v10, v8, v9, s2 v_subrev_nc_u32_e32 v11, s8, v10 v_cmp_le_u32_e64 s2, s8, v10 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cndmask_b32_e64 v10, v10, v11, s2 v_xor_b32_e32 v10, v10, v7 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_sub_nc_u32_e32 v10, v10, v7 v_sub_nc_u32_e32 v11, 0, v10 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_max_i32_e32 v10, v10, v11 v_sub_nc_u32_e32 v10, s4, v10 s_or_b32 exec_lo, exec_lo, s11 v_add_nc_u32_e32 v11, s10, v4 s_mov_b32 s11, exec_lo s_delay_alu instid0(VALU_DEP_1) v_cmpx_gt_i32_e32 0, v11 s_cbranch_execz .LBB0_4 v_ashrrev_i32_e32 v11, 31, v11 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_add3_u32 v12, v4, s10, v11 v_xor_b32_e32 v12, v12, v11 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_hi_u32 v13, v12, v2 v_mul_lo_u32 v13, v13, s3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_sub_nc_u32_e32 v12, v12, v13 v_subrev_nc_u32_e32 v13, s3, v12 v_cmp_le_u32_e64 s2, s3, v12 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cndmask_b32_e64 v12, v12, v13, s2 v_subrev_nc_u32_e32 v13, s3, v12 v_cmp_le_u32_e64 s2, s3, v12 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_cndmask_b32_e64 v12, v12, v13, s2 v_xor_b32_e32 v12, v12, v11 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_sub_nc_u32_e32 v11, v12, v11 v_sub_nc_u32_e32 v12, 0, v11 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_max_i32_e32 v11, v11, v12 v_sub_nc_u32_e32 v11, s5, v11 s_branch .LBB0_4 .LBB0_11: s_mov_b32 s2, 0 s_branch .LBB0_5 .LBB0_12: v_ashrrev_i32_e32 v2, 31, v1 s_mov_b32 s2, 0 s_mov_b32 s3, exec_lo s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[3:4], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v3, vcc_lo, s6, v3 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v4, vcc_lo, s7, v4, vcc_lo global_load_b32 v4, v[3:4], off s_waitcnt vmcnt(0) v_cmpx_lt_i32_e32 0, v4 s_xor_b32 s3, exec_lo, s3 v_cmp_eq_u32_e32 vcc_lo, 1, v4 s_mov_b32 s4, 0 s_and_b32 s2, vcc_lo, exec_lo s_or_saveexec_b32 s3, s3 v_mov_b32_e32 v3, s4 s_xor_b32 exec_lo, exec_lo, s3 s_cbranch_execz .LBB0_24 v_mov_b32_e32 v3, 0 s_mov_b32 s5, s2 s_mov_b32 s4, exec_lo v_cmpx_eq_u32_e32 0, v4 s_cbranch_execz .LBB0_23 v_mov_b32_e32 v3, 0 s_mov_b32 s5, 0 s_mov_b32 s6, exec_lo v_cmpx_lt_i32_e32 5, v0 s_xor_b32 s6, exec_lo, s6 v_cmp_ne_u32_e32 vcc_lo, 6, v0 v_mov_b32_e32 v3, 1 s_and_b32 s5, vcc_lo, exec_lo s_and_not1_saveexec_b32 s6, s6 v_cmp_ne_u32_e32 vcc_lo, 3, v0 v_mov_b32_e32 v3, 1 s_and_not1_b32 s5, s5, exec_lo s_and_b32 s7, vcc_lo, exec_lo s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 s5, s5, s7 s_or_b32 exec_lo, exec_lo, s6 s_mov_b32 s6, s2 s_and_saveexec_b32 s7, s5 v_cmp_eq_u32_e32 vcc_lo, 1, v4 v_mov_b32_e32 v3, 0 s_and_not1_b32 s5, s2, exec_lo s_and_b32 s6, vcc_lo, exec_lo s_delay_alu instid0(SALU_CYCLE_1) s_or_b32 s6, s5, s6 s_or_b32 exec_lo, exec_lo, s7 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) s_and_not1_b32 s5, s2, exec_lo s_and_b32 s6, s6, exec_lo s_or_b32 s5, s5, s6 .LBB0_23: s_or_b32 exec_lo, exec_lo, s4 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) s_and_not1_b32 s2, s2, exec_lo s_and_b32 s4, s5, exec_lo s_or_b32 s2, s2, s4 .LBB0_24: s_or_b32 exec_lo, exec_lo, s3 s_and_saveexec_b32 s3, s2 v_and_b32_e32 v0, -2, v0 s_delay_alu instid0(VALU_DEP_1) v_cmp_eq_u32_e32 vcc_lo, 2, v0 v_cndmask_b32_e64 v3, 0, 1, vcc_lo s_or_b32 exec_lo, exec_lo, s3 s_load_b64 s[0:1], s[0:1], 0x8 v_lshlrev_b64 v[0:1], 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 v0, vcc_lo, s0, v0 v_add_co_ci_u32_e32 v1, vcc_lo, s1, v1, vcc_lo global_store_b32 v[0:1], v3, off .LBB0_27: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z6updatePiS_ii .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 16 .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 _Z6updatePiS_ii, .Lfunc_end0-_Z6updatePiS_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 - .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: _Z6updatePiS_ii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z6updatePiS_ii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 16 .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 <hip/hip_runtime_api.h> #include <cstdio> // Device input vectors int *d_a; //Device output vector int *d_b; __device__ int mod(int a, int b) { return a >= 0 ? a%b : ( b - abs ( a%b ) ) % b; } __global__ void update(int *A, int *B, int height, int width) { const int index = blockIdx.x * blockDim.x + threadIdx.x; if (index < height*width) { int aliveNeighbours = 0; const int h = index / width; const int w = index % width; /*The neighbours of the cell are checked*/ for (int i = -1; i < 2; ++i) { for (int j = -1; j < 2; ++j) { aliveNeighbours += ( (i | j) && A[mod(h+i,height)*width + mod(w+j,width)]); } } B[index] = (A[index] == 0 && (aliveNeighbours == 3 || aliveNeighbours == 6)) || A[index] == 1 && (aliveNeighbours == 2 || aliveNeighbours == 3); } } void setUp(int height, int width) { // Allocate memory for each vector on GPU hipMalloc((void **) &d_a, height*width*sizeof(int)); hipMalloc((void **) &d_b, height*width*sizeof(int)); } void destroy() { // Release device memory hipFree(d_a); hipFree(d_b); } void updateCuda(int *A, int *B, int height, int width) { // Size, in bytes, of each vector size_t bytes = height*width*sizeof(int); // Copy host vectors to device hipMemcpy(d_a, A, bytes, hipMemcpyHostToDevice); int blockSize, gridSize, n; // // Tamaño de la matriz. // n = height*width; // // // Tamaño del bloque. Elegir entre 32 y 31. // //blockSize = 32; // blockSize = 32; // // // Number of thread blocks in grid // gridSize = (int)ceil((float)n/blockSize); // Execute the kernel update<<< width*height, 1 >>>(d_a, d_b, height, width); // Copy array back to host hipMemcpy( B, d_b, bytes, hipMemcpyDeviceToHost ); }
.text .file "kernel.hip" .globl _Z21__device_stub__updatePiS_ii # -- Begin function _Z21__device_stub__updatePiS_ii .p2align 4, 0x90 .type _Z21__device_stub__updatePiS_ii,@function _Z21__device_stub__updatePiS_ii: # @_Z21__device_stub__updatePiS_ii .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%rsp) movl %ecx, 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 $_Z6updatePiS_ii, %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 _Z21__device_stub__updatePiS_ii, .Lfunc_end0-_Z21__device_stub__updatePiS_ii .cfi_endproc # -- End function .globl _Z5setUpii # -- Begin function _Z5setUpii .p2align 4, 0x90 .type _Z5setUpii,@function _Z5setUpii: # @_Z5setUpii .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset %rbx, -16 imull %esi, %edi movslq %edi, %rbx shlq $2, %rbx movl $d_a, %edi movq %rbx, %rsi callq hipMalloc movl $d_b, %edi movq %rbx, %rsi popq %rbx .cfi_def_cfa_offset 8 jmp hipMalloc # TAILCALL .Lfunc_end1: .size _Z5setUpii, .Lfunc_end1-_Z5setUpii .cfi_endproc # -- End function .globl _Z7destroyv # -- Begin function _Z7destroyv .p2align 4, 0x90 .type _Z7destroyv,@function _Z7destroyv: # @_Z7destroyv .cfi_startproc # %bb.0: pushq %rax .cfi_def_cfa_offset 16 movq d_a(%rip), %rdi callq hipFree movq d_b(%rip), %rdi popq %rax .cfi_def_cfa_offset 8 jmp hipFree # TAILCALL .Lfunc_end2: .size _Z7destroyv, .Lfunc_end2-_Z7destroyv .cfi_endproc # -- End function .globl _Z10updateCudaPiS_ii # -- Begin function _Z10updateCudaPiS_ii .p2align 4, 0x90 .type _Z10updateCudaPiS_ii,@function _Z10updateCudaPiS_ii: # @_Z10updateCudaPiS_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 %r12 .cfi_def_cfa_offset 40 pushq %rbx .cfi_def_cfa_offset 48 subq $112, %rsp .cfi_def_cfa_offset 160 .cfi_offset %rbx, -48 .cfi_offset %r12, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %ecx, %ebp movl %edx, %r15d movq %rsi, %rbx movq %rdi, %rsi movl %ecx, %r12d imull %edx, %r12d movslq %r12d, %r14 shlq $2, %r14 movq d_a(%rip), %rdi movq %r14, %rdx movl $1, %ecx callq hipMemcpy movabsq $4294967296, %rdx # imm = 0x100000000 orq %rdx, %r12 orq $1, %rdx movq %r12, %rdi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB3_2 # %bb.1: movq d_a(%rip), %rax movq d_b(%rip), %rcx movq %rax, 72(%rsp) movq %rcx, 64(%rsp) movl %r15d, 12(%rsp) movl %ebp, 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 $_Z6updatePiS_ii, %edi pushq 16(%rsp) .cfi_adjust_cfa_offset 8 pushq 32(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB3_2: movq d_b(%rip), %rsi movq %rbx, %rdi movq %r14, %rdx movl $2, %ecx callq hipMemcpy addq $112, %rsp .cfi_def_cfa_offset 48 popq %rbx .cfi_def_cfa_offset 40 popq %r12 .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 _Z10updateCudaPiS_ii, .Lfunc_end3-_Z10updateCudaPiS_ii .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 .LBB4_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB4_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z6updatePiS_ii, %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_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 d_a,@object # @d_a .bss .globl d_a .p2align 3, 0x0 d_a: .quad 0 .size d_a, 8 .type d_b,@object # @d_b .globl d_b .p2align 3, 0x0 d_b: .quad 0 .size d_b, 8 .type _Z6updatePiS_ii,@object # @_Z6updatePiS_ii .section .rodata,"a",@progbits .globl _Z6updatePiS_ii .p2align 3, 0x0 _Z6updatePiS_ii: .quad _Z21__device_stub__updatePiS_ii .size _Z6updatePiS_ii, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z6updatePiS_ii" .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 _Z21__device_stub__updatePiS_ii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym d_a .addrsig_sym d_b .addrsig_sym _Z6updatePiS_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_000207cd_00000000-6_kernel.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2063: .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 .LFE2063: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z3modii .type _Z3modii, @function _Z3modii: .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 _Z3modii, .-_Z3modii .globl _Z5setUpii .type _Z5setUpii, @function _Z5setUpii: .LFB2058: .cfi_startproc endbr64 pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset 3, -16 imull %esi, %edi movslq %edi, %rbx salq $2, %rbx movq %rbx, %rsi leaq d_a(%rip), %rdi call cudaMalloc@PLT movq %rbx, %rsi leaq d_b(%rip), %rdi call cudaMalloc@PLT popq %rbx .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2058: .size _Z5setUpii, .-_Z5setUpii .globl _Z7destroyv .type _Z7destroyv, @function _Z7destroyv: .LFB2059: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 movq d_a(%rip), %rdi call cudaFree@PLT movq d_b(%rip), %rdi call cudaFree@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2059: .size _Z7destroyv, .-_Z7destroyv .globl _Z29__device_stub__Z6updatePiS_iiPiS_ii .type _Z29__device_stub__Z6updatePiS_iiPiS_ii, @function _Z29__device_stub__Z6updatePiS_iiPiS_ii: .LFB2085: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%rsp) movl %ecx, 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 .L13 .L9: movq 136(%rsp), %rax subq %fs:40, %rax jne .L14 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L13: .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 _Z6updatePiS_ii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L9 .L14: call __stack_chk_fail@PLT .cfi_endproc .LFE2085: .size _Z29__device_stub__Z6updatePiS_iiPiS_ii, .-_Z29__device_stub__Z6updatePiS_iiPiS_ii .globl _Z6updatePiS_ii .type _Z6updatePiS_ii, @function _Z6updatePiS_ii: .LFB2086: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z6updatePiS_iiPiS_ii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2086: .size _Z6updatePiS_ii, .-_Z6updatePiS_ii .globl _Z10updateCudaPiS_ii .type _Z10updateCudaPiS_ii, @function _Z10updateCudaPiS_ii: .LFB2060: .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 $32, %rsp .cfi_def_cfa_offset 80 movq %rsi, %r13 movl %edx, %ebp movl %ecx, %r12d movl %edx, %r14d imull %ecx, %r14d movslq %r14d, %rbx salq $2, %rbx movl $1, %ecx movq %rbx, %rdx movq %rdi, %rsi movq d_a(%rip), %rdi call cudaMemcpy@PLT movl $1, 20(%rsp) movl $1, 24(%rsp) movl %r14d, 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 .L20 .L18: movl $2, %ecx movq %rbx, %rdx movq d_b(%rip), %rsi movq %r13, %rdi call cudaMemcpy@PLT addq $32, %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 .L20: .cfi_restore_state movl %r12d, %ecx movl %ebp, %edx movq d_b(%rip), %rsi movq d_a(%rip), %rdi call _Z29__device_stub__Z6updatePiS_iiPiS_ii jmp .L18 .cfi_endproc .LFE2060: .size _Z10updateCudaPiS_ii, .-_Z10updateCudaPiS_ii .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z6updatePiS_ii" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2088: .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 _Z6updatePiS_ii(%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 .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 .globl d_b .bss .align 8 .type d_b, @object .size d_b, 8 d_b: .zero 8 .globl d_a .align 8 .type d_a, @object .size d_a, 8 d_a: .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 "kernel.hip" .globl _Z21__device_stub__updatePiS_ii # -- Begin function _Z21__device_stub__updatePiS_ii .p2align 4, 0x90 .type _Z21__device_stub__updatePiS_ii,@function _Z21__device_stub__updatePiS_ii: # @_Z21__device_stub__updatePiS_ii .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%rsp) movl %ecx, 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 $_Z6updatePiS_ii, %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 _Z21__device_stub__updatePiS_ii, .Lfunc_end0-_Z21__device_stub__updatePiS_ii .cfi_endproc # -- End function .globl _Z5setUpii # -- Begin function _Z5setUpii .p2align 4, 0x90 .type _Z5setUpii,@function _Z5setUpii: # @_Z5setUpii .cfi_startproc # %bb.0: pushq %rbx .cfi_def_cfa_offset 16 .cfi_offset %rbx, -16 imull %esi, %edi movslq %edi, %rbx shlq $2, %rbx movl $d_a, %edi movq %rbx, %rsi callq hipMalloc movl $d_b, %edi movq %rbx, %rsi popq %rbx .cfi_def_cfa_offset 8 jmp hipMalloc # TAILCALL .Lfunc_end1: .size _Z5setUpii, .Lfunc_end1-_Z5setUpii .cfi_endproc # -- End function .globl _Z7destroyv # -- Begin function _Z7destroyv .p2align 4, 0x90 .type _Z7destroyv,@function _Z7destroyv: # @_Z7destroyv .cfi_startproc # %bb.0: pushq %rax .cfi_def_cfa_offset 16 movq d_a(%rip), %rdi callq hipFree movq d_b(%rip), %rdi popq %rax .cfi_def_cfa_offset 8 jmp hipFree # TAILCALL .Lfunc_end2: .size _Z7destroyv, .Lfunc_end2-_Z7destroyv .cfi_endproc # -- End function .globl _Z10updateCudaPiS_ii # -- Begin function _Z10updateCudaPiS_ii .p2align 4, 0x90 .type _Z10updateCudaPiS_ii,@function _Z10updateCudaPiS_ii: # @_Z10updateCudaPiS_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 %r12 .cfi_def_cfa_offset 40 pushq %rbx .cfi_def_cfa_offset 48 subq $112, %rsp .cfi_def_cfa_offset 160 .cfi_offset %rbx, -48 .cfi_offset %r12, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movl %ecx, %ebp movl %edx, %r15d movq %rsi, %rbx movq %rdi, %rsi movl %ecx, %r12d imull %edx, %r12d movslq %r12d, %r14 shlq $2, %r14 movq d_a(%rip), %rdi movq %r14, %rdx movl $1, %ecx callq hipMemcpy movabsq $4294967296, %rdx # imm = 0x100000000 orq %rdx, %r12 orq $1, %rdx movq %r12, %rdi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB3_2 # %bb.1: movq d_a(%rip), %rax movq d_b(%rip), %rcx movq %rax, 72(%rsp) movq %rcx, 64(%rsp) movl %r15d, 12(%rsp) movl %ebp, 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 $_Z6updatePiS_ii, %edi pushq 16(%rsp) .cfi_adjust_cfa_offset 8 pushq 32(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB3_2: movq d_b(%rip), %rsi movq %rbx, %rdi movq %r14, %rdx movl $2, %ecx callq hipMemcpy addq $112, %rsp .cfi_def_cfa_offset 48 popq %rbx .cfi_def_cfa_offset 40 popq %r12 .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 _Z10updateCudaPiS_ii, .Lfunc_end3-_Z10updateCudaPiS_ii .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 .LBB4_2 # %bb.1: movl $__hip_fatbin_wrapper, %edi callq __hipRegisterFatBinary movq %rax, __hip_gpubin_handle(%rip) .LBB4_2: movq __hip_gpubin_handle(%rip), %rdi xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z6updatePiS_ii, %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_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 d_a,@object # @d_a .bss .globl d_a .p2align 3, 0x0 d_a: .quad 0 .size d_a, 8 .type d_b,@object # @d_b .globl d_b .p2align 3, 0x0 d_b: .quad 0 .size d_b, 8 .type _Z6updatePiS_ii,@object # @_Z6updatePiS_ii .section .rodata,"a",@progbits .globl _Z6updatePiS_ii .p2align 3, 0x0 _Z6updatePiS_ii: .quad _Z21__device_stub__updatePiS_ii .size _Z6updatePiS_ii, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z6updatePiS_ii" .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 _Z21__device_stub__updatePiS_ii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym d_a .addrsig_sym d_b .addrsig_sym _Z6updatePiS_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> /* printf, scanf, puts, NULL */ #include <stdlib.h> /* srand, rand */ #include <time.h> /* time */ #include <iostream> // From: https://a248.e.akamai.net/f/862/5658/3/developer.download.nvidia.com/assets/cuda/files/reduction.pdf template<unsigned int blockSize> __device__ void warpReduce(volatile double *sdata, unsigned int tid){ if (blockSize >= 64) { sdata[tid] += sdata[tid + 32] ; } if (blockSize >= 32) { sdata[tid] += sdata[tid + 16] ; } if (blockSize >= 16) { sdata[tid] += sdata[tid + 8]; } if (blockSize >= 8) { sdata[tid] += sdata[tid + 4]; } if (blockSize >= 4) {sdata[tid] += sdata[tid + 2]; } if (blockSize >= 2) {sdata[tid] += sdata[tid + 1]; } } template <unsigned int blockSize> __global__ void reduce6( double *g_idata, double *g_odata, unsigned int n) { extern __shared__ double sdata[]; unsigned int tid = threadIdx.x; unsigned int i = blockIdx.x*(blockSize*2) + tid ; unsigned int gridSize = blockSize*2*gridDim.x; sdata[tid] = 0; while (i < n) { sdata[tid] += g_idata[i] + g_idata[i + blockSize]; i += gridSize; } __syncthreads(); if ( blockSize >= 512) { if ( tid < 256) { sdata[tid] += sdata[tid + 256]; } __syncthreads(); } if ( blockSize >= 256) { if ( tid < 128) { sdata[tid] += sdata[tid + 128]; } __syncthreads(); } if ( blockSize >= 128) { if ( tid < 64) { sdata[tid] += sdata[tid + 64]; } __syncthreads(); } if (tid < 32) { warpReduce<blockSize>(sdata,tid); } __syncthreads(); if (tid == 0) {g_odata[blockIdx.x] = sdata[0]; } } int main() { for (int t = 0; t <= 4; t++) { for (int iter = 1; iter <= 15; iter++) { for (int bl = 1; bl <= 4; bl*=2) { printf("Iter %d, Blocks %d",t,bl); int blocks = 32*bl; int threads = 1024; int arraySize = 3200*1024*iter; int smemSize = threads * sizeof(double); int arrayBytes = arraySize * sizeof(double); printf("=====\n"); printf("Input Size = %d\n",arraySize); double *h_in, *h_out; // host arrays h_in = (double*) malloc(arrayBytes); double *d_in, *d_out; //device arrays h_out = (double*) malloc(smemSize); for (int i = 0; i < threads; i++) h_out[i] = 0; double result = 0; for (int i = 0; i < arraySize; i++) { h_in[i] = i; result += i; } cudaEvent_t start, stop, startT, stopT; float time,full; cudaEventCreate(&start); cudaEventCreate(&startT); cudaEventCreate(&stop); cudaEventCreate(&stopT); //allocate memory on device and copy cudaEventRecord(startT,0); cudaMalloc((void**)&d_in, arrayBytes); cudaMalloc((void**)&d_out, smemSize); cudaMemcpy(d_in, h_in, arrayBytes, cudaMemcpyHostToDevice); cudaEventRecord(start,0); reduce6<512><<<blocks,threads,smemSize>>>(d_in,d_out,arraySize); cudaDeviceSynchronize(); cudaEventRecord(stop, 0); cudaEventSynchronize(stop); cudaMemcpy(h_out, d_out, smemSize, cudaMemcpyDeviceToHost); double res = 0; for (int i = 0; i < blocks;i++){ res += h_out[i]; } cudaEventRecord(stopT,0); cudaEventSynchronize(stopT); std::cout << "Device Result is: " << h_out[0] << std::endl; std::cout << "Host Result is: " << result << std::endl; std::cout << "Result is: " << res << std::endl; printf("Result correct? %s\n", res == result ? "true" : "false"); cudaEventElapsedTime(&time, start, stop); cudaEventElapsedTime(&full, startT, stopT); printf ("Time for the kernel: %f ms\n", time); printf ("Time Full: %f ms\n", full); free(h_in); free(h_out); cudaFree(d_in); cudaFree(d_out); } } } }
code for sm_80 Function : _Z7reduce6ILj512EEvPdS0_j .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_TID.X ; /* 0x0000000000007919 */ /* 0x000e220000002100 */ /*0020*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe20000000a00 */ /*0030*/ BSSY B0, 0x1a0 ; /* 0x0000016000007945 */ /* 0x000fe40003800000 */ /*0040*/ S2R R3, SR_CTAID.X ; /* 0x0000000000037919 */ /* 0x000e680000002500 */ /*0050*/ STS.64 [R0.X8], RZ ; /* 0x000000ff00007388 */ /* 0x0011e20000008a00 */ /*0060*/ IMAD R2, R3, 0x400, R0 ; /* 0x0000040003027824 */ /* 0x002fca00078e0200 */ /*0070*/ ISETP.GE.U32.AND P0, PT, R2, c[0x0][0x170], PT ; /* 0x00005c0002007a0c */ /* 0x000fda0003f06070 */ /*0080*/ @P0 BRA 0x190 ; /* 0x0000010000000947 */ /* 0x000fea0003800000 */ /*0090*/ BSSY B1, 0x180 ; /* 0x000000e000017945 */ /* 0x001fe20003800000 */ /*00a0*/ CS2R R4, SRZ ; /* 0x0000000000047805 */ /* 0x000fe4000001ff00 */ /*00b0*/ IADD3 R8, R2, 0x200, RZ ; /* 0x0000020002087810 */ /* 0x000fe20007ffe0ff */ /*00c0*/ IMAD.MOV.U32 R7, RZ, RZ, 0x8 ; /* 0x00000008ff077424 */ /* 0x000fc800078e00ff */ /*00d0*/ IMAD.WIDE.U32 R8, R8, R7, c[0x0][0x160] ; /* 0x0000580008087625 */ /* 0x000fc800078e0007 */ /*00e0*/ IMAD.WIDE.U32 R6, R2, R7, c[0x0][0x160] ; /* 0x0000580002067625 */ /* 0x000fe400078e0007 */ /*00f0*/ LDG.E.64 R8, [R8.64] ; /* 0x0000000408087981 */ /* 0x000ea8000c1e1b00 */ /*0100*/ LDG.E.64 R6, [R6.64] ; /* 0x0000000406067981 */ /* 0x000ea2000c1e1b00 */ /*0110*/ MOV R11, c[0x0][0xc] ; /* 0x00000300000b7a02 */ /* 0x001fca0000000f00 */ /*0120*/ IMAD R2, R11, 0x400, R2 ; /* 0x000004000b027824 */ /* 0x000fca00078e0202 */ /*0130*/ ISETP.GE.U32.AND P0, PT, R2, c[0x0][0x170], PT ; /* 0x00005c0002007a0c */ /* 0x000fe20003f06070 */ /*0140*/ DADD R10, R8, R6 ; /* 0x00000000080a7229 */ /* 0x004e0c0000000006 */ /*0150*/ DADD R4, R10, R4 ; /* 0x000000000a047229 */ /* 0x00304c0000000004 */ /*0160*/ @!P0 BRA 0xb0 ; /* 0xffffff4000008947 */ /* 0x000fea000383ffff */ /*0170*/ BSYNC B1 ; /* 0x0000000000017941 */ /* 0x000fea0003800000 */ /*0180*/ STS.64 [R0.X8], R4 ; /* 0x0000000400007388 */ /* 0x0023e40000008a00 */ /*0190*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x001fea0003800000 */ /*01a0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe20000010000 */ /*01b0*/ ISETP.GT.U32.AND P0, PT, R0.reuse, 0xff, PT ; /* 0x000000ff0000780c */ /* 0x040fe40003f04070 */ /*01c0*/ ISETP.GT.U32.AND P1, PT, R0.reuse, 0x7f, PT ; /* 0x0000007f0000780c */ /* 0x040fe40003f24070 */ /*01d0*/ ISETP.NE.AND P2, PT, R0, RZ, PT ; /* 0x000000ff0000720c */ /* 0x000fe20003f45270 */ /*01e0*/ BSSY B0, 0x4a0 ; /* 0x000002b000007945 */ /* 0x000ff00003800000 */ /*01f0*/ @!P0 LDS.64 R4, [R0.X8] ; /* 0x0000000000048984 */ /* 0x002fe80000008a00 */ /*0200*/ @!P0 LDS.64 R6, [R0.X8+0x800] ; /* 0x0008000000068984 */ /* 0x000e240000008a00 */ /*0210*/ @!P0 DADD R4, R4, R6 ; /* 0x0000000004048229 */ /* 0x001e0e0000000006 */ /*0220*/ @!P0 STS.64 [R0.X8], R4 ; /* 0x0000000400008388 */ /* 0x001fe80000008a00 */ /*0230*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe20000010000 */ /*0240*/ ISETP.GT.U32.AND P0, PT, R0, 0x3f, PT ; /* 0x0000003f0000780c */ /* 0x000fca0003f04070 */ /*0250*/ @!P1 LDS.64 R6, [R0.X8] ; /* 0x0000000000069984 */ /* 0x000fe80000008a00 */ /*0260*/ @!P1 LDS.64 R8, [R0.X8+0x400] ; /* 0x0004000000089984 */ /* 0x000e240000008a00 */ /*0270*/ @!P1 DADD R6, R6, R8 ; /* 0x0000000006069229 */ /* 0x001e0e0000000008 */ /*0280*/ @!P1 STS.64 [R0.X8], R6 ; /* 0x0000000600009388 */ /* 0x001fe80000008a00 */ /*0290*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe20000010000 */ /*02a0*/ ISETP.GT.U32.AND P1, PT, R0, 0x1f, PT ; /* 0x0000001f0000780c */ /* 0x000fca0003f24070 */ /*02b0*/ @!P0 LDS.64 R8, [R0.X8] ; /* 0x0000000000088984 */ /* 0x000fe80000008a00 */ /*02c0*/ @!P0 LDS.64 R10, [R0.X8+0x200] ; /* 0x00020000000a8984 */ /* 0x000e240000008a00 */ /*02d0*/ @!P0 DADD R8, R8, R10 ; /* 0x0000000008088229 */ /* 0x001e0e000000000a */ /*02e0*/ @!P0 STS.64 [R0.X8], R8 ; /* 0x0000000800008388 */ /* 0x0011e80000008a00 */ /*02f0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fec0000010000 */ /*0300*/ @P1 BRA 0x490 ; /* 0x0000018000001947 */ /* 0x000fea0003800000 */ /*0310*/ LDS.64 R4, [R0.X8] ; /* 0x0000000000047984 */ /* 0x000fe80000008a00 */ /*0320*/ LDS.64 R6, [R0.X8+0x100] ; /* 0x0001000000067984 */ /* 0x000e640000008a00 */ /*0330*/ DADD R4, R4, R6 ; /* 0x0000000004047229 */ /* 0x002e4e0000000006 */ /*0340*/ STS.64 [R0.X8], R4 ; /* 0x0000000400007388 */ /* 0x002fe80000008a00 */ /*0350*/ LDS.64 R6, [R0.X8] ; /* 0x0000000000067984 */ /* 0x000fe80000008a00 */ /*0360*/ LDS.64 R8, [R0.X8+0x80] ; /* 0x0000800000087984 */ /* 0x001e240000008a00 */ /*0370*/ DADD R6, R6, R8 ; /* 0x0000000006067229 */ /* 0x001e0e0000000008 */ /*0380*/ STS.64 [R0.X8], R6 ; /* 0x0000000600007388 */ /* 0x001fe80000008a00 */ /*0390*/ LDS.64 R8, [R0.X8] ; /* 0x0000000000087984 */ /* 0x000fe80000008a00 */ /*03a0*/ LDS.64 R10, [R0.X8+0x40] ; /* 0x00004000000a7984 */ /* 0x000e240000008a00 */ /*03b0*/ DADD R8, R8, R10 ; /* 0x0000000008087229 */ /* 0x001e0e000000000a */ /*03c0*/ STS.64 [R0.X8], R8 ; /* 0x0000000800007388 */ /* 0x001fe80000008a00 */ /*03d0*/ LDS.64 R10, [R0.X8] ; /* 0x00000000000a7984 */ /* 0x000fe80000008a00 */ /*03e0*/ LDS.64 R12, [R0.X8+0x20] ; /* 0x00002000000c7984 */ /* 0x000e240000008a00 */ /*03f0*/ DADD R10, R10, R12 ; /* 0x000000000a0a7229 */ /* 0x001e0e000000000c */ /*0400*/ STS.64 [R0.X8], R10 ; /* 0x0000000a00007388 */ /* 0x001fe80000008a00 */ /*0410*/ LDS.64 R4, [R0.X8] ; /* 0x0000000000047984 */ /* 0x000fe80000008a00 */ /*0420*/ LDS.64 R12, [R0.X8+0x10] ; /* 0x00001000000c7984 */ /* 0x000e240000008a00 */ /*0430*/ DADD R4, R4, R12 ; /* 0x0000000004047229 */ /* 0x001e0e000000000c */ /*0440*/ STS.64 [R0.X8], R4 ; /* 0x0000000400007388 */ /* 0x001fe80000008a00 */ /*0450*/ LDS.64 R6, [R0.X8] ; /* 0x0000000000067984 */ /* 0x000fe80000008a00 */ /*0460*/ LDS.64 R12, [R0.X8+0x8] ; /* 0x00000800000c7984 */ /* 0x000e240000008a00 */ /*0470*/ DADD R6, R6, R12 ; /* 0x0000000006067229 */ /* 0x001e0e000000000c */ /*0480*/ STS.64 [R0.X8], R6 ; /* 0x0000000600007388 */ /* 0x0011e80000008a00 */ /*0490*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*04a0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fec0000010000 */ /*04b0*/ @P2 EXIT ; /* 0x000000000000294d */ /* 0x000fea0003800000 */ /*04c0*/ LDS.64 R4, [RZ] ; /* 0x00000000ff047984 */ /* 0x000e620000000a00 */ /*04d0*/ MOV R2, 0x8 ; /* 0x0000000800027802 */ /* 0x000fca0000000f00 */ /*04e0*/ IMAD.WIDE.U32 R2, R3, R2, c[0x0][0x168] ; /* 0x00005a0003027625 */ /* 0x000fca00078e0002 */ /*04f0*/ STG.E.64 [R2.64], R4 ; /* 0x0000000402007986 */ /* 0x002fe2000c101b04 */ /*0500*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0510*/ BRA 0x510; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0520*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0530*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0540*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0550*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0560*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0570*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0580*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0590*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include <stdio.h> /* printf, scanf, puts, NULL */ #include <stdlib.h> /* srand, rand */ #include <time.h> /* time */ #include <iostream> // From: https://a248.e.akamai.net/f/862/5658/3/developer.download.nvidia.com/assets/cuda/files/reduction.pdf template<unsigned int blockSize> __device__ void warpReduce(volatile double *sdata, unsigned int tid){ if (blockSize >= 64) { sdata[tid] += sdata[tid + 32] ; } if (blockSize >= 32) { sdata[tid] += sdata[tid + 16] ; } if (blockSize >= 16) { sdata[tid] += sdata[tid + 8]; } if (blockSize >= 8) { sdata[tid] += sdata[tid + 4]; } if (blockSize >= 4) {sdata[tid] += sdata[tid + 2]; } if (blockSize >= 2) {sdata[tid] += sdata[tid + 1]; } } template <unsigned int blockSize> __global__ void reduce6( double *g_idata, double *g_odata, unsigned int n) { extern __shared__ double sdata[]; unsigned int tid = threadIdx.x; unsigned int i = blockIdx.x*(blockSize*2) + tid ; unsigned int gridSize = blockSize*2*gridDim.x; sdata[tid] = 0; while (i < n) { sdata[tid] += g_idata[i] + g_idata[i + blockSize]; i += gridSize; } __syncthreads(); if ( blockSize >= 512) { if ( tid < 256) { sdata[tid] += sdata[tid + 256]; } __syncthreads(); } if ( blockSize >= 256) { if ( tid < 128) { sdata[tid] += sdata[tid + 128]; } __syncthreads(); } if ( blockSize >= 128) { if ( tid < 64) { sdata[tid] += sdata[tid + 64]; } __syncthreads(); } if (tid < 32) { warpReduce<blockSize>(sdata,tid); } __syncthreads(); if (tid == 0) {g_odata[blockIdx.x] = sdata[0]; } } int main() { for (int t = 0; t <= 4; t++) { for (int iter = 1; iter <= 15; iter++) { for (int bl = 1; bl <= 4; bl*=2) { printf("Iter %d, Blocks %d",t,bl); int blocks = 32*bl; int threads = 1024; int arraySize = 3200*1024*iter; int smemSize = threads * sizeof(double); int arrayBytes = arraySize * sizeof(double); printf("=====\n"); printf("Input Size = %d\n",arraySize); double *h_in, *h_out; // host arrays h_in = (double*) malloc(arrayBytes); double *d_in, *d_out; //device arrays h_out = (double*) malloc(smemSize); for (int i = 0; i < threads; i++) h_out[i] = 0; double result = 0; for (int i = 0; i < arraySize; i++) { h_in[i] = i; result += i; } cudaEvent_t start, stop, startT, stopT; float time,full; cudaEventCreate(&start); cudaEventCreate(&startT); cudaEventCreate(&stop); cudaEventCreate(&stopT); //allocate memory on device and copy cudaEventRecord(startT,0); cudaMalloc((void**)&d_in, arrayBytes); cudaMalloc((void**)&d_out, smemSize); cudaMemcpy(d_in, h_in, arrayBytes, cudaMemcpyHostToDevice); cudaEventRecord(start,0); reduce6<512><<<blocks,threads,smemSize>>>(d_in,d_out,arraySize); cudaDeviceSynchronize(); cudaEventRecord(stop, 0); cudaEventSynchronize(stop); cudaMemcpy(h_out, d_out, smemSize, cudaMemcpyDeviceToHost); double res = 0; for (int i = 0; i < blocks;i++){ res += h_out[i]; } cudaEventRecord(stopT,0); cudaEventSynchronize(stopT); std::cout << "Device Result is: " << h_out[0] << std::endl; std::cout << "Host Result is: " << result << std::endl; std::cout << "Result is: " << res << std::endl; printf("Result correct? %s\n", res == result ? "true" : "false"); cudaEventElapsedTime(&time, start, stop); cudaEventElapsedTime(&full, startT, stopT); printf ("Time for the kernel: %f ms\n", time); printf ("Time Full: %f ms\n", full); free(h_in); free(h_out); cudaFree(d_in); cudaFree(d_out); } } } }
.file "tmpxft_000c4427_00000000-6_reduce.cudafe1.cpp" .text #APP .globl _ZSt21ios_base_library_initv #NO_APP .type _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j, @function _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j: .LFB3697: .cfi_startproc subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%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 12(%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 .L5 .L1: movq 120(%rsp), %rax subq %fs:40, %rax jne .L6 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L5: .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 _Z7reduce6ILj512EEvPdS0_j(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L1 .L6: call __stack_chk_fail@PLT .cfi_endproc .LFE3697: .size _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j, .-_ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j .section .text._Z7reduce6ILj512EEvPdS0_j,"axG",@progbits,_Z7reduce6ILj512EEvPdS0_j,comdat .weak _Z7reduce6ILj512EEvPdS0_j .type _Z7reduce6ILj512EEvPdS0_j, @function _Z7reduce6ILj512EEvPdS0_j: .LFB4000: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE4000: .size _Z7reduce6ILj512EEvPdS0_j, .-_Z7reduce6ILj512EEvPdS0_j .text .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB3675: .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 .LFE3675: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .section .rodata.str1.1,"aMS",@progbits,1 .LC1: .string "true" .LC2: .string "false" .LC3: .string "Iter %d, Blocks %d" .LC4: .string "=====\n" .LC5: .string "Input Size = %d\n" .LC6: .string "Device Result is: " .LC7: .string "Host Result is: " .LC8: .string "Result is: " .LC9: .string "Result correct? %s\n" .LC10: .string "Time for the kernel: %f ms\n" .LC11: .string "Time Full: %f ms\n" .text .globl main .type main, @function main: .LFB3672: .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 %fs:40, %rax movq %rax, 104(%rsp) xorl %eax, %eax movl $0, %r8d jmp .L12 .L38: movq $0x000000000, 8(%rsp) jmp .L14 .L51: movl 28(%rsp), %edx movq 40(%rsp), %rsi movq 32(%rsp), %rdi call _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j jmp .L16 .L39: movq $0x000000000, (%rsp) jmp .L17 .L52: movq 104(%rsp), %rax subq %fs:40, %rax jne .L47 call _ZSt16__throw_bad_castv@PLT .L47: call __stack_chk_fail@PLT .L21: movq %r15, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq (%r15), %rax movl $10, %esi movq %r15, %rdi call *48(%rax) jmp .L22 .L53: movq 104(%rsp), %rax subq %fs:40, %rax jne .L48 call _ZSt16__throw_bad_castv@PLT .L48: call __stack_chk_fail@PLT .L25: movq %r15, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq (%r15), %rax movl $10, %esi movq %r15, %rdi call *48(%rax) jmp .L26 .L54: movq 104(%rsp), %rax subq %fs:40, %rax jne .L49 call _ZSt16__throw_bad_castv@PLT .L49: call __stack_chk_fail@PLT .L29: movq %r15, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq (%r15), %rax movl $10, %esi movq %r15, %rdi call *48(%rax) jmp .L30 .L40: leaq .LC2(%rip), %rdx .L31: leaq .LC9(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq 80(%rsp), %rdi movq 56(%rsp), %rdx movq 48(%rsp), %rsi call cudaEventElapsedTime@PLT leaq 92(%rsp), %rdi movq 72(%rsp), %rdx movq 64(%rsp), %rsi call cudaEventElapsedTime@PLT pxor %xmm0, %xmm0 cvtss2sd 80(%rsp), %xmm0 leaq .LC10(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT pxor %xmm0, %xmm0 cvtss2sd 92(%rsp), %xmm0 leaq .LC11(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movq %rbp, %rdi call free@PLT movq %r13, %rdi call free@PLT movq 32(%rsp), %rdi call cudaFree@PLT movq 40(%rsp), %rdi call cudaFree@PLT addl %r14d, %r14d cmpl $4, %r14d jg .L50 .L33: movl %r14d, %ecx movl 24(%rsp), %edx leaq .LC3(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %r14d, %r15d sall $5, %r15d leaq .LC4(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %r12d, %edx leaq .LC5(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq 16(%rsp), %rdi call malloc@PLT movq %rax, %rbp movl $8192, %edi call malloc@PLT movq %rax, %r13 movq %rax, %rbx leaq 8192(%rax), %rdx .L13: movq $0x000000000, (%rax) addq $8, %rax cmpq %rdx, %rax jne .L13 testl %r12d, %r12d jle .L38 movl $0, %eax movq $0x000000000, 8(%rsp) .L15: pxor %xmm0, %xmm0 cvtsi2sdl %eax, %xmm0 movsd %xmm0, 0(%rbp,%rax,8) addsd 8(%rsp), %xmm0 movsd %xmm0, 8(%rsp) addq $1, %rax cmpl %eax, %r12d jg .L15 .L14: leaq 48(%rsp), %rdi call cudaEventCreate@PLT leaq 64(%rsp), %rdi call cudaEventCreate@PLT leaq 56(%rsp), %rdi call cudaEventCreate@PLT leaq 72(%rsp), %rdi call cudaEventCreate@PLT movl $0, %esi movq 64(%rsp), %rdi call cudaEventRecord@PLT leaq 32(%rsp), %rdi movq 16(%rsp), %rsi call cudaMalloc@PLT leaq 40(%rsp), %rdi movl $8192, %esi call cudaMalloc@PLT movl $1, %ecx movq 16(%rsp), %rdx movq %rbp, %rsi movq 32(%rsp), %rdi call cudaMemcpy@PLT movl $0, %esi movq 48(%rsp), %rdi call cudaEventRecord@PLT movl $1024, 92(%rsp) movl $1, 96(%rsp) movl $1, 100(%rsp) movl %r15d, 80(%rsp) movl $1, 84(%rsp) movl $1, 88(%rsp) movl $0, %r9d movl $8192, %r8d movq 92(%rsp), %rdx movl $1, %ecx movq 80(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L51 .L16: call cudaDeviceSynchronize@PLT movl $0, %esi movq 56(%rsp), %rdi call cudaEventRecord@PLT movq 56(%rsp), %rdi call cudaEventSynchronize@PLT movl $2, %ecx movl $8192, %edx movq 40(%rsp), %rsi movq %r13, %rdi call cudaMemcpy@PLT testl %r15d, %r15d jle .L39 leal -1(%r15), %eax leaq 8(%r13,%rax,8), %rax movq $0x000000000, (%rsp) .L18: movsd (%rsp), %xmm1 addsd (%rbx), %xmm1 movsd %xmm1, (%rsp) addq $8, %rbx cmpq %rax, %rbx jne .L18 .L17: movl $0, %esi movq 72(%rsp), %rdi call cudaEventRecord@PLT movq 72(%rsp), %rdi call cudaEventSynchronize@PLT movl $18, %edx leaq .LC6(%rip), %rsi leaq _ZSt4cout(%rip), %rbx movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movsd 0(%r13), %xmm0 movq %rbx, %rdi call _ZNSo9_M_insertIdEERSoT_@PLT movq %rax, %rbx movq (%rax), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %r15 testq %r15, %r15 je .L52 cmpb $0, 56(%r15) je .L21 movzbl 67(%r15), %eax .L22: movsbl %al, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT movl $16, %edx leaq .LC7(%rip), %rsi leaq _ZSt4cout(%rip), %rbx movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movsd 8(%rsp), %xmm0 movq %rbx, %rdi call _ZNSo9_M_insertIdEERSoT_@PLT movq %rax, %rbx movq (%rax), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %r15 testq %r15, %r15 je .L53 cmpb $0, 56(%r15) je .L25 movzbl 67(%r15), %eax .L26: movsbl %al, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT movl $11, %edx leaq .LC8(%rip), %rsi leaq _ZSt4cout(%rip), %rbx movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movsd (%rsp), %xmm0 movq %rbx, %rdi call _ZNSo9_M_insertIdEERSoT_@PLT movq %rax, %rbx movq (%rax), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %r15 testq %r15, %r15 je .L54 cmpb $0, 56(%r15) je .L29 movzbl 67(%r15), %eax .L30: movsbl %al, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT movsd 8(%rsp), %xmm3 movsd (%rsp), %xmm4 ucomisd %xmm4, %xmm3 jp .L40 leaq .LC1(%rip), %rdx je .L31 jmp .L40 .L50: movl 28(%rsp), %r9d movq 16(%rsp), %rcx addl $3276800, %r9d addq $26214400, %rcx cmpl $52428800, %r9d je .L34 .L36: movl %r9d, %r12d movl $1, %r14d movl %r9d, 28(%rsp) movq %rcx, 16(%rsp) jmp .L33 .L34: movl 24(%rsp), %r8d addl $1, %r8d cmpl $5, %r8d je .L35 .L12: movl $26214400, %ecx movl $3276800, %r9d movl %r8d, 24(%rsp) jmp .L36 .L35: movq 104(%rsp), %rax subq %fs:40, %rax jne .L55 movl $0, %eax 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 .L55: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE3672: .size main, .-main .section .rodata.str1.1 .LC12: .string "_Z7reduce6ILj512EEvPdS0_j" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB3700: .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 _Z7reduce6ILj512EEvPdS0_j(%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 .LFE3700: .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> /* printf, scanf, puts, NULL */ #include <stdlib.h> /* srand, rand */ #include <time.h> /* time */ #include <iostream> // From: https://a248.e.akamai.net/f/862/5658/3/developer.download.nvidia.com/assets/cuda/files/reduction.pdf template<unsigned int blockSize> __device__ void warpReduce(volatile double *sdata, unsigned int tid){ if (blockSize >= 64) { sdata[tid] += sdata[tid + 32] ; } if (blockSize >= 32) { sdata[tid] += sdata[tid + 16] ; } if (blockSize >= 16) { sdata[tid] += sdata[tid + 8]; } if (blockSize >= 8) { sdata[tid] += sdata[tid + 4]; } if (blockSize >= 4) {sdata[tid] += sdata[tid + 2]; } if (blockSize >= 2) {sdata[tid] += sdata[tid + 1]; } } template <unsigned int blockSize> __global__ void reduce6( double *g_idata, double *g_odata, unsigned int n) { extern __shared__ double sdata[]; unsigned int tid = threadIdx.x; unsigned int i = blockIdx.x*(blockSize*2) + tid ; unsigned int gridSize = blockSize*2*gridDim.x; sdata[tid] = 0; while (i < n) { sdata[tid] += g_idata[i] + g_idata[i + blockSize]; i += gridSize; } __syncthreads(); if ( blockSize >= 512) { if ( tid < 256) { sdata[tid] += sdata[tid + 256]; } __syncthreads(); } if ( blockSize >= 256) { if ( tid < 128) { sdata[tid] += sdata[tid + 128]; } __syncthreads(); } if ( blockSize >= 128) { if ( tid < 64) { sdata[tid] += sdata[tid + 64]; } __syncthreads(); } if (tid < 32) { warpReduce<blockSize>(sdata,tid); } __syncthreads(); if (tid == 0) {g_odata[blockIdx.x] = sdata[0]; } } int main() { for (int t = 0; t <= 4; t++) { for (int iter = 1; iter <= 15; iter++) { for (int bl = 1; bl <= 4; bl*=2) { printf("Iter %d, Blocks %d",t,bl); int blocks = 32*bl; int threads = 1024; int arraySize = 3200*1024*iter; int smemSize = threads * sizeof(double); int arrayBytes = arraySize * sizeof(double); printf("=====\n"); printf("Input Size = %d\n",arraySize); double *h_in, *h_out; // host arrays h_in = (double*) malloc(arrayBytes); double *d_in, *d_out; //device arrays h_out = (double*) malloc(smemSize); for (int i = 0; i < threads; i++) h_out[i] = 0; double result = 0; for (int i = 0; i < arraySize; i++) { h_in[i] = i; result += i; } cudaEvent_t start, stop, startT, stopT; float time,full; cudaEventCreate(&start); cudaEventCreate(&startT); cudaEventCreate(&stop); cudaEventCreate(&stopT); //allocate memory on device and copy cudaEventRecord(startT,0); cudaMalloc((void**)&d_in, arrayBytes); cudaMalloc((void**)&d_out, smemSize); cudaMemcpy(d_in, h_in, arrayBytes, cudaMemcpyHostToDevice); cudaEventRecord(start,0); reduce6<512><<<blocks,threads,smemSize>>>(d_in,d_out,arraySize); cudaDeviceSynchronize(); cudaEventRecord(stop, 0); cudaEventSynchronize(stop); cudaMemcpy(h_out, d_out, smemSize, cudaMemcpyDeviceToHost); double res = 0; for (int i = 0; i < blocks;i++){ res += h_out[i]; } cudaEventRecord(stopT,0); cudaEventSynchronize(stopT); std::cout << "Device Result is: " << h_out[0] << std::endl; std::cout << "Host Result is: " << result << std::endl; std::cout << "Result is: " << res << std::endl; printf("Result correct? %s\n", res == result ? "true" : "false"); cudaEventElapsedTime(&time, start, stop); cudaEventElapsedTime(&full, startT, stopT); printf ("Time for the kernel: %f ms\n", time); printf ("Time Full: %f ms\n", full); free(h_in); free(h_out); cudaFree(d_in); cudaFree(d_out); } } } }
#include <hip/hip_runtime.h> #include <stdio.h> /* printf, scanf, puts, NULL */ #include <stdlib.h> /* srand, rand */ #include <time.h> /* time */ #include <iostream> // From: https://a248.e.akamai.net/f/862/5658/3/developer.download.nvidia.com/assets/cuda/files/reduction.pdf template<unsigned int blockSize> __device__ void warpReduce(volatile double *sdata, unsigned int tid){ if (blockSize >= 64) { sdata[tid] += sdata[tid + 32] ; } if (blockSize >= 32) { sdata[tid] += sdata[tid + 16] ; } if (blockSize >= 16) { sdata[tid] += sdata[tid + 8]; } if (blockSize >= 8) { sdata[tid] += sdata[tid + 4]; } if (blockSize >= 4) {sdata[tid] += sdata[tid + 2]; } if (blockSize >= 2) {sdata[tid] += sdata[tid + 1]; } } template <unsigned int blockSize> __global__ void reduce6( double *g_idata, double *g_odata, unsigned int n) { extern __shared__ double sdata[]; unsigned int tid = threadIdx.x; unsigned int i = blockIdx.x*(blockSize*2) + tid ; unsigned int gridSize = blockSize*2*gridDim.x; sdata[tid] = 0; while (i < n) { sdata[tid] += g_idata[i] + g_idata[i + blockSize]; i += gridSize; } __syncthreads(); if ( blockSize >= 512) { if ( tid < 256) { sdata[tid] += sdata[tid + 256]; } __syncthreads(); } if ( blockSize >= 256) { if ( tid < 128) { sdata[tid] += sdata[tid + 128]; } __syncthreads(); } if ( blockSize >= 128) { if ( tid < 64) { sdata[tid] += sdata[tid + 64]; } __syncthreads(); } if (tid < 32) { warpReduce<blockSize>(sdata,tid); } __syncthreads(); if (tid == 0) {g_odata[blockIdx.x] = sdata[0]; } } int main() { for (int t = 0; t <= 4; t++) { for (int iter = 1; iter <= 15; iter++) { for (int bl = 1; bl <= 4; bl*=2) { printf("Iter %d, Blocks %d",t,bl); int blocks = 32*bl; int threads = 1024; int arraySize = 3200*1024*iter; int smemSize = threads * sizeof(double); int arrayBytes = arraySize * sizeof(double); printf("=====\n"); printf("Input Size = %d\n",arraySize); double *h_in, *h_out; // host arrays h_in = (double*) malloc(arrayBytes); double *d_in, *d_out; //device arrays h_out = (double*) malloc(smemSize); for (int i = 0; i < threads; i++) h_out[i] = 0; double result = 0; for (int i = 0; i < arraySize; i++) { h_in[i] = i; result += i; } hipEvent_t start, stop, startT, stopT; float time,full; hipEventCreate(&start); hipEventCreate(&startT); hipEventCreate(&stop); hipEventCreate(&stopT); //allocate memory on device and copy hipEventRecord(startT,0); hipMalloc((void**)&d_in, arrayBytes); hipMalloc((void**)&d_out, smemSize); hipMemcpy(d_in, h_in, arrayBytes, hipMemcpyHostToDevice); hipEventRecord(start,0); reduce6<512><<<blocks,threads,smemSize>>>(d_in,d_out,arraySize); hipDeviceSynchronize(); hipEventRecord(stop, 0); hipEventSynchronize(stop); hipMemcpy(h_out, d_out, smemSize, hipMemcpyDeviceToHost); double res = 0; for (int i = 0; i < blocks;i++){ res += h_out[i]; } hipEventRecord(stopT,0); hipEventSynchronize(stopT); std::cout << "Device Result is: " << h_out[0] << std::endl; std::cout << "Host Result is: " << result << std::endl; std::cout << "Result is: " << res << std::endl; printf("Result correct? %s\n", res == result ? "true" : "false"); hipEventElapsedTime(&time, start, stop); hipEventElapsedTime(&full, startT, stopT); printf ("Time for the kernel: %f ms\n", time); printf ("Time Full: %f ms\n", full); free(h_in); free(h_out); hipFree(d_in); hipFree(d_out); } } } }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <hip/hip_runtime.h> #include <stdio.h> /* printf, scanf, puts, NULL */ #include <stdlib.h> /* srand, rand */ #include <time.h> /* time */ #include <iostream> // From: https://a248.e.akamai.net/f/862/5658/3/developer.download.nvidia.com/assets/cuda/files/reduction.pdf template<unsigned int blockSize> __device__ void warpReduce(volatile double *sdata, unsigned int tid){ if (blockSize >= 64) { sdata[tid] += sdata[tid + 32] ; } if (blockSize >= 32) { sdata[tid] += sdata[tid + 16] ; } if (blockSize >= 16) { sdata[tid] += sdata[tid + 8]; } if (blockSize >= 8) { sdata[tid] += sdata[tid + 4]; } if (blockSize >= 4) {sdata[tid] += sdata[tid + 2]; } if (blockSize >= 2) {sdata[tid] += sdata[tid + 1]; } } template <unsigned int blockSize> __global__ void reduce6( double *g_idata, double *g_odata, unsigned int n) { extern __shared__ double sdata[]; unsigned int tid = threadIdx.x; unsigned int i = blockIdx.x*(blockSize*2) + tid ; unsigned int gridSize = blockSize*2*gridDim.x; sdata[tid] = 0; while (i < n) { sdata[tid] += g_idata[i] + g_idata[i + blockSize]; i += gridSize; } __syncthreads(); if ( blockSize >= 512) { if ( tid < 256) { sdata[tid] += sdata[tid + 256]; } __syncthreads(); } if ( blockSize >= 256) { if ( tid < 128) { sdata[tid] += sdata[tid + 128]; } __syncthreads(); } if ( blockSize >= 128) { if ( tid < 64) { sdata[tid] += sdata[tid + 64]; } __syncthreads(); } if (tid < 32) { warpReduce<blockSize>(sdata,tid); } __syncthreads(); if (tid == 0) {g_odata[blockIdx.x] = sdata[0]; } } int main() { for (int t = 0; t <= 4; t++) { for (int iter = 1; iter <= 15; iter++) { for (int bl = 1; bl <= 4; bl*=2) { printf("Iter %d, Blocks %d",t,bl); int blocks = 32*bl; int threads = 1024; int arraySize = 3200*1024*iter; int smemSize = threads * sizeof(double); int arrayBytes = arraySize * sizeof(double); printf("=====\n"); printf("Input Size = %d\n",arraySize); double *h_in, *h_out; // host arrays h_in = (double*) malloc(arrayBytes); double *d_in, *d_out; //device arrays h_out = (double*) malloc(smemSize); for (int i = 0; i < threads; i++) h_out[i] = 0; double result = 0; for (int i = 0; i < arraySize; i++) { h_in[i] = i; result += i; } hipEvent_t start, stop, startT, stopT; float time,full; hipEventCreate(&start); hipEventCreate(&startT); hipEventCreate(&stop); hipEventCreate(&stopT); //allocate memory on device and copy hipEventRecord(startT,0); hipMalloc((void**)&d_in, arrayBytes); hipMalloc((void**)&d_out, smemSize); hipMemcpy(d_in, h_in, arrayBytes, hipMemcpyHostToDevice); hipEventRecord(start,0); reduce6<512><<<blocks,threads,smemSize>>>(d_in,d_out,arraySize); hipDeviceSynchronize(); hipEventRecord(stop, 0); hipEventSynchronize(stop); hipMemcpy(h_out, d_out, smemSize, hipMemcpyDeviceToHost); double res = 0; for (int i = 0; i < blocks;i++){ res += h_out[i]; } hipEventRecord(stopT,0); hipEventSynchronize(stopT); std::cout << "Device Result is: " << h_out[0] << std::endl; std::cout << "Host Result is: " << result << std::endl; std::cout << "Result is: " << res << std::endl; printf("Result correct? %s\n", res == result ? "true" : "false"); hipEventElapsedTime(&time, start, stop); hipEventElapsedTime(&full, startT, stopT); printf ("Time for the kernel: %f ms\n", time); printf ("Time Full: %f ms\n", full); free(h_in); free(h_out); hipFree(d_in); hipFree(d_out); } } } }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .section .text._Z7reduce6ILj512EEvPdS0_j,"axG",@progbits,_Z7reduce6ILj512EEvPdS0_j,comdat .protected _Z7reduce6ILj512EEvPdS0_j .globl _Z7reduce6ILj512EEvPdS0_j .p2align 8 .type _Z7reduce6ILj512EEvPdS0_j,@function _Z7reduce6ILj512EEvPdS0_j: s_load_b32 s5, s[0:1], 0x10 v_mov_b32_e32 v2, 0 s_mov_b32 s4, s15 v_lshl_add_u32 v5, v0, 3, 0 v_lshl_or_b32 v1, s4, 10, v0 s_mov_b32 s7, 0 v_mov_b32_e32 v3, v2 s_mov_b32 s6, exec_lo ds_store_b64 v5, v[2:3] s_waitcnt lgkmcnt(0) v_cmpx_gt_u32_e64 s5, v1 s_cbranch_execz .LBB0_4 s_clause 0x1 s_load_b32 s8, s[0:1], 0x18 s_load_b64 s[2:3], s[0:1], 0x0 ds_load_b64 v[3:4], v5 s_waitcnt lgkmcnt(0) s_lshl_b32 s8, s8, 10 .p2align 6 .LBB0_2: v_mov_b32_e32 v2, 0 v_add_nc_u32_e32 v6, 0x200, v1 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(VALU_DEP_3) v_lshlrev_b64 v[8:9], 3, v[1:2] v_add_nc_u32_e32 v1, s8, v1 v_mov_b32_e32 v7, v2 v_add_co_u32 v8, vcc_lo, s2, v8 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_lshlrev_b64 v[6:7], 3, v[6:7] v_add_co_ci_u32_e32 v9, vcc_lo, s3, v9, vcc_lo v_add_co_u32 v6, vcc_lo, s2, v6 s_delay_alu instid0(VALU_DEP_3) v_add_co_ci_u32_e32 v7, vcc_lo, s3, v7, vcc_lo s_clause 0x1 global_load_b64 v[8:9], v[8:9], off global_load_b64 v[6:7], v[6:7], off v_cmp_le_u32_e32 vcc_lo, s5, v1 s_or_b32 s7, vcc_lo, s7 s_waitcnt vmcnt(0) v_add_f64 v[6:7], v[8:9], v[6:7] s_delay_alu instid0(VALU_DEP_1) v_add_f64 v[3:4], v[3:4], v[6:7] s_and_not1_b32 exec_lo, exec_lo, s7 s_cbranch_execnz .LBB0_2 s_or_b32 exec_lo, exec_lo, s7 ds_store_b64 v5, v[3:4] .LBB0_4: s_or_b32 exec_lo, exec_lo, s6 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s2, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e32 0x100, v0 s_cbranch_execz .LBB0_6 v_lshl_add_u32 v1, v0, 3, 0 ds_load_b64 v[1:2], v1 offset:2048 ds_load_b64 v[3:4], v5 s_waitcnt lgkmcnt(0) v_add_f64 v[1:2], v[1:2], v[3:4] ds_store_b64 v5, v[1:2] .LBB0_6: s_or_b32 exec_lo, exec_lo, s2 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s2, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e32 0x80, v0 s_cbranch_execz .LBB0_8 v_lshl_add_u32 v1, v0, 3, 0 ds_load_b64 v[1:2], v1 offset:1024 ds_load_b64 v[3:4], v5 s_waitcnt lgkmcnt(0) v_add_f64 v[1:2], v[1:2], v[3:4] ds_store_b64 v5, v[1:2] .LBB0_8: s_or_b32 exec_lo, exec_lo, s2 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s2, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e32 64, v0 s_cbranch_execz .LBB0_10 v_lshl_add_u32 v1, v0, 3, 0 ds_load_b64 v[1:2], v1 offset:512 ds_load_b64 v[3:4], v5 s_waitcnt lgkmcnt(0) v_add_f64 v[1:2], v[1:2], v[3:4] ds_store_b64 v5, v[1:2] .LBB0_10: s_or_b32 exec_lo, exec_lo, s2 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s3, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e32 32, v0 s_cbranch_execz .LBB0_12 v_lshl_add_u32 v7, v0, 3, 0 s_mov_b64 s[6:7], src_shared_base s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v2, 0x100, v7 v_cmp_ne_u32_e64 s2, -1, v7 v_cmp_ne_u32_e32 vcc_lo, -1, v2 s_delay_alu instid0(VALU_DEP_2) v_cndmask_b32_e64 v1, 0, v7, s2 v_cndmask_b32_e32 v3, 0, v2, vcc_lo v_cndmask_b32_e64 v4, 0, s7, vcc_lo v_cndmask_b32_e64 v2, 0, s7, s2 flat_load_b64 v[3:4], v[3:4] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 0x80, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 64, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 32, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 16, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 8, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 .LBB0_12: s_or_b32 exec_lo, exec_lo, s3 s_waitcnt lgkmcnt(0) s_waitcnt_vscnt null, 0x0 s_barrier buffer_gl0_inv s_mov_b32 s2, exec_lo v_cmpx_eq_u32_e32 0, v0 s_cbranch_execz .LBB0_14 v_mov_b32_e32 v0, 0 s_load_b64 s[0:1], s[0:1], 0x8 s_mov_b32 s5, 0 v_mov_b32_e32 v2, 0 s_lshl_b64 s[2:3], s[4:5], 3 ds_load_b64 v[0:1], v0 s_waitcnt lgkmcnt(0) s_add_u32 s0, s0, s2 s_addc_u32 s1, s1, s3 global_store_b64 v2, v[0:1], s[0:1] .LBB0_14: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z7reduce6ILj512EEvPdS0_j .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 10 .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._Z7reduce6ILj512EEvPdS0_j,"axG",@progbits,_Z7reduce6ILj512EEvPdS0_j,comdat .Lfunc_end0: .size _Z7reduce6ILj512EEvPdS0_j, .Lfunc_end0-_Z7reduce6ILj512EEvPdS0_j .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 sdata .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: 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 - .offset: 144 .size: 4 .value_kind: hidden_dynamic_lds_size .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: _Z7reduce6ILj512EEvPdS0_j .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z7reduce6ILj512EEvPdS0_j.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 10 .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> /* printf, scanf, puts, NULL */ #include <stdlib.h> /* srand, rand */ #include <time.h> /* time */ #include <iostream> // From: https://a248.e.akamai.net/f/862/5658/3/developer.download.nvidia.com/assets/cuda/files/reduction.pdf template<unsigned int blockSize> __device__ void warpReduce(volatile double *sdata, unsigned int tid){ if (blockSize >= 64) { sdata[tid] += sdata[tid + 32] ; } if (blockSize >= 32) { sdata[tid] += sdata[tid + 16] ; } if (blockSize >= 16) { sdata[tid] += sdata[tid + 8]; } if (blockSize >= 8) { sdata[tid] += sdata[tid + 4]; } if (blockSize >= 4) {sdata[tid] += sdata[tid + 2]; } if (blockSize >= 2) {sdata[tid] += sdata[tid + 1]; } } template <unsigned int blockSize> __global__ void reduce6( double *g_idata, double *g_odata, unsigned int n) { extern __shared__ double sdata[]; unsigned int tid = threadIdx.x; unsigned int i = blockIdx.x*(blockSize*2) + tid ; unsigned int gridSize = blockSize*2*gridDim.x; sdata[tid] = 0; while (i < n) { sdata[tid] += g_idata[i] + g_idata[i + blockSize]; i += gridSize; } __syncthreads(); if ( blockSize >= 512) { if ( tid < 256) { sdata[tid] += sdata[tid + 256]; } __syncthreads(); } if ( blockSize >= 256) { if ( tid < 128) { sdata[tid] += sdata[tid + 128]; } __syncthreads(); } if ( blockSize >= 128) { if ( tid < 64) { sdata[tid] += sdata[tid + 64]; } __syncthreads(); } if (tid < 32) { warpReduce<blockSize>(sdata,tid); } __syncthreads(); if (tid == 0) {g_odata[blockIdx.x] = sdata[0]; } } int main() { for (int t = 0; t <= 4; t++) { for (int iter = 1; iter <= 15; iter++) { for (int bl = 1; bl <= 4; bl*=2) { printf("Iter %d, Blocks %d",t,bl); int blocks = 32*bl; int threads = 1024; int arraySize = 3200*1024*iter; int smemSize = threads * sizeof(double); int arrayBytes = arraySize * sizeof(double); printf("=====\n"); printf("Input Size = %d\n",arraySize); double *h_in, *h_out; // host arrays h_in = (double*) malloc(arrayBytes); double *d_in, *d_out; //device arrays h_out = (double*) malloc(smemSize); for (int i = 0; i < threads; i++) h_out[i] = 0; double result = 0; for (int i = 0; i < arraySize; i++) { h_in[i] = i; result += i; } hipEvent_t start, stop, startT, stopT; float time,full; hipEventCreate(&start); hipEventCreate(&startT); hipEventCreate(&stop); hipEventCreate(&stopT); //allocate memory on device and copy hipEventRecord(startT,0); hipMalloc((void**)&d_in, arrayBytes); hipMalloc((void**)&d_out, smemSize); hipMemcpy(d_in, h_in, arrayBytes, hipMemcpyHostToDevice); hipEventRecord(start,0); reduce6<512><<<blocks,threads,smemSize>>>(d_in,d_out,arraySize); hipDeviceSynchronize(); hipEventRecord(stop, 0); hipEventSynchronize(stop); hipMemcpy(h_out, d_out, smemSize, hipMemcpyDeviceToHost); double res = 0; for (int i = 0; i < blocks;i++){ res += h_out[i]; } hipEventRecord(stopT,0); hipEventSynchronize(stopT); std::cout << "Device Result is: " << h_out[0] << std::endl; std::cout << "Host Result is: " << result << std::endl; std::cout << "Result is: " << res << std::endl; printf("Result correct? %s\n", res == result ? "true" : "false"); hipEventElapsedTime(&time, start, stop); hipEventElapsedTime(&full, startT, stopT); printf ("Time for the kernel: %f ms\n", time); printf ("Time Full: %f ms\n", full); free(h_in); free(h_out); hipFree(d_in); hipFree(d_out); } } } }
.text .file "reduce.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl main # -- Begin function 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 $200, %rsp .cfi_def_cfa_offset 256 .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, %rax # imm = 0x100000000 addq $1024, %rax # imm = 0x400 movq %rax, 144(%rsp) # 8-byte Spill movl $0, 8(%rsp) # 4-byte Folded Spill jmp .LBB0_1 .p2align 4, 0x90 .LBB0_12: # in Loop: Header=BB0_1 Depth=1 movl 8(%rsp), %eax # 4-byte Reload incl %eax movl %eax, 8(%rsp) # 4-byte Spill cmpl $5, %eax je .LBB0_13 .LBB0_1: # %.preheader75 # =>This Loop Header: Depth=1 # Child Loop BB0_2 Depth 2 # Child Loop BB0_3 Depth 3 # Child Loop BB0_4 Depth 4 # Child Loop BB0_8 Depth 4 movl $3276800, %r13d # imm = 0x320000 movl $1, %eax jmp .LBB0_2 .p2align 4, 0x90 .LBB0_24: # in Loop: Header=BB0_2 Depth=2 movq 136(%rsp), %rax # 8-byte Reload incq %rax addq $3276800, %r13 # imm = 0x320000 cmpq $16, %rax je .LBB0_12 .LBB0_2: # %.preheader74 # Parent Loop BB0_1 Depth=1 # => This Loop Header: Depth=2 # Child Loop BB0_3 Depth 3 # Child Loop BB0_4 Depth 4 # Child Loop BB0_8 Depth 4 imull $3276800, %eax, %ecx # imm = 0x320000 movl %ecx, 12(%rsp) # 4-byte Spill movq %rax, 136(%rsp) # 8-byte Spill imulq $26214400, %rax, %rax # imm = 0x1900000 movq %rax, 72(%rsp) # 8-byte Spill movl $1, %r12d jmp .LBB0_3 .p2align 4, 0x90 .LBB0_22: # in Loop: Header=BB0_3 Depth=3 movq %r14, %rdi movq %rax, %rbp callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) movl %eax, %ecx movq %rbp, %rax .LBB0_23: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit70 # in Loop: Header=BB0_3 Depth=3 movsbl %cl, %esi movq %rax, %rdi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movsd 80(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero ucomisd 88(%rsp), %xmm0 # 8-byte Folded Reload movl $.L.str.7, %esi movl $.L.str.8, %eax cmovneq %rax, %rsi cmovpq %rax, %rsi movl $.L.str.6, %edi xorl %eax, %eax callq printf movq 104(%rsp), %rsi movq 24(%rsp), %rdx leaq 112(%rsp), %rdi callq hipEventElapsedTime movq 96(%rsp), %rsi movq 16(%rsp), %rdx leaq 56(%rsp), %rdi callq hipEventElapsedTime movss 112(%rsp), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str.9, %edi movb $1, %al callq printf movss 56(%rsp), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str.10, %edi movb $1, %al callq printf movq %r15, %rdi callq free movq %rbx, %rdi callq free movq 40(%rsp), %rdi callq hipFree movq 32(%rsp), %rdi callq hipFree leal (%r12,%r12), %eax cmpl $3, %r12d movl %eax, %r12d jae .LBB0_24 .LBB0_3: # %.preheader.preheader # Parent Loop BB0_1 Depth=1 # Parent Loop BB0_2 Depth=2 # => This Loop Header: Depth=3 # Child Loop BB0_4 Depth 4 # Child Loop BB0_8 Depth 4 movl $.L.str, %edi movl 8(%rsp), %esi # 4-byte Reload movl %r12d, %edx xorl %eax, %eax callq printf movl $.Lstr, %edi callq puts@PLT movl $.L.str.2, %edi movl 12(%rsp), %esi # 4-byte Reload xorl %eax, %eax callq printf movq 72(%rsp), %rdi # 8-byte Reload callq malloc movq %rax, %r15 movl $8192, %edi # imm = 0x2000 callq malloc movq %rax, %rbx movl $8192, %edx # imm = 0x2000 movq %rax, %rdi xorl %esi, %esi callq memset@PLT xorpd %xmm1, %xmm1 xorl %eax, %eax .p2align 4, 0x90 .LBB0_4: # %.preheader # Parent Loop BB0_1 Depth=1 # Parent Loop BB0_2 Depth=2 # Parent Loop BB0_3 Depth=3 # => This Inner Loop Header: Depth=4 xorps %xmm0, %xmm0 cvtsi2sd %eax, %xmm0 movsd %xmm0, (%r15,%rax,8) addsd %xmm0, %xmm1 incq %rax cmpq %rax, %r13 jne .LBB0_4 # %bb.5: # in Loop: Header=BB0_3 Depth=3 movsd %xmm1, 88(%rsp) # 8-byte Spill movl %r12d, %r14d shll $5, %r14d leaq 104(%rsp), %rdi callq hipEventCreate leaq 96(%rsp), %rdi callq hipEventCreate leaq 24(%rsp), %rdi callq hipEventCreate leaq 16(%rsp), %rdi callq hipEventCreate movq 96(%rsp), %rdi xorl %esi, %esi callq hipEventRecord leaq 40(%rsp), %rdi movq 72(%rsp), %rbp # 8-byte Reload movq %rbp, %rsi callq hipMalloc movl $8192, %esi # imm = 0x2000 leaq 32(%rsp), %rdi callq hipMalloc movq 40(%rsp), %rdi movq %r15, %rsi movq %rbp, %rdx movl $1, %ecx callq hipMemcpy movq 104(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movabsq $4294967296, %rax # imm = 0x100000000 orq %rax, %r14 movl $8192, %r8d # imm = 0x2000 movq %r14, %rdi movl $1, %esi movq 144(%rsp), %rdx # 8-byte Reload movl $1, %ecx xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB0_7 # %bb.6: # in Loop: Header=BB0_3 Depth=3 movq 40(%rsp), %rax movq 32(%rsp), %rcx movq %rax, 192(%rsp) movq %rcx, 184(%rsp) movl 12(%rsp), %eax # 4-byte Reload movl %eax, 52(%rsp) leaq 192(%rsp), %rax movq %rax, 112(%rsp) leaq 184(%rsp), %rax movq %rax, 120(%rsp) leaq 52(%rsp), %rax movq %rax, 128(%rsp) leaq 56(%rsp), %rdi leaq 168(%rsp), %rsi leaq 160(%rsp), %rdx leaq 152(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 168(%rsp), %rcx movl 176(%rsp), %r8d movl $_Z7reduce6ILj512EEvPdS0_j, %edi leaq 112(%rsp), %r9 pushq 152(%rsp) .cfi_adjust_cfa_offset 8 pushq 168(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB0_7: # in Loop: Header=BB0_3 Depth=3 callq hipDeviceSynchronize movq 24(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq 24(%rsp), %rdi callq hipEventSynchronize movq 32(%rsp), %rsi movl $8192, %edx # imm = 0x2000 movq %rbx, %rdi movl $2, %ecx callq hipMemcpy movl %r12d, %eax shlq $8, %rax xorpd %xmm0, %xmm0 xorl %ecx, %ecx .p2align 4, 0x90 .LBB0_8: # Parent Loop BB0_1 Depth=1 # Parent Loop BB0_2 Depth=2 # Parent Loop BB0_3 Depth=3 # => This Inner Loop Header: Depth=4 addsd (%rbx,%rcx), %xmm0 addq $8, %rcx cmpq %rcx, %rax jne .LBB0_8 # %bb.9: # in Loop: Header=BB0_3 Depth=3 movsd %xmm0, 80(%rsp) # 8-byte Spill movq 16(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq 16(%rsp), %rdi callq hipEventSynchronize movl $_ZSt4cout, %edi movl $.L.str.3, %esi movl $18, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movsd (%rbx), %xmm0 # xmm0 = mem[0],zero movl $_ZSt4cout, %edi callq _ZNSo9_M_insertIdEERSoT_ movq (%rax), %rcx movq -24(%rcx), %rcx movq 240(%rax,%rcx), %r14 testq %r14, %r14 je .LBB0_25 # %bb.10: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i # in Loop: Header=BB0_3 Depth=3 cmpb $0, 56(%r14) je .LBB0_14 # %bb.11: # in Loop: Header=BB0_3 Depth=3 movzbl 67(%r14), %ecx jmp .LBB0_15 .p2align 4, 0x90 .LBB0_14: # in Loop: Header=BB0_3 Depth=3 movq %r14, %rdi movq %rax, %rbp callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) movl %eax, %ecx movq %rbp, %rax .LBB0_15: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit # in Loop: Header=BB0_3 Depth=3 movsbl %cl, %esi movq %rax, %rdi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movl $_ZSt4cout, %edi movl $.L.str.4, %esi movl $16, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl $_ZSt4cout, %edi movsd 88(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero callq _ZNSo9_M_insertIdEERSoT_ movq (%rax), %rcx movq -24(%rcx), %rcx movq 240(%rax,%rcx), %r14 testq %r14, %r14 je .LBB0_25 # %bb.16: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i62 # in Loop: Header=BB0_3 Depth=3 cmpb $0, 56(%r14) je .LBB0_18 # %bb.17: # in Loop: Header=BB0_3 Depth=3 movzbl 67(%r14), %ecx jmp .LBB0_19 .p2align 4, 0x90 .LBB0_18: # in Loop: Header=BB0_3 Depth=3 movq %r14, %rdi movq %rax, %rbp callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) movl %eax, %ecx movq %rbp, %rax .LBB0_19: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit65 # in Loop: Header=BB0_3 Depth=3 movsbl %cl, %esi movq %rax, %rdi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movl $_ZSt4cout, %edi movl $.L.str.5, %esi movl $11, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl $_ZSt4cout, %edi movsd 80(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero callq _ZNSo9_M_insertIdEERSoT_ movq (%rax), %rcx movq -24(%rcx), %rcx movq 240(%rax,%rcx), %r14 testq %r14, %r14 je .LBB0_25 # %bb.20: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i67 # in Loop: Header=BB0_3 Depth=3 cmpb $0, 56(%r14) je .LBB0_22 # %bb.21: # in Loop: Header=BB0_3 Depth=3 movzbl 67(%r14), %ecx jmp .LBB0_23 .LBB0_13: xorl %eax, %eax addq $200, %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 .LBB0_25: .cfi_def_cfa_offset 256 callq _ZSt16__throw_bad_castv .Lfunc_end0: .size main, .Lfunc_end0-main .cfi_endproc # -- End function .section .text._Z22__device_stub__reduce6ILj512EEvPdS0_j,"axG",@progbits,_Z22__device_stub__reduce6ILj512EEvPdS0_j,comdat .weak _Z22__device_stub__reduce6ILj512EEvPdS0_j # -- Begin function _Z22__device_stub__reduce6ILj512EEvPdS0_j .p2align 4, 0x90 .type _Z22__device_stub__reduce6ILj512EEvPdS0_j,@function _Z22__device_stub__reduce6ILj512EEvPdS0_j: # @_Z22__device_stub__reduce6ILj512EEvPdS0_j .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%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 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 $_Z7reduce6ILj512EEvPdS0_j, %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_end1: .size _Z22__device_stub__reduce6ILj512EEvPdS0_j, .Lfunc_end1-_Z22__device_stub__reduce6ILj512EEvPdS0_j .cfi_endproc # -- End function .text .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 $_Z7reduce6ILj512EEvPdS0_j, %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 .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "Iter %d, Blocks %d" .size .L.str, 19 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "Input Size = %d\n" .size .L.str.2, 17 .type _Z7reduce6ILj512EEvPdS0_j,@object # @_Z7reduce6ILj512EEvPdS0_j .section .rodata._Z7reduce6ILj512EEvPdS0_j,"aG",@progbits,_Z7reduce6ILj512EEvPdS0_j,comdat .weak _Z7reduce6ILj512EEvPdS0_j .p2align 3, 0x0 _Z7reduce6ILj512EEvPdS0_j: .quad _Z22__device_stub__reduce6ILj512EEvPdS0_j .size _Z7reduce6ILj512EEvPdS0_j, 8 .type .L.str.3,@object # @.str.3 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.3: .asciz "Device Result is: " .size .L.str.3, 19 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "Host Result is: " .size .L.str.4, 17 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "Result is: " .size .L.str.5, 12 .type .L.str.6,@object # @.str.6 .L.str.6: .asciz "Result correct? %s\n" .size .L.str.6, 20 .type .L.str.7,@object # @.str.7 .L.str.7: .asciz "true" .size .L.str.7, 5 .type .L.str.8,@object # @.str.8 .L.str.8: .asciz "false" .size .L.str.8, 6 .type .L.str.9,@object # @.str.9 .L.str.9: .asciz "Time for the kernel: %f ms\n" .size .L.str.9, 28 .type .L.str.10,@object # @.str.10 .L.str.10: .asciz "Time Full: %f ms\n" .size .L.str.10, 18 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z7reduce6ILj512EEvPdS0_j" .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 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "=====" .size .Lstr, 6 .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 _Z22__device_stub__reduce6ILj512EEvPdS0_j .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z7reduce6ILj512EEvPdS0_j .addrsig_sym _ZSt4cout .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 : _Z7reduce6ILj512EEvPdS0_j .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_TID.X ; /* 0x0000000000007919 */ /* 0x000e220000002100 */ /*0020*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe20000000a00 */ /*0030*/ BSSY B0, 0x1a0 ; /* 0x0000016000007945 */ /* 0x000fe40003800000 */ /*0040*/ S2R R3, SR_CTAID.X ; /* 0x0000000000037919 */ /* 0x000e680000002500 */ /*0050*/ STS.64 [R0.X8], RZ ; /* 0x000000ff00007388 */ /* 0x0011e20000008a00 */ /*0060*/ IMAD R2, R3, 0x400, R0 ; /* 0x0000040003027824 */ /* 0x002fca00078e0200 */ /*0070*/ ISETP.GE.U32.AND P0, PT, R2, c[0x0][0x170], PT ; /* 0x00005c0002007a0c */ /* 0x000fda0003f06070 */ /*0080*/ @P0 BRA 0x190 ; /* 0x0000010000000947 */ /* 0x000fea0003800000 */ /*0090*/ BSSY B1, 0x180 ; /* 0x000000e000017945 */ /* 0x001fe20003800000 */ /*00a0*/ CS2R R4, SRZ ; /* 0x0000000000047805 */ /* 0x000fe4000001ff00 */ /*00b0*/ IADD3 R8, R2, 0x200, RZ ; /* 0x0000020002087810 */ /* 0x000fe20007ffe0ff */ /*00c0*/ IMAD.MOV.U32 R7, RZ, RZ, 0x8 ; /* 0x00000008ff077424 */ /* 0x000fc800078e00ff */ /*00d0*/ IMAD.WIDE.U32 R8, R8, R7, c[0x0][0x160] ; /* 0x0000580008087625 */ /* 0x000fc800078e0007 */ /*00e0*/ IMAD.WIDE.U32 R6, R2, R7, c[0x0][0x160] ; /* 0x0000580002067625 */ /* 0x000fe400078e0007 */ /*00f0*/ LDG.E.64 R8, [R8.64] ; /* 0x0000000408087981 */ /* 0x000ea8000c1e1b00 */ /*0100*/ LDG.E.64 R6, [R6.64] ; /* 0x0000000406067981 */ /* 0x000ea2000c1e1b00 */ /*0110*/ MOV R11, c[0x0][0xc] ; /* 0x00000300000b7a02 */ /* 0x001fca0000000f00 */ /*0120*/ IMAD R2, R11, 0x400, R2 ; /* 0x000004000b027824 */ /* 0x000fca00078e0202 */ /*0130*/ ISETP.GE.U32.AND P0, PT, R2, c[0x0][0x170], PT ; /* 0x00005c0002007a0c */ /* 0x000fe20003f06070 */ /*0140*/ DADD R10, R8, R6 ; /* 0x00000000080a7229 */ /* 0x004e0c0000000006 */ /*0150*/ DADD R4, R10, R4 ; /* 0x000000000a047229 */ /* 0x00304c0000000004 */ /*0160*/ @!P0 BRA 0xb0 ; /* 0xffffff4000008947 */ /* 0x000fea000383ffff */ /*0170*/ BSYNC B1 ; /* 0x0000000000017941 */ /* 0x000fea0003800000 */ /*0180*/ STS.64 [R0.X8], R4 ; /* 0x0000000400007388 */ /* 0x0023e40000008a00 */ /*0190*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x001fea0003800000 */ /*01a0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe20000010000 */ /*01b0*/ ISETP.GT.U32.AND P0, PT, R0.reuse, 0xff, PT ; /* 0x000000ff0000780c */ /* 0x040fe40003f04070 */ /*01c0*/ ISETP.GT.U32.AND P1, PT, R0.reuse, 0x7f, PT ; /* 0x0000007f0000780c */ /* 0x040fe40003f24070 */ /*01d0*/ ISETP.NE.AND P2, PT, R0, RZ, PT ; /* 0x000000ff0000720c */ /* 0x000fe20003f45270 */ /*01e0*/ BSSY B0, 0x4a0 ; /* 0x000002b000007945 */ /* 0x000ff00003800000 */ /*01f0*/ @!P0 LDS.64 R4, [R0.X8] ; /* 0x0000000000048984 */ /* 0x002fe80000008a00 */ /*0200*/ @!P0 LDS.64 R6, [R0.X8+0x800] ; /* 0x0008000000068984 */ /* 0x000e240000008a00 */ /*0210*/ @!P0 DADD R4, R4, R6 ; /* 0x0000000004048229 */ /* 0x001e0e0000000006 */ /*0220*/ @!P0 STS.64 [R0.X8], R4 ; /* 0x0000000400008388 */ /* 0x001fe80000008a00 */ /*0230*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe20000010000 */ /*0240*/ ISETP.GT.U32.AND P0, PT, R0, 0x3f, PT ; /* 0x0000003f0000780c */ /* 0x000fca0003f04070 */ /*0250*/ @!P1 LDS.64 R6, [R0.X8] ; /* 0x0000000000069984 */ /* 0x000fe80000008a00 */ /*0260*/ @!P1 LDS.64 R8, [R0.X8+0x400] ; /* 0x0004000000089984 */ /* 0x000e240000008a00 */ /*0270*/ @!P1 DADD R6, R6, R8 ; /* 0x0000000006069229 */ /* 0x001e0e0000000008 */ /*0280*/ @!P1 STS.64 [R0.X8], R6 ; /* 0x0000000600009388 */ /* 0x001fe80000008a00 */ /*0290*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe20000010000 */ /*02a0*/ ISETP.GT.U32.AND P1, PT, R0, 0x1f, PT ; /* 0x0000001f0000780c */ /* 0x000fca0003f24070 */ /*02b0*/ @!P0 LDS.64 R8, [R0.X8] ; /* 0x0000000000088984 */ /* 0x000fe80000008a00 */ /*02c0*/ @!P0 LDS.64 R10, [R0.X8+0x200] ; /* 0x00020000000a8984 */ /* 0x000e240000008a00 */ /*02d0*/ @!P0 DADD R8, R8, R10 ; /* 0x0000000008088229 */ /* 0x001e0e000000000a */ /*02e0*/ @!P0 STS.64 [R0.X8], R8 ; /* 0x0000000800008388 */ /* 0x0011e80000008a00 */ /*02f0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fec0000010000 */ /*0300*/ @P1 BRA 0x490 ; /* 0x0000018000001947 */ /* 0x000fea0003800000 */ /*0310*/ LDS.64 R4, [R0.X8] ; /* 0x0000000000047984 */ /* 0x000fe80000008a00 */ /*0320*/ LDS.64 R6, [R0.X8+0x100] ; /* 0x0001000000067984 */ /* 0x000e640000008a00 */ /*0330*/ DADD R4, R4, R6 ; /* 0x0000000004047229 */ /* 0x002e4e0000000006 */ /*0340*/ STS.64 [R0.X8], R4 ; /* 0x0000000400007388 */ /* 0x002fe80000008a00 */ /*0350*/ LDS.64 R6, [R0.X8] ; /* 0x0000000000067984 */ /* 0x000fe80000008a00 */ /*0360*/ LDS.64 R8, [R0.X8+0x80] ; /* 0x0000800000087984 */ /* 0x001e240000008a00 */ /*0370*/ DADD R6, R6, R8 ; /* 0x0000000006067229 */ /* 0x001e0e0000000008 */ /*0380*/ STS.64 [R0.X8], R6 ; /* 0x0000000600007388 */ /* 0x001fe80000008a00 */ /*0390*/ LDS.64 R8, [R0.X8] ; /* 0x0000000000087984 */ /* 0x000fe80000008a00 */ /*03a0*/ LDS.64 R10, [R0.X8+0x40] ; /* 0x00004000000a7984 */ /* 0x000e240000008a00 */ /*03b0*/ DADD R8, R8, R10 ; /* 0x0000000008087229 */ /* 0x001e0e000000000a */ /*03c0*/ STS.64 [R0.X8], R8 ; /* 0x0000000800007388 */ /* 0x001fe80000008a00 */ /*03d0*/ LDS.64 R10, [R0.X8] ; /* 0x00000000000a7984 */ /* 0x000fe80000008a00 */ /*03e0*/ LDS.64 R12, [R0.X8+0x20] ; /* 0x00002000000c7984 */ /* 0x000e240000008a00 */ /*03f0*/ DADD R10, R10, R12 ; /* 0x000000000a0a7229 */ /* 0x001e0e000000000c */ /*0400*/ STS.64 [R0.X8], R10 ; /* 0x0000000a00007388 */ /* 0x001fe80000008a00 */ /*0410*/ LDS.64 R4, [R0.X8] ; /* 0x0000000000047984 */ /* 0x000fe80000008a00 */ /*0420*/ LDS.64 R12, [R0.X8+0x10] ; /* 0x00001000000c7984 */ /* 0x000e240000008a00 */ /*0430*/ DADD R4, R4, R12 ; /* 0x0000000004047229 */ /* 0x001e0e000000000c */ /*0440*/ STS.64 [R0.X8], R4 ; /* 0x0000000400007388 */ /* 0x001fe80000008a00 */ /*0450*/ LDS.64 R6, [R0.X8] ; /* 0x0000000000067984 */ /* 0x000fe80000008a00 */ /*0460*/ LDS.64 R12, [R0.X8+0x8] ; /* 0x00000800000c7984 */ /* 0x000e240000008a00 */ /*0470*/ DADD R6, R6, R12 ; /* 0x0000000006067229 */ /* 0x001e0e000000000c */ /*0480*/ STS.64 [R0.X8], R6 ; /* 0x0000000600007388 */ /* 0x0011e80000008a00 */ /*0490*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*04a0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fec0000010000 */ /*04b0*/ @P2 EXIT ; /* 0x000000000000294d */ /* 0x000fea0003800000 */ /*04c0*/ LDS.64 R4, [RZ] ; /* 0x00000000ff047984 */ /* 0x000e620000000a00 */ /*04d0*/ MOV R2, 0x8 ; /* 0x0000000800027802 */ /* 0x000fca0000000f00 */ /*04e0*/ IMAD.WIDE.U32 R2, R3, R2, c[0x0][0x168] ; /* 0x00005a0003027625 */ /* 0x000fca00078e0002 */ /*04f0*/ STG.E.64 [R2.64], R4 ; /* 0x0000000402007986 */ /* 0x002fe2000c101b04 */ /*0500*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0510*/ BRA 0x510; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0520*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0530*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0540*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0550*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0560*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0570*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0580*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0590*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*05f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .section .text._Z7reduce6ILj512EEvPdS0_j,"axG",@progbits,_Z7reduce6ILj512EEvPdS0_j,comdat .protected _Z7reduce6ILj512EEvPdS0_j .globl _Z7reduce6ILj512EEvPdS0_j .p2align 8 .type _Z7reduce6ILj512EEvPdS0_j,@function _Z7reduce6ILj512EEvPdS0_j: s_load_b32 s5, s[0:1], 0x10 v_mov_b32_e32 v2, 0 s_mov_b32 s4, s15 v_lshl_add_u32 v5, v0, 3, 0 v_lshl_or_b32 v1, s4, 10, v0 s_mov_b32 s7, 0 v_mov_b32_e32 v3, v2 s_mov_b32 s6, exec_lo ds_store_b64 v5, v[2:3] s_waitcnt lgkmcnt(0) v_cmpx_gt_u32_e64 s5, v1 s_cbranch_execz .LBB0_4 s_clause 0x1 s_load_b32 s8, s[0:1], 0x18 s_load_b64 s[2:3], s[0:1], 0x0 ds_load_b64 v[3:4], v5 s_waitcnt lgkmcnt(0) s_lshl_b32 s8, s8, 10 .p2align 6 .LBB0_2: v_mov_b32_e32 v2, 0 v_add_nc_u32_e32 v6, 0x200, v1 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(VALU_DEP_3) v_lshlrev_b64 v[8:9], 3, v[1:2] v_add_nc_u32_e32 v1, s8, v1 v_mov_b32_e32 v7, v2 v_add_co_u32 v8, vcc_lo, s2, v8 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_lshlrev_b64 v[6:7], 3, v[6:7] v_add_co_ci_u32_e32 v9, vcc_lo, s3, v9, vcc_lo v_add_co_u32 v6, vcc_lo, s2, v6 s_delay_alu instid0(VALU_DEP_3) v_add_co_ci_u32_e32 v7, vcc_lo, s3, v7, vcc_lo s_clause 0x1 global_load_b64 v[8:9], v[8:9], off global_load_b64 v[6:7], v[6:7], off v_cmp_le_u32_e32 vcc_lo, s5, v1 s_or_b32 s7, vcc_lo, s7 s_waitcnt vmcnt(0) v_add_f64 v[6:7], v[8:9], v[6:7] s_delay_alu instid0(VALU_DEP_1) v_add_f64 v[3:4], v[3:4], v[6:7] s_and_not1_b32 exec_lo, exec_lo, s7 s_cbranch_execnz .LBB0_2 s_or_b32 exec_lo, exec_lo, s7 ds_store_b64 v5, v[3:4] .LBB0_4: s_or_b32 exec_lo, exec_lo, s6 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s2, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e32 0x100, v0 s_cbranch_execz .LBB0_6 v_lshl_add_u32 v1, v0, 3, 0 ds_load_b64 v[1:2], v1 offset:2048 ds_load_b64 v[3:4], v5 s_waitcnt lgkmcnt(0) v_add_f64 v[1:2], v[1:2], v[3:4] ds_store_b64 v5, v[1:2] .LBB0_6: s_or_b32 exec_lo, exec_lo, s2 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s2, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e32 0x80, v0 s_cbranch_execz .LBB0_8 v_lshl_add_u32 v1, v0, 3, 0 ds_load_b64 v[1:2], v1 offset:1024 ds_load_b64 v[3:4], v5 s_waitcnt lgkmcnt(0) v_add_f64 v[1:2], v[1:2], v[3:4] ds_store_b64 v5, v[1:2] .LBB0_8: s_or_b32 exec_lo, exec_lo, s2 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s2, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e32 64, v0 s_cbranch_execz .LBB0_10 v_lshl_add_u32 v1, v0, 3, 0 ds_load_b64 v[1:2], v1 offset:512 ds_load_b64 v[3:4], v5 s_waitcnt lgkmcnt(0) v_add_f64 v[1:2], v[1:2], v[3:4] ds_store_b64 v5, v[1:2] .LBB0_10: s_or_b32 exec_lo, exec_lo, s2 s_delay_alu instid0(SALU_CYCLE_1) s_mov_b32 s3, exec_lo s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv v_cmpx_gt_u32_e32 32, v0 s_cbranch_execz .LBB0_12 v_lshl_add_u32 v7, v0, 3, 0 s_mov_b64 s[6:7], src_shared_base s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_add_nc_u32_e32 v2, 0x100, v7 v_cmp_ne_u32_e64 s2, -1, v7 v_cmp_ne_u32_e32 vcc_lo, -1, v2 s_delay_alu instid0(VALU_DEP_2) v_cndmask_b32_e64 v1, 0, v7, s2 v_cndmask_b32_e32 v3, 0, v2, vcc_lo v_cndmask_b32_e64 v4, 0, s7, vcc_lo v_cndmask_b32_e64 v2, 0, s7, s2 flat_load_b64 v[3:4], v[3:4] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 0x80, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 64, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 32, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 16, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] v_add_nc_u32_e32 v5, 8, v7 s_delay_alu instid0(VALU_DEP_1) v_cmp_ne_u32_e32 vcc_lo, -1, v5 v_cndmask_b32_e32 v5, 0, v5, vcc_lo v_cndmask_b32_e64 v6, 0, s7, vcc_lo flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 flat_load_b64 v[3:4], v[5:6] glc dlc s_waitcnt vmcnt(0) flat_load_b64 v[5:6], v[1:2] glc dlc s_waitcnt vmcnt(0) lgkmcnt(0) v_add_f64 v[3:4], v[3:4], v[5:6] flat_store_b64 v[1:2], v[3:4] dlc s_waitcnt_vscnt null, 0x0 .LBB0_12: s_or_b32 exec_lo, exec_lo, s3 s_waitcnt lgkmcnt(0) s_waitcnt_vscnt null, 0x0 s_barrier buffer_gl0_inv s_mov_b32 s2, exec_lo v_cmpx_eq_u32_e32 0, v0 s_cbranch_execz .LBB0_14 v_mov_b32_e32 v0, 0 s_load_b64 s[0:1], s[0:1], 0x8 s_mov_b32 s5, 0 v_mov_b32_e32 v2, 0 s_lshl_b64 s[2:3], s[4:5], 3 ds_load_b64 v[0:1], v0 s_waitcnt lgkmcnt(0) s_add_u32 s0, s0, s2 s_addc_u32 s1, s1, s3 global_store_b64 v2, v[0:1], s[0:1] .LBB0_14: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z7reduce6ILj512EEvPdS0_j .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 10 .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._Z7reduce6ILj512EEvPdS0_j,"axG",@progbits,_Z7reduce6ILj512EEvPdS0_j,comdat .Lfunc_end0: .size _Z7reduce6ILj512EEvPdS0_j, .Lfunc_end0-_Z7reduce6ILj512EEvPdS0_j .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 sdata .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: 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 - .offset: 144 .size: 4 .value_kind: hidden_dynamic_lds_size .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: _Z7reduce6ILj512EEvPdS0_j .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z7reduce6ILj512EEvPdS0_j.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 10 .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_000c4427_00000000-6_reduce.cudafe1.cpp" .text #APP .globl _ZSt21ios_base_library_initv #NO_APP .type _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j, @function _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j: .LFB3697: .cfi_startproc subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%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 12(%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 .L5 .L1: movq 120(%rsp), %rax subq %fs:40, %rax jne .L6 addq $136, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L5: .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 _Z7reduce6ILj512EEvPdS0_j(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 144 jmp .L1 .L6: call __stack_chk_fail@PLT .cfi_endproc .LFE3697: .size _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j, .-_ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j .section .text._Z7reduce6ILj512EEvPdS0_j,"axG",@progbits,_Z7reduce6ILj512EEvPdS0_j,comdat .weak _Z7reduce6ILj512EEvPdS0_j .type _Z7reduce6ILj512EEvPdS0_j, @function _Z7reduce6ILj512EEvPdS0_j: .LFB4000: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE4000: .size _Z7reduce6ILj512EEvPdS0_j, .-_Z7reduce6ILj512EEvPdS0_j .text .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB3675: .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 .LFE3675: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .section .rodata.str1.1,"aMS",@progbits,1 .LC1: .string "true" .LC2: .string "false" .LC3: .string "Iter %d, Blocks %d" .LC4: .string "=====\n" .LC5: .string "Input Size = %d\n" .LC6: .string "Device Result is: " .LC7: .string "Host Result is: " .LC8: .string "Result is: " .LC9: .string "Result correct? %s\n" .LC10: .string "Time for the kernel: %f ms\n" .LC11: .string "Time Full: %f ms\n" .text .globl main .type main, @function main: .LFB3672: .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 %fs:40, %rax movq %rax, 104(%rsp) xorl %eax, %eax movl $0, %r8d jmp .L12 .L38: movq $0x000000000, 8(%rsp) jmp .L14 .L51: movl 28(%rsp), %edx movq 40(%rsp), %rsi movq 32(%rsp), %rdi call _ZL39__device_stub__Z7reduce6ILj512EEvPdS0_jPdS_j jmp .L16 .L39: movq $0x000000000, (%rsp) jmp .L17 .L52: movq 104(%rsp), %rax subq %fs:40, %rax jne .L47 call _ZSt16__throw_bad_castv@PLT .L47: call __stack_chk_fail@PLT .L21: movq %r15, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq (%r15), %rax movl $10, %esi movq %r15, %rdi call *48(%rax) jmp .L22 .L53: movq 104(%rsp), %rax subq %fs:40, %rax jne .L48 call _ZSt16__throw_bad_castv@PLT .L48: call __stack_chk_fail@PLT .L25: movq %r15, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq (%r15), %rax movl $10, %esi movq %r15, %rdi call *48(%rax) jmp .L26 .L54: movq 104(%rsp), %rax subq %fs:40, %rax jne .L49 call _ZSt16__throw_bad_castv@PLT .L49: call __stack_chk_fail@PLT .L29: movq %r15, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq (%r15), %rax movl $10, %esi movq %r15, %rdi call *48(%rax) jmp .L30 .L40: leaq .LC2(%rip), %rdx .L31: leaq .LC9(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT leaq 80(%rsp), %rdi movq 56(%rsp), %rdx movq 48(%rsp), %rsi call cudaEventElapsedTime@PLT leaq 92(%rsp), %rdi movq 72(%rsp), %rdx movq 64(%rsp), %rsi call cudaEventElapsedTime@PLT pxor %xmm0, %xmm0 cvtss2sd 80(%rsp), %xmm0 leaq .LC10(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT pxor %xmm0, %xmm0 cvtss2sd 92(%rsp), %xmm0 leaq .LC11(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT movq %rbp, %rdi call free@PLT movq %r13, %rdi call free@PLT movq 32(%rsp), %rdi call cudaFree@PLT movq 40(%rsp), %rdi call cudaFree@PLT addl %r14d, %r14d cmpl $4, %r14d jg .L50 .L33: movl %r14d, %ecx movl 24(%rsp), %edx leaq .LC3(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %r14d, %r15d sall $5, %r15d leaq .LC4(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movl %r12d, %edx leaq .LC5(%rip), %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT movq 16(%rsp), %rdi call malloc@PLT movq %rax, %rbp movl $8192, %edi call malloc@PLT movq %rax, %r13 movq %rax, %rbx leaq 8192(%rax), %rdx .L13: movq $0x000000000, (%rax) addq $8, %rax cmpq %rdx, %rax jne .L13 testl %r12d, %r12d jle .L38 movl $0, %eax movq $0x000000000, 8(%rsp) .L15: pxor %xmm0, %xmm0 cvtsi2sdl %eax, %xmm0 movsd %xmm0, 0(%rbp,%rax,8) addsd 8(%rsp), %xmm0 movsd %xmm0, 8(%rsp) addq $1, %rax cmpl %eax, %r12d jg .L15 .L14: leaq 48(%rsp), %rdi call cudaEventCreate@PLT leaq 64(%rsp), %rdi call cudaEventCreate@PLT leaq 56(%rsp), %rdi call cudaEventCreate@PLT leaq 72(%rsp), %rdi call cudaEventCreate@PLT movl $0, %esi movq 64(%rsp), %rdi call cudaEventRecord@PLT leaq 32(%rsp), %rdi movq 16(%rsp), %rsi call cudaMalloc@PLT leaq 40(%rsp), %rdi movl $8192, %esi call cudaMalloc@PLT movl $1, %ecx movq 16(%rsp), %rdx movq %rbp, %rsi movq 32(%rsp), %rdi call cudaMemcpy@PLT movl $0, %esi movq 48(%rsp), %rdi call cudaEventRecord@PLT movl $1024, 92(%rsp) movl $1, 96(%rsp) movl $1, 100(%rsp) movl %r15d, 80(%rsp) movl $1, 84(%rsp) movl $1, 88(%rsp) movl $0, %r9d movl $8192, %r8d movq 92(%rsp), %rdx movl $1, %ecx movq 80(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L51 .L16: call cudaDeviceSynchronize@PLT movl $0, %esi movq 56(%rsp), %rdi call cudaEventRecord@PLT movq 56(%rsp), %rdi call cudaEventSynchronize@PLT movl $2, %ecx movl $8192, %edx movq 40(%rsp), %rsi movq %r13, %rdi call cudaMemcpy@PLT testl %r15d, %r15d jle .L39 leal -1(%r15), %eax leaq 8(%r13,%rax,8), %rax movq $0x000000000, (%rsp) .L18: movsd (%rsp), %xmm1 addsd (%rbx), %xmm1 movsd %xmm1, (%rsp) addq $8, %rbx cmpq %rax, %rbx jne .L18 .L17: movl $0, %esi movq 72(%rsp), %rdi call cudaEventRecord@PLT movq 72(%rsp), %rdi call cudaEventSynchronize@PLT movl $18, %edx leaq .LC6(%rip), %rsi leaq _ZSt4cout(%rip), %rbx movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movsd 0(%r13), %xmm0 movq %rbx, %rdi call _ZNSo9_M_insertIdEERSoT_@PLT movq %rax, %rbx movq (%rax), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %r15 testq %r15, %r15 je .L52 cmpb $0, 56(%r15) je .L21 movzbl 67(%r15), %eax .L22: movsbl %al, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT movl $16, %edx leaq .LC7(%rip), %rsi leaq _ZSt4cout(%rip), %rbx movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movsd 8(%rsp), %xmm0 movq %rbx, %rdi call _ZNSo9_M_insertIdEERSoT_@PLT movq %rax, %rbx movq (%rax), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %r15 testq %r15, %r15 je .L53 cmpb $0, 56(%r15) je .L25 movzbl 67(%r15), %eax .L26: movsbl %al, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT movl $11, %edx leaq .LC8(%rip), %rsi leaq _ZSt4cout(%rip), %rbx movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movsd (%rsp), %xmm0 movq %rbx, %rdi call _ZNSo9_M_insertIdEERSoT_@PLT movq %rax, %rbx movq (%rax), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %r15 testq %r15, %r15 je .L54 cmpb $0, 56(%r15) je .L29 movzbl 67(%r15), %eax .L30: movsbl %al, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT movsd 8(%rsp), %xmm3 movsd (%rsp), %xmm4 ucomisd %xmm4, %xmm3 jp .L40 leaq .LC1(%rip), %rdx je .L31 jmp .L40 .L50: movl 28(%rsp), %r9d movq 16(%rsp), %rcx addl $3276800, %r9d addq $26214400, %rcx cmpl $52428800, %r9d je .L34 .L36: movl %r9d, %r12d movl $1, %r14d movl %r9d, 28(%rsp) movq %rcx, 16(%rsp) jmp .L33 .L34: movl 24(%rsp), %r8d addl $1, %r8d cmpl $5, %r8d je .L35 .L12: movl $26214400, %ecx movl $3276800, %r9d movl %r8d, 24(%rsp) jmp .L36 .L35: movq 104(%rsp), %rax subq %fs:40, %rax jne .L55 movl $0, %eax 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 .L55: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE3672: .size main, .-main .section .rodata.str1.1 .LC12: .string "_Z7reduce6ILj512EEvPdS0_j" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB3700: .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 _Z7reduce6ILj512EEvPdS0_j(%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 .LFE3700: .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 "reduce.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl main # -- Begin function 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 $200, %rsp .cfi_def_cfa_offset 256 .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, %rax # imm = 0x100000000 addq $1024, %rax # imm = 0x400 movq %rax, 144(%rsp) # 8-byte Spill movl $0, 8(%rsp) # 4-byte Folded Spill jmp .LBB0_1 .p2align 4, 0x90 .LBB0_12: # in Loop: Header=BB0_1 Depth=1 movl 8(%rsp), %eax # 4-byte Reload incl %eax movl %eax, 8(%rsp) # 4-byte Spill cmpl $5, %eax je .LBB0_13 .LBB0_1: # %.preheader75 # =>This Loop Header: Depth=1 # Child Loop BB0_2 Depth 2 # Child Loop BB0_3 Depth 3 # Child Loop BB0_4 Depth 4 # Child Loop BB0_8 Depth 4 movl $3276800, %r13d # imm = 0x320000 movl $1, %eax jmp .LBB0_2 .p2align 4, 0x90 .LBB0_24: # in Loop: Header=BB0_2 Depth=2 movq 136(%rsp), %rax # 8-byte Reload incq %rax addq $3276800, %r13 # imm = 0x320000 cmpq $16, %rax je .LBB0_12 .LBB0_2: # %.preheader74 # Parent Loop BB0_1 Depth=1 # => This Loop Header: Depth=2 # Child Loop BB0_3 Depth 3 # Child Loop BB0_4 Depth 4 # Child Loop BB0_8 Depth 4 imull $3276800, %eax, %ecx # imm = 0x320000 movl %ecx, 12(%rsp) # 4-byte Spill movq %rax, 136(%rsp) # 8-byte Spill imulq $26214400, %rax, %rax # imm = 0x1900000 movq %rax, 72(%rsp) # 8-byte Spill movl $1, %r12d jmp .LBB0_3 .p2align 4, 0x90 .LBB0_22: # in Loop: Header=BB0_3 Depth=3 movq %r14, %rdi movq %rax, %rbp callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) movl %eax, %ecx movq %rbp, %rax .LBB0_23: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit70 # in Loop: Header=BB0_3 Depth=3 movsbl %cl, %esi movq %rax, %rdi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movsd 80(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero ucomisd 88(%rsp), %xmm0 # 8-byte Folded Reload movl $.L.str.7, %esi movl $.L.str.8, %eax cmovneq %rax, %rsi cmovpq %rax, %rsi movl $.L.str.6, %edi xorl %eax, %eax callq printf movq 104(%rsp), %rsi movq 24(%rsp), %rdx leaq 112(%rsp), %rdi callq hipEventElapsedTime movq 96(%rsp), %rsi movq 16(%rsp), %rdx leaq 56(%rsp), %rdi callq hipEventElapsedTime movss 112(%rsp), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str.9, %edi movb $1, %al callq printf movss 56(%rsp), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str.10, %edi movb $1, %al callq printf movq %r15, %rdi callq free movq %rbx, %rdi callq free movq 40(%rsp), %rdi callq hipFree movq 32(%rsp), %rdi callq hipFree leal (%r12,%r12), %eax cmpl $3, %r12d movl %eax, %r12d jae .LBB0_24 .LBB0_3: # %.preheader.preheader # Parent Loop BB0_1 Depth=1 # Parent Loop BB0_2 Depth=2 # => This Loop Header: Depth=3 # Child Loop BB0_4 Depth 4 # Child Loop BB0_8 Depth 4 movl $.L.str, %edi movl 8(%rsp), %esi # 4-byte Reload movl %r12d, %edx xorl %eax, %eax callq printf movl $.Lstr, %edi callq puts@PLT movl $.L.str.2, %edi movl 12(%rsp), %esi # 4-byte Reload xorl %eax, %eax callq printf movq 72(%rsp), %rdi # 8-byte Reload callq malloc movq %rax, %r15 movl $8192, %edi # imm = 0x2000 callq malloc movq %rax, %rbx movl $8192, %edx # imm = 0x2000 movq %rax, %rdi xorl %esi, %esi callq memset@PLT xorpd %xmm1, %xmm1 xorl %eax, %eax .p2align 4, 0x90 .LBB0_4: # %.preheader # Parent Loop BB0_1 Depth=1 # Parent Loop BB0_2 Depth=2 # Parent Loop BB0_3 Depth=3 # => This Inner Loop Header: Depth=4 xorps %xmm0, %xmm0 cvtsi2sd %eax, %xmm0 movsd %xmm0, (%r15,%rax,8) addsd %xmm0, %xmm1 incq %rax cmpq %rax, %r13 jne .LBB0_4 # %bb.5: # in Loop: Header=BB0_3 Depth=3 movsd %xmm1, 88(%rsp) # 8-byte Spill movl %r12d, %r14d shll $5, %r14d leaq 104(%rsp), %rdi callq hipEventCreate leaq 96(%rsp), %rdi callq hipEventCreate leaq 24(%rsp), %rdi callq hipEventCreate leaq 16(%rsp), %rdi callq hipEventCreate movq 96(%rsp), %rdi xorl %esi, %esi callq hipEventRecord leaq 40(%rsp), %rdi movq 72(%rsp), %rbp # 8-byte Reload movq %rbp, %rsi callq hipMalloc movl $8192, %esi # imm = 0x2000 leaq 32(%rsp), %rdi callq hipMalloc movq 40(%rsp), %rdi movq %r15, %rsi movq %rbp, %rdx movl $1, %ecx callq hipMemcpy movq 104(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movabsq $4294967296, %rax # imm = 0x100000000 orq %rax, %r14 movl $8192, %r8d # imm = 0x2000 movq %r14, %rdi movl $1, %esi movq 144(%rsp), %rdx # 8-byte Reload movl $1, %ecx xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB0_7 # %bb.6: # in Loop: Header=BB0_3 Depth=3 movq 40(%rsp), %rax movq 32(%rsp), %rcx movq %rax, 192(%rsp) movq %rcx, 184(%rsp) movl 12(%rsp), %eax # 4-byte Reload movl %eax, 52(%rsp) leaq 192(%rsp), %rax movq %rax, 112(%rsp) leaq 184(%rsp), %rax movq %rax, 120(%rsp) leaq 52(%rsp), %rax movq %rax, 128(%rsp) leaq 56(%rsp), %rdi leaq 168(%rsp), %rsi leaq 160(%rsp), %rdx leaq 152(%rsp), %rcx callq __hipPopCallConfiguration movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 168(%rsp), %rcx movl 176(%rsp), %r8d movl $_Z7reduce6ILj512EEvPdS0_j, %edi leaq 112(%rsp), %r9 pushq 152(%rsp) .cfi_adjust_cfa_offset 8 pushq 168(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB0_7: # in Loop: Header=BB0_3 Depth=3 callq hipDeviceSynchronize movq 24(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq 24(%rsp), %rdi callq hipEventSynchronize movq 32(%rsp), %rsi movl $8192, %edx # imm = 0x2000 movq %rbx, %rdi movl $2, %ecx callq hipMemcpy movl %r12d, %eax shlq $8, %rax xorpd %xmm0, %xmm0 xorl %ecx, %ecx .p2align 4, 0x90 .LBB0_8: # Parent Loop BB0_1 Depth=1 # Parent Loop BB0_2 Depth=2 # Parent Loop BB0_3 Depth=3 # => This Inner Loop Header: Depth=4 addsd (%rbx,%rcx), %xmm0 addq $8, %rcx cmpq %rcx, %rax jne .LBB0_8 # %bb.9: # in Loop: Header=BB0_3 Depth=3 movsd %xmm0, 80(%rsp) # 8-byte Spill movq 16(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movq 16(%rsp), %rdi callq hipEventSynchronize movl $_ZSt4cout, %edi movl $.L.str.3, %esi movl $18, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movsd (%rbx), %xmm0 # xmm0 = mem[0],zero movl $_ZSt4cout, %edi callq _ZNSo9_M_insertIdEERSoT_ movq (%rax), %rcx movq -24(%rcx), %rcx movq 240(%rax,%rcx), %r14 testq %r14, %r14 je .LBB0_25 # %bb.10: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i # in Loop: Header=BB0_3 Depth=3 cmpb $0, 56(%r14) je .LBB0_14 # %bb.11: # in Loop: Header=BB0_3 Depth=3 movzbl 67(%r14), %ecx jmp .LBB0_15 .p2align 4, 0x90 .LBB0_14: # in Loop: Header=BB0_3 Depth=3 movq %r14, %rdi movq %rax, %rbp callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) movl %eax, %ecx movq %rbp, %rax .LBB0_15: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit # in Loop: Header=BB0_3 Depth=3 movsbl %cl, %esi movq %rax, %rdi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movl $_ZSt4cout, %edi movl $.L.str.4, %esi movl $16, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl $_ZSt4cout, %edi movsd 88(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero callq _ZNSo9_M_insertIdEERSoT_ movq (%rax), %rcx movq -24(%rcx), %rcx movq 240(%rax,%rcx), %r14 testq %r14, %r14 je .LBB0_25 # %bb.16: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i62 # in Loop: Header=BB0_3 Depth=3 cmpb $0, 56(%r14) je .LBB0_18 # %bb.17: # in Loop: Header=BB0_3 Depth=3 movzbl 67(%r14), %ecx jmp .LBB0_19 .p2align 4, 0x90 .LBB0_18: # in Loop: Header=BB0_3 Depth=3 movq %r14, %rdi movq %rax, %rbp callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) movl %eax, %ecx movq %rbp, %rax .LBB0_19: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit65 # in Loop: Header=BB0_3 Depth=3 movsbl %cl, %esi movq %rax, %rdi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movl $_ZSt4cout, %edi movl $.L.str.5, %esi movl $11, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl $_ZSt4cout, %edi movsd 80(%rsp), %xmm0 # 8-byte Reload # xmm0 = mem[0],zero callq _ZNSo9_M_insertIdEERSoT_ movq (%rax), %rcx movq -24(%rcx), %rcx movq 240(%rax,%rcx), %r14 testq %r14, %r14 je .LBB0_25 # %bb.20: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i67 # in Loop: Header=BB0_3 Depth=3 cmpb $0, 56(%r14) je .LBB0_22 # %bb.21: # in Loop: Header=BB0_3 Depth=3 movzbl 67(%r14), %ecx jmp .LBB0_23 .LBB0_13: xorl %eax, %eax addq $200, %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 .LBB0_25: .cfi_def_cfa_offset 256 callq _ZSt16__throw_bad_castv .Lfunc_end0: .size main, .Lfunc_end0-main .cfi_endproc # -- End function .section .text._Z22__device_stub__reduce6ILj512EEvPdS0_j,"axG",@progbits,_Z22__device_stub__reduce6ILj512EEvPdS0_j,comdat .weak _Z22__device_stub__reduce6ILj512EEvPdS0_j # -- Begin function _Z22__device_stub__reduce6ILj512EEvPdS0_j .p2align 4, 0x90 .type _Z22__device_stub__reduce6ILj512EEvPdS0_j,@function _Z22__device_stub__reduce6ILj512EEvPdS0_j: # @_Z22__device_stub__reduce6ILj512EEvPdS0_j .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%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 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 $_Z7reduce6ILj512EEvPdS0_j, %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_end1: .size _Z22__device_stub__reduce6ILj512EEvPdS0_j, .Lfunc_end1-_Z22__device_stub__reduce6ILj512EEvPdS0_j .cfi_endproc # -- End function .text .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 $_Z7reduce6ILj512EEvPdS0_j, %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 .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "Iter %d, Blocks %d" .size .L.str, 19 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "Input Size = %d\n" .size .L.str.2, 17 .type _Z7reduce6ILj512EEvPdS0_j,@object # @_Z7reduce6ILj512EEvPdS0_j .section .rodata._Z7reduce6ILj512EEvPdS0_j,"aG",@progbits,_Z7reduce6ILj512EEvPdS0_j,comdat .weak _Z7reduce6ILj512EEvPdS0_j .p2align 3, 0x0 _Z7reduce6ILj512EEvPdS0_j: .quad _Z22__device_stub__reduce6ILj512EEvPdS0_j .size _Z7reduce6ILj512EEvPdS0_j, 8 .type .L.str.3,@object # @.str.3 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.3: .asciz "Device Result is: " .size .L.str.3, 19 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz "Host Result is: " .size .L.str.4, 17 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "Result is: " .size .L.str.5, 12 .type .L.str.6,@object # @.str.6 .L.str.6: .asciz "Result correct? %s\n" .size .L.str.6, 20 .type .L.str.7,@object # @.str.7 .L.str.7: .asciz "true" .size .L.str.7, 5 .type .L.str.8,@object # @.str.8 .L.str.8: .asciz "false" .size .L.str.8, 6 .type .L.str.9,@object # @.str.9 .L.str.9: .asciz "Time for the kernel: %f ms\n" .size .L.str.9, 28 .type .L.str.10,@object # @.str.10 .L.str.10: .asciz "Time Full: %f ms\n" .size .L.str.10, 18 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z7reduce6ILj512EEvPdS0_j" .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 .type .Lstr,@object # @str .section .rodata.str1.1,"aMS",@progbits,1 .Lstr: .asciz "=====" .size .Lstr, 6 .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 _Z22__device_stub__reduce6ILj512EEvPdS0_j .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z7reduce6ILj512EEvPdS0_j .addrsig_sym _ZSt4cout .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 reduceNeighboredLess(int *g_idata, int *g_odata, unsigned int n) { // set the thread id. unsigned int tid = threadIdx.x; unsigned int idx = threadIdx.x + blockIdx.x * blockDim.x; // convert global data pointer to the local pointer of this block. int *idata = g_idata + blockIdx.x * blockDim.x; // boundary check. if (idx >= n) return; for (int stride = 1; stride < blockDim.x; stride *= 2) { // convert tid into local array index. int index = 2 * stride * tid; if (index < blockDim.x) { idata[index] += idata[index + stride]; } // synchronize within threadblock. __syncthreads(); } // write result for this block to global mem. if (tid == 0) { g_odata[blockIdx.x] = idata[0]; } }
code for sm_80 Function : _Z20reduceNeighboredLessPiS_j .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 R8, SR_CTAID.X ; /* 0x0000000000087919 */ /* 0x000e280000002500 */ /*0020*/ S2R R11, SR_TID.X ; /* 0x00000000000b7919 */ /* 0x000e620000002100 */ /*0030*/ IMAD R2, R8, c[0x0][0x0], RZ ; /* 0x0000000008027a24 */ /* 0x001fc800078e02ff */ /*0040*/ IMAD.IADD R0, R2, 0x1, R11 ; /* 0x0000000102007824 */ /* 0x002fca00078e020b */ /*0050*/ ISETP.GE.U32.AND P0, PT, R0, c[0x0][0x170], PT ; /* 0x00005c0000007a0c */ /* 0x000fda0003f06070 */ /*0060*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0070*/ MOV R0, c[0x0][0x0] ; /* 0x0000000000007a02 */ /* 0x000fe20000000f00 */ /*0080*/ IMAD.MOV.U32 R3, RZ, RZ, 0x4 ; /* 0x00000004ff037424 */ /* 0x000fe200078e00ff */ /*0090*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe40000000a00 */ /*00a0*/ ISETP.GE.U32.AND P0, PT, R0, 0x2, PT ; /* 0x000000020000780c */ /* 0x000fe20003f06070 */ /*00b0*/ IMAD.WIDE.U32 R2, R2, R3, c[0x0][0x160] ; /* 0x0000580002027625 */ /* 0x000fd800078e0003 */ /*00c0*/ @!P0 BRA 0x1f0 ; /* 0x0000012000008947 */ /* 0x000fea0003800000 */ /*00d0*/ HFMA2.MMA R0, -RZ, RZ, 0, 5.9604644775390625e-08 ; /* 0x00000001ff007435 */ /* 0x000fd400000001ff */ /*00e0*/ IMAD.SHL.U32 R10, R0, 0x2, RZ ; /* 0x00000002000a7824 */ /* 0x000fe200078e00ff */ /*00f0*/ BSSY B0, 0x1b0 ; /* 0x000000b000007945 */ /* 0x000fe60003800000 */ /*0100*/ IMAD R7, R10, R11, RZ ; /* 0x0000000b0a077224 */ /* 0x001fca00078e02ff */ /*0110*/ ISETP.GE.U32.AND P0, PT, R7, c[0x0][0x0], PT ; /* 0x0000000007007a0c */ /* 0x000fda0003f06070 */ /*0120*/ @P0 BRA 0x1a0 ; /* 0x0000007000000947 */ /* 0x000fea0003800000 */ /*0130*/ IADD3 R5, R7.reuse, R0, RZ ; /* 0x0000000007057210 */ /* 0x040fe20007ffe0ff */ /*0140*/ IMAD.WIDE R6, R7, 0x4, R2 ; /* 0x0000000407067825 */ /* 0x000fc800078e0202 */ /*0150*/ IMAD.WIDE R4, R5, 0x4, R2 ; /* 0x0000000405047825 */ /* 0x000fe200078e0202 */ /*0160*/ LDG.E R0, [R6.64] ; /* 0x0000000406007981 */ /* 0x000eaa000c1e1900 */ /*0170*/ LDG.E R5, [R4.64] ; /* 0x0000000404057981 */ /* 0x000ea4000c1e1900 */ /*0180*/ IMAD.IADD R9, R0, 0x1, R5 ; /* 0x0000000100097824 */ /* 0x004fca00078e0205 */ /*0190*/ STG.E [R6.64], R9 ; /* 0x0000000906007986 */ /* 0x0001e4000c101904 */ /*01a0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*01b0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe20000010000 */ /*01c0*/ ISETP.GE.U32.AND P0, PT, R10, c[0x0][0x0], PT ; /* 0x000000000a007a0c */ /* 0x000fe40003f06070 */ /*01d0*/ MOV R0, R10 ; /* 0x0000000a00007202 */ /* 0x000fd60000000f00 */ /*01e0*/ @!P0 BRA 0xe0 ; /* 0xfffffef000008947 */ /* 0x000fea000383ffff */ /*01f0*/ ISETP.NE.AND P0, PT, R11, RZ, PT ; /* 0x000000ff0b00720c */ /* 0x000fda0003f05270 */ /*0200*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0210*/ LDG.E R3, [R2.64] ; /* 0x0000000402037981 */ /* 0x000ea2000c1e1900 */ /*0220*/ HFMA2.MMA R5, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff057435 */ /* 0x000fd400000001ff */ /*0230*/ IMAD.WIDE.U32 R4, R8, R5, c[0x0][0x168] ; /* 0x00005a0008047625 */ /* 0x000fca00078e0005 */ /*0240*/ STG.E [R4.64], R3 ; /* 0x0000000304007986 */ /* 0x004fe2000c101904 */ /*0250*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0260*/ BRA 0x260; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0270*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0280*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0290*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include "includes.h" __global__ void reduceNeighboredLess(int *g_idata, int *g_odata, unsigned int n) { // set the thread id. unsigned int tid = threadIdx.x; unsigned int idx = threadIdx.x + blockIdx.x * blockDim.x; // convert global data pointer to the local pointer of this block. int *idata = g_idata + blockIdx.x * blockDim.x; // boundary check. if (idx >= n) return; for (int stride = 1; stride < blockDim.x; stride *= 2) { // convert tid into local array index. int index = 2 * stride * tid; if (index < blockDim.x) { idata[index] += idata[index + stride]; } // synchronize within threadblock. __syncthreads(); } // write result for this block to global mem. if (tid == 0) { g_odata[blockIdx.x] = idata[0]; } }
.file "tmpxft_00196245_00000000-6_reduceNeighboredLess.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 _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j .type _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j, @function _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j: .LFB2051: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%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 12(%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 _Z20reduceNeighboredLessPiS_j(%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 _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j, .-_Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j .globl _Z20reduceNeighboredLessPiS_j .type _Z20reduceNeighboredLessPiS_j, @function _Z20reduceNeighboredLessPiS_j: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z20reduceNeighboredLessPiS_j, .-_Z20reduceNeighboredLessPiS_j .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z20reduceNeighboredLessPiS_j" .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 _Z20reduceNeighboredLessPiS_j(%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 reduceNeighboredLess(int *g_idata, int *g_odata, unsigned int n) { // set the thread id. unsigned int tid = threadIdx.x; unsigned int idx = threadIdx.x + blockIdx.x * blockDim.x; // convert global data pointer to the local pointer of this block. int *idata = g_idata + blockIdx.x * blockDim.x; // boundary check. if (idx >= n) return; for (int stride = 1; stride < blockDim.x; stride *= 2) { // convert tid into local array index. int index = 2 * stride * tid; if (index < blockDim.x) { idata[index] += idata[index + stride]; } // synchronize within threadblock. __syncthreads(); } // write result for this block to global mem. if (tid == 0) { g_odata[blockIdx.x] = idata[0]; } }
#include <hip/hip_runtime.h> #include "includes.h" __global__ void reduceNeighboredLess(int *g_idata, int *g_odata, unsigned int n) { // set the thread id. unsigned int tid = threadIdx.x; unsigned int idx = threadIdx.x + blockIdx.x * blockDim.x; // convert global data pointer to the local pointer of this block. int *idata = g_idata + blockIdx.x * blockDim.x; // boundary check. if (idx >= n) return; for (int stride = 1; stride < blockDim.x; stride *= 2) { // convert tid into local array index. int index = 2 * stride * tid; if (index < blockDim.x) { idata[index] += idata[index + stride]; } // synchronize within threadblock. __syncthreads(); } // write result for this block to global mem. if (tid == 0) { g_odata[blockIdx.x] = idata[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 reduceNeighboredLess(int *g_idata, int *g_odata, unsigned int n) { // set the thread id. unsigned int tid = threadIdx.x; unsigned int idx = threadIdx.x + blockIdx.x * blockDim.x; // convert global data pointer to the local pointer of this block. int *idata = g_idata + blockIdx.x * blockDim.x; // boundary check. if (idx >= n) return; for (int stride = 1; stride < blockDim.x; stride *= 2) { // convert tid into local array index. int index = 2 * stride * tid; if (index < blockDim.x) { idata[index] += idata[index + stride]; } // synchronize within threadblock. __syncthreads(); } // write result for this block to global mem. if (tid == 0) { g_odata[blockIdx.x] = idata[0]; } }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z20reduceNeighboredLessPiS_j .globl _Z20reduceNeighboredLessPiS_j .p2align 8 .type _Z20reduceNeighboredLessPiS_j,@function _Z20reduceNeighboredLessPiS_j: s_clause 0x1 s_load_b32 s3, s[0:1], 0x24 s_load_b32 s5, s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_and_b32 s3, s3, 0xffff s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_mul_i32 s4, s15, s3 v_add_nc_u32_e32 v1, s4, v0 s_delay_alu instid0(VALU_DEP_1) v_cmp_gt_u32_e32 vcc_lo, s5, v1 s_and_saveexec_b32 s5, vcc_lo s_cbranch_execz .LBB0_8 s_load_b64 s[6:7], s[0:1], 0x0 s_mov_b32 s5, 0 s_mov_b32 s2, s15 s_lshl_b64 s[4:5], s[4:5], 2 s_waitcnt lgkmcnt(0) s_add_u32 s4, s6, s4 s_addc_u32 s5, s7, s5 s_cmp_lt_u32 s3, 2 s_cbranch_scc1 .LBB0_6 s_mov_b32 s8, 1 s_set_inst_prefetch_distance 0x1 s_branch .LBB0_4 .p2align 6 .LBB0_3: s_or_b32 exec_lo, exec_lo, s7 s_cmp_ge_u32 s6, s3 s_mov_b32 s8, s6 s_waitcnt_vscnt null, 0x0 s_barrier buffer_gl0_inv s_cbranch_scc1 .LBB0_6 .LBB0_4: s_lshl_b32 s6, s8, 1 s_mov_b32 s7, exec_lo v_mul_lo_u32 v1, s6, v0 s_delay_alu instid0(VALU_DEP_1) v_cmpx_gt_u32_e64 s3, v1 s_cbranch_execz .LBB0_3 v_dual_mov_b32 v4, 0 :: v_dual_add_nc_u32 v3, s8, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_mov_b32_e32 v2, v4 v_lshlrev_b64 v[3:4], 2, v[3:4] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[1:2], 2, v[1:2] v_add_co_u32 v3, vcc_lo, s4, v3 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_ci_u32_e32 v4, vcc_lo, s5, v4, vcc_lo v_add_co_u32 v1, vcc_lo, s4, v1 s_delay_alu instid0(VALU_DEP_4) v_add_co_ci_u32_e32 v2, vcc_lo, s5, v2, vcc_lo s_clause 0x1 global_load_b32 v3, v[3:4], off global_load_b32 v4, v[1:2], off s_waitcnt vmcnt(0) v_add_nc_u32_e32 v3, v4, v3 global_store_b32 v[1:2], v3, off s_branch .LBB0_3 .LBB0_6: s_set_inst_prefetch_distance 0x2 v_cmp_eq_u32_e32 vcc_lo, 0, v0 s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_8 v_mov_b32_e32 v0, 0 s_load_b64 s[0:1], s[0:1], 0x8 s_mov_b32 s3, 0 s_delay_alu instid0(SALU_CYCLE_1) s_lshl_b64 s[2:3], s[2:3], 2 global_load_b32 v1, v0, s[4:5] s_waitcnt lgkmcnt(0) s_add_u32 s0, s0, s2 s_addc_u32 s1, s1, s3 s_waitcnt vmcnt(0) global_store_b32 v0, v1, s[0:1] .LBB0_8: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z20reduceNeighboredLessPiS_j .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 5 .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 _Z20reduceNeighboredLessPiS_j, .Lfunc_end0-_Z20reduceNeighboredLessPiS_j .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: 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: _Z20reduceNeighboredLessPiS_j .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z20reduceNeighboredLessPiS_j.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 5 .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 reduceNeighboredLess(int *g_idata, int *g_odata, unsigned int n) { // set the thread id. unsigned int tid = threadIdx.x; unsigned int idx = threadIdx.x + blockIdx.x * blockDim.x; // convert global data pointer to the local pointer of this block. int *idata = g_idata + blockIdx.x * blockDim.x; // boundary check. if (idx >= n) return; for (int stride = 1; stride < blockDim.x; stride *= 2) { // convert tid into local array index. int index = 2 * stride * tid; if (index < blockDim.x) { idata[index] += idata[index + stride]; } // synchronize within threadblock. __syncthreads(); } // write result for this block to global mem. if (tid == 0) { g_odata[blockIdx.x] = idata[0]; } }
.text .file "reduceNeighboredLess.hip" .globl _Z35__device_stub__reduceNeighboredLessPiS_j # -- Begin function _Z35__device_stub__reduceNeighboredLessPiS_j .p2align 4, 0x90 .type _Z35__device_stub__reduceNeighboredLessPiS_j,@function _Z35__device_stub__reduceNeighboredLessPiS_j: # @_Z35__device_stub__reduceNeighboredLessPiS_j .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%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 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 $_Z20reduceNeighboredLessPiS_j, %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 _Z35__device_stub__reduceNeighboredLessPiS_j, .Lfunc_end0-_Z35__device_stub__reduceNeighboredLessPiS_j .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 $_Z20reduceNeighboredLessPiS_j, %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 _Z20reduceNeighboredLessPiS_j,@object # @_Z20reduceNeighboredLessPiS_j .section .rodata,"a",@progbits .globl _Z20reduceNeighboredLessPiS_j .p2align 3, 0x0 _Z20reduceNeighboredLessPiS_j: .quad _Z35__device_stub__reduceNeighboredLessPiS_j .size _Z20reduceNeighboredLessPiS_j, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z20reduceNeighboredLessPiS_j" .size .L__unnamed_1, 30 .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 _Z35__device_stub__reduceNeighboredLessPiS_j .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z20reduceNeighboredLessPiS_j .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 : _Z20reduceNeighboredLessPiS_j .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 R8, SR_CTAID.X ; /* 0x0000000000087919 */ /* 0x000e280000002500 */ /*0020*/ S2R R11, SR_TID.X ; /* 0x00000000000b7919 */ /* 0x000e620000002100 */ /*0030*/ IMAD R2, R8, c[0x0][0x0], RZ ; /* 0x0000000008027a24 */ /* 0x001fc800078e02ff */ /*0040*/ IMAD.IADD R0, R2, 0x1, R11 ; /* 0x0000000102007824 */ /* 0x002fca00078e020b */ /*0050*/ ISETP.GE.U32.AND P0, PT, R0, c[0x0][0x170], PT ; /* 0x00005c0000007a0c */ /* 0x000fda0003f06070 */ /*0060*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0070*/ MOV R0, c[0x0][0x0] ; /* 0x0000000000007a02 */ /* 0x000fe20000000f00 */ /*0080*/ IMAD.MOV.U32 R3, RZ, RZ, 0x4 ; /* 0x00000004ff037424 */ /* 0x000fe200078e00ff */ /*0090*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe40000000a00 */ /*00a0*/ ISETP.GE.U32.AND P0, PT, R0, 0x2, PT ; /* 0x000000020000780c */ /* 0x000fe20003f06070 */ /*00b0*/ IMAD.WIDE.U32 R2, R2, R3, c[0x0][0x160] ; /* 0x0000580002027625 */ /* 0x000fd800078e0003 */ /*00c0*/ @!P0 BRA 0x1f0 ; /* 0x0000012000008947 */ /* 0x000fea0003800000 */ /*00d0*/ HFMA2.MMA R0, -RZ, RZ, 0, 5.9604644775390625e-08 ; /* 0x00000001ff007435 */ /* 0x000fd400000001ff */ /*00e0*/ IMAD.SHL.U32 R10, R0, 0x2, RZ ; /* 0x00000002000a7824 */ /* 0x000fe200078e00ff */ /*00f0*/ BSSY B0, 0x1b0 ; /* 0x000000b000007945 */ /* 0x000fe60003800000 */ /*0100*/ IMAD R7, R10, R11, RZ ; /* 0x0000000b0a077224 */ /* 0x001fca00078e02ff */ /*0110*/ ISETP.GE.U32.AND P0, PT, R7, c[0x0][0x0], PT ; /* 0x0000000007007a0c */ /* 0x000fda0003f06070 */ /*0120*/ @P0 BRA 0x1a0 ; /* 0x0000007000000947 */ /* 0x000fea0003800000 */ /*0130*/ IADD3 R5, R7.reuse, R0, RZ ; /* 0x0000000007057210 */ /* 0x040fe20007ffe0ff */ /*0140*/ IMAD.WIDE R6, R7, 0x4, R2 ; /* 0x0000000407067825 */ /* 0x000fc800078e0202 */ /*0150*/ IMAD.WIDE R4, R5, 0x4, R2 ; /* 0x0000000405047825 */ /* 0x000fe200078e0202 */ /*0160*/ LDG.E R0, [R6.64] ; /* 0x0000000406007981 */ /* 0x000eaa000c1e1900 */ /*0170*/ LDG.E R5, [R4.64] ; /* 0x0000000404057981 */ /* 0x000ea4000c1e1900 */ /*0180*/ IMAD.IADD R9, R0, 0x1, R5 ; /* 0x0000000100097824 */ /* 0x004fca00078e0205 */ /*0190*/ STG.E [R6.64], R9 ; /* 0x0000000906007986 */ /* 0x0001e4000c101904 */ /*01a0*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x000fea0003800000 */ /*01b0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe20000010000 */ /*01c0*/ ISETP.GE.U32.AND P0, PT, R10, c[0x0][0x0], PT ; /* 0x000000000a007a0c */ /* 0x000fe40003f06070 */ /*01d0*/ MOV R0, R10 ; /* 0x0000000a00007202 */ /* 0x000fd60000000f00 */ /*01e0*/ @!P0 BRA 0xe0 ; /* 0xfffffef000008947 */ /* 0x000fea000383ffff */ /*01f0*/ ISETP.NE.AND P0, PT, R11, RZ, PT ; /* 0x000000ff0b00720c */ /* 0x000fda0003f05270 */ /*0200*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0210*/ LDG.E R3, [R2.64] ; /* 0x0000000402037981 */ /* 0x000ea2000c1e1900 */ /*0220*/ HFMA2.MMA R5, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff057435 */ /* 0x000fd400000001ff */ /*0230*/ IMAD.WIDE.U32 R4, R8, R5, c[0x0][0x168] ; /* 0x00005a0008047625 */ /* 0x000fca00078e0005 */ /*0240*/ STG.E [R4.64], R3 ; /* 0x0000000304007986 */ /* 0x004fe2000c101904 */ /*0250*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0260*/ BRA 0x260; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0270*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0280*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0290*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*02f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z20reduceNeighboredLessPiS_j .globl _Z20reduceNeighboredLessPiS_j .p2align 8 .type _Z20reduceNeighboredLessPiS_j,@function _Z20reduceNeighboredLessPiS_j: s_clause 0x1 s_load_b32 s3, s[0:1], 0x24 s_load_b32 s5, s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_and_b32 s3, s3, 0xffff s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_mul_i32 s4, s15, s3 v_add_nc_u32_e32 v1, s4, v0 s_delay_alu instid0(VALU_DEP_1) v_cmp_gt_u32_e32 vcc_lo, s5, v1 s_and_saveexec_b32 s5, vcc_lo s_cbranch_execz .LBB0_8 s_load_b64 s[6:7], s[0:1], 0x0 s_mov_b32 s5, 0 s_mov_b32 s2, s15 s_lshl_b64 s[4:5], s[4:5], 2 s_waitcnt lgkmcnt(0) s_add_u32 s4, s6, s4 s_addc_u32 s5, s7, s5 s_cmp_lt_u32 s3, 2 s_cbranch_scc1 .LBB0_6 s_mov_b32 s8, 1 s_set_inst_prefetch_distance 0x1 s_branch .LBB0_4 .p2align 6 .LBB0_3: s_or_b32 exec_lo, exec_lo, s7 s_cmp_ge_u32 s6, s3 s_mov_b32 s8, s6 s_waitcnt_vscnt null, 0x0 s_barrier buffer_gl0_inv s_cbranch_scc1 .LBB0_6 .LBB0_4: s_lshl_b32 s6, s8, 1 s_mov_b32 s7, exec_lo v_mul_lo_u32 v1, s6, v0 s_delay_alu instid0(VALU_DEP_1) v_cmpx_gt_u32_e64 s3, v1 s_cbranch_execz .LBB0_3 v_dual_mov_b32 v4, 0 :: v_dual_add_nc_u32 v3, s8, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_mov_b32_e32 v2, v4 v_lshlrev_b64 v[3:4], 2, v[3:4] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[1:2], 2, v[1:2] v_add_co_u32 v3, vcc_lo, s4, v3 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_ci_u32_e32 v4, vcc_lo, s5, v4, vcc_lo v_add_co_u32 v1, vcc_lo, s4, v1 s_delay_alu instid0(VALU_DEP_4) v_add_co_ci_u32_e32 v2, vcc_lo, s5, v2, vcc_lo s_clause 0x1 global_load_b32 v3, v[3:4], off global_load_b32 v4, v[1:2], off s_waitcnt vmcnt(0) v_add_nc_u32_e32 v3, v4, v3 global_store_b32 v[1:2], v3, off s_branch .LBB0_3 .LBB0_6: s_set_inst_prefetch_distance 0x2 v_cmp_eq_u32_e32 vcc_lo, 0, v0 s_and_b32 exec_lo, exec_lo, vcc_lo s_cbranch_execz .LBB0_8 v_mov_b32_e32 v0, 0 s_load_b64 s[0:1], s[0:1], 0x8 s_mov_b32 s3, 0 s_delay_alu instid0(SALU_CYCLE_1) s_lshl_b64 s[2:3], s[2:3], 2 global_load_b32 v1, v0, s[4:5] s_waitcnt lgkmcnt(0) s_add_u32 s0, s0, s2 s_addc_u32 s1, s1, s3 s_waitcnt vmcnt(0) global_store_b32 v0, v1, s[0:1] .LBB0_8: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z20reduceNeighboredLessPiS_j .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 5 .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 _Z20reduceNeighboredLessPiS_j, .Lfunc_end0-_Z20reduceNeighboredLessPiS_j .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: 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: _Z20reduceNeighboredLessPiS_j .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z20reduceNeighboredLessPiS_j.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 5 .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_00196245_00000000-6_reduceNeighboredLess.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 _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j .type _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j, @function _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j: .LFB2051: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%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 12(%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 _Z20reduceNeighboredLessPiS_j(%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 _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j, .-_Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j .globl _Z20reduceNeighboredLessPiS_j .type _Z20reduceNeighboredLessPiS_j, @function _Z20reduceNeighboredLessPiS_j: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z43__device_stub__Z20reduceNeighboredLessPiS_jPiS_j addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z20reduceNeighboredLessPiS_j, .-_Z20reduceNeighboredLessPiS_j .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z20reduceNeighboredLessPiS_j" .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 _Z20reduceNeighboredLessPiS_j(%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 "reduceNeighboredLess.hip" .globl _Z35__device_stub__reduceNeighboredLessPiS_j # -- Begin function _Z35__device_stub__reduceNeighboredLessPiS_j .p2align 4, 0x90 .type _Z35__device_stub__reduceNeighboredLessPiS_j,@function _Z35__device_stub__reduceNeighboredLessPiS_j: # @_Z35__device_stub__reduceNeighboredLessPiS_j .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%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 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 $_Z20reduceNeighboredLessPiS_j, %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 _Z35__device_stub__reduceNeighboredLessPiS_j, .Lfunc_end0-_Z35__device_stub__reduceNeighboredLessPiS_j .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 $_Z20reduceNeighboredLessPiS_j, %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 _Z20reduceNeighboredLessPiS_j,@object # @_Z20reduceNeighboredLessPiS_j .section .rodata,"a",@progbits .globl _Z20reduceNeighboredLessPiS_j .p2align 3, 0x0 _Z20reduceNeighboredLessPiS_j: .quad _Z35__device_stub__reduceNeighboredLessPiS_j .size _Z20reduceNeighboredLessPiS_j, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z20reduceNeighboredLessPiS_j" .size .L__unnamed_1, 30 .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 _Z35__device_stub__reduceNeighboredLessPiS_j .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z20reduceNeighboredLessPiS_j .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 <cuda_runtime.h> #include <iostream> __global__ void add(int *d_a,int *d_b,int *d_c,int n) { int idx = threadIdx.x; d_c[idx] = d_a[idx] + d_b[idx]; } int main() { int n = 0; int blag = 1;//标志位 do{ std::cout << "请输入数组的长度:" << std::endl; std::cin >> n; if(n <= 0) { std::cout << "你输入的数组长度为为正数,请重新输入:" << std::endl; }else { blag = 0; } }while(blag); /******申请主机内存******/ int * h_a = (int*)malloc(sizeof(int) * n); int * h_b = (int*)malloc(sizeof(int) * n); int * h_c = (int*)malloc(sizeof(int) * n); /******主机内存赋值********/ for(int i = 0; i < n; ++i) { h_a[i] = i + 1; h_b[i] = i + 3; } /******申请设备内存**********/ int *d_a,*d_b,*d_c; cudaMalloc((void**)&d_a,sizeof(int) * n); cudaMalloc((void**)&d_b,sizeof(int) * n); cudaMalloc((void**)&d_c,sizeof(int) * n); /******主机内存数据复制到设备内存********/ cudaMemcpy(d_a,h_a,sizeof(int) * n,cudaMemcpyHostToDevice); cudaMemcpy(d_b,h_b,sizeof(int) * n,cudaMemcpyHostToDevice); /*****启动核函数********/ add<<<1,n>>>(d_a,d_b,d_c,n); /*****设备内存数据复制到主机内存*********/ cudaMemcpy(h_c,d_c,sizeof(int) * n,cudaMemcpyDeviceToHost); for(int i = 0; i < n; ++i) { std::cout << "h_c[" << i << "] = " << h_c[i] << " "; } std::cout << std::endl; /*******释放设备内存*****/ cudaFree(d_a); cudaFree(d_b); cudaFree(d_c); /*****释放主机内存*****/ free(h_a); free(h_b); free(h_c); std::cout << "运行结束!" << std::endl; return 0; }
code for sm_80 Function : _Z3addPiS_S_i .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_TID.X ; /* 0x0000000000067919 */ /* 0x000e220000002100 */ /*0020*/ HFMA2.MMA R7, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff077435 */ /* 0x000fe200000001ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd20000000a00 */ /*0040*/ IMAD.WIDE R2, R6, R7, c[0x0][0x160] ; /* 0x0000580006027625 */ /* 0x001fc800078e0207 */ /*0050*/ IMAD.WIDE R4, R6.reuse, R7.reuse, c[0x0][0x168] ; /* 0x00005a0006047625 */ /* 0x0c0fe400078e0207 */ /*0060*/ LDG.E R2, [R2.64] ; /* 0x0000000402027981 */ /* 0x000ea8000c1e1900 */ /*0070*/ LDG.E R5, [R4.64] ; /* 0x0000000404057981 */ /* 0x000ea2000c1e1900 */ /*0080*/ IMAD.WIDE R6, R6, R7, c[0x0][0x170] ; /* 0x00005c0006067625 */ /* 0x000fe200078e0207 */ /*0090*/ IADD3 R9, R2, R5, RZ ; /* 0x0000000502097210 */ /* 0x004fca0007ffe0ff */ /*00a0*/ STG.E [R6.64], R9 ; /* 0x0000000906007986 */ /* 0x000fe2000c101904 */ /*00b0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00c0*/ BRA 0xc0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*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 <cuda_runtime.h> #include <iostream> __global__ void add(int *d_a,int *d_b,int *d_c,int n) { int idx = threadIdx.x; d_c[idx] = d_a[idx] + d_b[idx]; } int main() { int n = 0; int blag = 1;//标志位 do{ std::cout << "请输入数组的长度:" << std::endl; std::cin >> n; if(n <= 0) { std::cout << "你输入的数组长度为为正数,请重新输入:" << std::endl; }else { blag = 0; } }while(blag); /******申请主机内存******/ int * h_a = (int*)malloc(sizeof(int) * n); int * h_b = (int*)malloc(sizeof(int) * n); int * h_c = (int*)malloc(sizeof(int) * n); /******主机内存赋值********/ for(int i = 0; i < n; ++i) { h_a[i] = i + 1; h_b[i] = i + 3; } /******申请设备内存**********/ int *d_a,*d_b,*d_c; cudaMalloc((void**)&d_a,sizeof(int) * n); cudaMalloc((void**)&d_b,sizeof(int) * n); cudaMalloc((void**)&d_c,sizeof(int) * n); /******主机内存数据复制到设备内存********/ cudaMemcpy(d_a,h_a,sizeof(int) * n,cudaMemcpyHostToDevice); cudaMemcpy(d_b,h_b,sizeof(int) * n,cudaMemcpyHostToDevice); /*****启动核函数********/ add<<<1,n>>>(d_a,d_b,d_c,n); /*****设备内存数据复制到主机内存*********/ cudaMemcpy(h_c,d_c,sizeof(int) * n,cudaMemcpyDeviceToHost); for(int i = 0; i < n; ++i) { std::cout << "h_c[" << i << "] = " << h_c[i] << " "; } std::cout << std::endl; /*******释放设备内存*****/ cudaFree(d_a); cudaFree(d_b); cudaFree(d_c); /*****释放主机内存*****/ free(h_a); free(h_b); free(h_c); std::cout << "运行结束!" << std::endl; return 0; }
.file "tmpxft_00019623_00000000-6_add.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__Z3addPiS_S_iPiS_S_i .type _Z27__device_stub__Z3addPiS_S_iPiS_S_i, @function _Z27__device_stub__Z3addPiS_S_iPiS_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 _Z3addPiS_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__Z3addPiS_S_iPiS_S_i, .-_Z27__device_stub__Z3addPiS_S_iPiS_S_i .globl _Z3addPiS_S_i .type _Z3addPiS_S_i, @function _Z3addPiS_S_i: .LFB3695: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z27__device_stub__Z3addPiS_S_iPiS_S_i addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3695: .size _Z3addPiS_S_i, .-_Z3addPiS_S_i .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "\350\257\267\350\276\223\345\205\245\346\225\260\347\273\204\347\232\204\351\225\277\345\272\246:" .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC1: .string "\344\275\240\350\276\223\345\205\245\347\232\204\346\225\260\347\273\204\351\225\277\345\272\246\344\270\272\344\270\272\346\255\243\346\225\260,\350\257\267\351\207\215\346\226\260\350\276\223\345\205\245:" .section .rodata.str1.1 .LC2: .string "h_c[" .LC3: .string "] = " .LC4: .string " " .LC5: .string "\350\277\220\350\241\214\347\273\223\346\235\237!" .text .globl main .type main, @function main: .LFB3669: .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 $72, %rsp .cfi_def_cfa_offset 128 movq %fs:40, %rax movq %rax, 56(%rsp) xorl %eax, %eax movl $0, 4(%rsp) leaq .LC0(%rip), %r12 leaq _ZSt4cout(%rip), %rbx leaq _ZSt3cin(%rip), %r13 leaq .LC1(%rip), %r14 jmp .L21 .L33: movq 56(%rsp), %rax subq %fs:40, %rax jne .L31 call _ZSt16__throw_bad_castv@PLT .L31: call __stack_chk_fail@PLT .L14: movq %rbp, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq 0(%rbp), %rax movl $10, %esi movq %rbp, %rdi call *48(%rax) movl %eax, %esi jmp .L15 .L34: movq 56(%rsp), %rax subq %fs:40, %rax jne .L32 call _ZSt16__throw_bad_castv@PLT .L32: call __stack_chk_fail@PLT .L19: movq %rbp, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq 0(%rbp), %rax movl $10, %esi movq %rbp, %rdi call *48(%rax) movl %eax, %esi .L20: movsbl %sil, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT .L21: movl $25, %edx movq %r12, %rsi movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movq (%rbx), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %rbp testq %rbp, %rbp je .L33 cmpb $0, 56(%rbp) je .L14 movzbl 67(%rbp), %esi .L15: movsbl %sil, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT leaq 4(%rsp), %rsi movq %r13, %rdi call _ZNSirsERi@PLT movl 4(%rsp), %eax testl %eax, %eax jg .L16 movl $53, %edx movq %r14, %rsi movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movq (%rbx), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %rbp testq %rbp, %rbp je .L34 cmpb $0, 56(%rbp) je .L19 movzbl 67(%rbp), %esi jmp .L20 .L35: movl 4(%rsp), %ecx movq 24(%rsp), %rdx movq 16(%rsp), %rsi movq 8(%rsp), %rdi call _Z27__device_stub__Z3addPiS_S_iPiS_S_i jmp .L23 .L16: movslq %eax, %rbx leaq 0(,%rbx,4), %rbp movq %rbp, %rdi call malloc@PLT movq %rax, %r15 movq %rbp, %rdi call malloc@PLT movq %rax, %r14 movq %rbp, %rdi call malloc@PLT movq %rax, %r13 movl $1, %eax .L22: movl %eax, -4(%r15,%rax,4) leal 2(%rax), %edx movl %edx, -4(%r14,%rax,4) movq %rax, %rdx addq $1, %rax cmpq %rdx, %rbx jne .L22 leaq 8(%rsp), %rdi movq %rbp, %rsi call cudaMalloc@PLT movslq 4(%rsp), %rsi salq $2, %rsi leaq 16(%rsp), %rdi call cudaMalloc@PLT movslq 4(%rsp), %rsi salq $2, %rsi leaq 24(%rsp), %rdi call cudaMalloc@PLT movslq 4(%rsp), %rdx salq $2, %rdx movl $1, %ecx movq %r15, %rsi movq 8(%rsp), %rdi call cudaMemcpy@PLT movslq 4(%rsp), %rdx salq $2, %rdx movl $1, %ecx movq %r14, %rsi movq 16(%rsp), %rdi call cudaMemcpy@PLT movl 4(%rsp), %eax movl %eax, 44(%rsp) movl $1, 48(%rsp) movl $1, 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 .L35 .L23: movslq 4(%rsp), %rdx salq $2, %rdx movl $2, %ecx movq 24(%rsp), %rsi movq %r13, %rdi call cudaMemcpy@PLT cmpl $0, 4(%rsp) jle .L24 movl $0, %ebx leaq _ZSt4cout(%rip), %r12 .L25: movl $4, %edx leaq .LC2(%rip), %rsi movq %r12, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movl %ebx, %esi movq %r12, %rdi call _ZNSolsEi@PLT movq %rax, %rbp movl $4, %edx leaq .LC3(%rip), %rsi movq %rax, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movl 0(%r13,%rbx,4), %esi movq %rbp, %rdi call _ZNSolsEi@PLT movq %rax, %rdi movl $2, %edx leaq .LC4(%rip), %rsi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT addq $1, %rbx cmpl %ebx, 4(%rsp) jg .L25 .L24: leaq _ZSt4cout(%rip), %rbx movq %rbx, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT movq 8(%rsp), %rdi call cudaFree@PLT movq 16(%rsp), %rdi call cudaFree@PLT movq 24(%rsp), %rdi call cudaFree@PLT movq %r15, %rdi call free@PLT movq %r14, %rdi call free@PLT movq %r13, %rdi call free@PLT leaq .LC5(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT movq 56(%rsp), %rax subq %fs:40, %rax jne .L36 movl $0, %eax addq $72, %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 .L36: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE3669: .size main, .-main .section .rodata.str1.1 .LC6: .string "_Z3addPiS_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 .LC6(%rip), %rdx movq %rdx, %rcx leaq _Z3addPiS_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 .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 <cuda_runtime.h> #include <iostream> __global__ void add(int *d_a,int *d_b,int *d_c,int n) { int idx = threadIdx.x; d_c[idx] = d_a[idx] + d_b[idx]; } int main() { int n = 0; int blag = 1;//标志位 do{ std::cout << "请输入数组的长度:" << std::endl; std::cin >> n; if(n <= 0) { std::cout << "你输入的数组长度为为正数,请重新输入:" << std::endl; }else { blag = 0; } }while(blag); /******申请主机内存******/ int * h_a = (int*)malloc(sizeof(int) * n); int * h_b = (int*)malloc(sizeof(int) * n); int * h_c = (int*)malloc(sizeof(int) * n); /******主机内存赋值********/ for(int i = 0; i < n; ++i) { h_a[i] = i + 1; h_b[i] = i + 3; } /******申请设备内存**********/ int *d_a,*d_b,*d_c; cudaMalloc((void**)&d_a,sizeof(int) * n); cudaMalloc((void**)&d_b,sizeof(int) * n); cudaMalloc((void**)&d_c,sizeof(int) * n); /******主机内存数据复制到设备内存********/ cudaMemcpy(d_a,h_a,sizeof(int) * n,cudaMemcpyHostToDevice); cudaMemcpy(d_b,h_b,sizeof(int) * n,cudaMemcpyHostToDevice); /*****启动核函数********/ add<<<1,n>>>(d_a,d_b,d_c,n); /*****设备内存数据复制到主机内存*********/ cudaMemcpy(h_c,d_c,sizeof(int) * n,cudaMemcpyDeviceToHost); for(int i = 0; i < n; ++i) { std::cout << "h_c[" << i << "] = " << h_c[i] << " "; } std::cout << std::endl; /*******释放设备内存*****/ cudaFree(d_a); cudaFree(d_b); cudaFree(d_c); /*****释放主机内存*****/ free(h_a); free(h_b); free(h_c); std::cout << "运行结束!" << std::endl; return 0; }
#include <hip/hip_runtime.h> #include <iostream> __global__ void add(int *d_a,int *d_b,int *d_c,int n) { int idx = threadIdx.x; d_c[idx] = d_a[idx] + d_b[idx]; } int main() { int n = 0; int blag = 1;//标志位 do{ std::cout << "请输入数组的长度:" << std::endl; std::cin >> n; if(n <= 0) { std::cout << "你输入的数组长度为为正数,请重新输入:" << std::endl; }else { blag = 0; } }while(blag); /******申请主机内存******/ int * h_a = (int*)malloc(sizeof(int) * n); int * h_b = (int*)malloc(sizeof(int) * n); int * h_c = (int*)malloc(sizeof(int) * n); /******主机内存赋值********/ for(int i = 0; i < n; ++i) { h_a[i] = i + 1; h_b[i] = i + 3; } /******申请设备内存**********/ int *d_a,*d_b,*d_c; hipMalloc((void**)&d_a,sizeof(int) * n); hipMalloc((void**)&d_b,sizeof(int) * n); hipMalloc((void**)&d_c,sizeof(int) * n); /******主机内存数据复制到设备内存********/ hipMemcpy(d_a,h_a,sizeof(int) * n,hipMemcpyHostToDevice); hipMemcpy(d_b,h_b,sizeof(int) * n,hipMemcpyHostToDevice); /*****启动核函数********/ add<<<1,n>>>(d_a,d_b,d_c,n); /*****设备内存数据复制到主机内存*********/ hipMemcpy(h_c,d_c,sizeof(int) * n,hipMemcpyDeviceToHost); for(int i = 0; i < n; ++i) { std::cout << "h_c[" << i << "] = " << h_c[i] << " "; } std::cout << std::endl; /*******释放设备内存*****/ hipFree(d_a); hipFree(d_b); hipFree(d_c); /*****释放主机内存*****/ free(h_a); free(h_b); free(h_c); std::cout << "运行结束!" << std::endl; 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> __global__ void add(int *d_a,int *d_b,int *d_c,int n) { int idx = threadIdx.x; d_c[idx] = d_a[idx] + d_b[idx]; } int main() { int n = 0; int blag = 1;//标志位 do{ std::cout << "请输入数组的长度:" << std::endl; std::cin >> n; if(n <= 0) { std::cout << "你输入的数组长度为为正数,请重新输入:" << std::endl; }else { blag = 0; } }while(blag); /******申请主机内存******/ int * h_a = (int*)malloc(sizeof(int) * n); int * h_b = (int*)malloc(sizeof(int) * n); int * h_c = (int*)malloc(sizeof(int) * n); /******主机内存赋值********/ for(int i = 0; i < n; ++i) { h_a[i] = i + 1; h_b[i] = i + 3; } /******申请设备内存**********/ int *d_a,*d_b,*d_c; hipMalloc((void**)&d_a,sizeof(int) * n); hipMalloc((void**)&d_b,sizeof(int) * n); hipMalloc((void**)&d_c,sizeof(int) * n); /******主机内存数据复制到设备内存********/ hipMemcpy(d_a,h_a,sizeof(int) * n,hipMemcpyHostToDevice); hipMemcpy(d_b,h_b,sizeof(int) * n,hipMemcpyHostToDevice); /*****启动核函数********/ add<<<1,n>>>(d_a,d_b,d_c,n); /*****设备内存数据复制到主机内存*********/ hipMemcpy(h_c,d_c,sizeof(int) * n,hipMemcpyDeviceToHost); for(int i = 0; i < n; ++i) { std::cout << "h_c[" << i << "] = " << h_c[i] << " "; } std::cout << std::endl; /*******释放设备内存*****/ hipFree(d_a); hipFree(d_b); hipFree(d_c); /*****释放主机内存*****/ free(h_a); free(h_b); free(h_c); std::cout << "运行结束!" << std::endl; return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z3addPiS_S_i .globl _Z3addPiS_S_i .p2align 8 .type _Z3addPiS_S_i,@function _Z3addPiS_S_i: s_load_b128 s[4:7], s[0:1], 0x0 v_lshlrev_b32_e32 v0, 2, v0 s_load_b64 s[0:1], s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_clause 0x1 global_load_b32 v1, v0, s[4:5] global_load_b32 v2, v0, s[6:7] s_waitcnt vmcnt(0) v_add_nc_u32_e32 v1, v2, 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 _Z3addPiS_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 _Z3addPiS_S_i, .Lfunc_end0-_Z3addPiS_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: _Z3addPiS_S_i .private_segment_fixed_size: 0 .sgpr_count: 8 .sgpr_spill_count: 0 .symbol: _Z3addPiS_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> __global__ void add(int *d_a,int *d_b,int *d_c,int n) { int idx = threadIdx.x; d_c[idx] = d_a[idx] + d_b[idx]; } int main() { int n = 0; int blag = 1;//标志位 do{ std::cout << "请输入数组的长度:" << std::endl; std::cin >> n; if(n <= 0) { std::cout << "你输入的数组长度为为正数,请重新输入:" << std::endl; }else { blag = 0; } }while(blag); /******申请主机内存******/ int * h_a = (int*)malloc(sizeof(int) * n); int * h_b = (int*)malloc(sizeof(int) * n); int * h_c = (int*)malloc(sizeof(int) * n); /******主机内存赋值********/ for(int i = 0; i < n; ++i) { h_a[i] = i + 1; h_b[i] = i + 3; } /******申请设备内存**********/ int *d_a,*d_b,*d_c; hipMalloc((void**)&d_a,sizeof(int) * n); hipMalloc((void**)&d_b,sizeof(int) * n); hipMalloc((void**)&d_c,sizeof(int) * n); /******主机内存数据复制到设备内存********/ hipMemcpy(d_a,h_a,sizeof(int) * n,hipMemcpyHostToDevice); hipMemcpy(d_b,h_b,sizeof(int) * n,hipMemcpyHostToDevice); /*****启动核函数********/ add<<<1,n>>>(d_a,d_b,d_c,n); /*****设备内存数据复制到主机内存*********/ hipMemcpy(h_c,d_c,sizeof(int) * n,hipMemcpyDeviceToHost); for(int i = 0; i < n; ++i) { std::cout << "h_c[" << i << "] = " << h_c[i] << " "; } std::cout << std::endl; /*******释放设备内存*****/ hipFree(d_a); hipFree(d_b); hipFree(d_c); /*****释放主机内存*****/ free(h_a); free(h_b); free(h_c); std::cout << "运行结束!" << std::endl; return 0; }
.text .file "add.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl _Z18__device_stub__addPiS_S_i # -- Begin function _Z18__device_stub__addPiS_S_i .p2align 4, 0x90 .type _Z18__device_stub__addPiS_S_i,@function _Z18__device_stub__addPiS_S_i: # @_Z18__device_stub__addPiS_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 $_Z3addPiS_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__addPiS_S_i, .Lfunc_end0-_Z18__device_stub__addPiS_S_i .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 %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 $0, 4(%rsp) movl $_ZSt4cout, %edi movl $.L.str, %esi movl $25, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB1_11 # %bb.1: leaq 4(%rsp), %rbx jmp .LBB1_2 .p2align 4, 0x90 .LBB1_9: # in Loop: Header=BB1_2 Depth=1 movq %r14, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) .LBB1_10: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit27 # in Loop: Header=BB1_2 Depth=1 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movl $_ZSt4cout, %edi movl $.L.str, %esi movl $25, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB1_11 .LBB1_2: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i # =>This Inner Loop Header: Depth=1 cmpb $0, 56(%r14) je .LBB1_4 # %bb.3: # in Loop: Header=BB1_2 Depth=1 movzbl 67(%r14), %eax jmp .LBB1_5 .p2align 4, 0x90 .LBB1_4: # in Loop: Header=BB1_2 Depth=1 movq %r14, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) .LBB1_5: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit # in Loop: Header=BB1_2 Depth=1 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movl $_ZSt3cin, %edi movq %rbx, %rsi callq _ZNSirsERi movl 4(%rsp), %r13d testl %r13d, %r13d jg .LBB1_12 # %bb.6: # in Loop: Header=BB1_2 Depth=1 movl $_ZSt4cout, %edi movl $.L.str.1, %esi movl $53, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB1_11 # %bb.7: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i24 # in Loop: Header=BB1_2 Depth=1 cmpb $0, 56(%r14) je .LBB1_9 # %bb.8: # in Loop: Header=BB1_2 Depth=1 movzbl 67(%r14), %eax jmp .LBB1_10 .LBB1_12: # %.critedge leaq (,%r13,4), %r12 movq %r12, %rdi callq malloc movq %rax, %rbx movq %r12, %rdi callq malloc movq %rax, %r14 movq %r12, %rdi callq malloc movq %rax, %r15 xorl %eax, %eax .p2align 4, 0x90 .LBB1_13: # =>This Inner Loop Header: Depth=1 leaq 1(%rax), %rcx movl %ecx, (%rbx,%rax,4) leal 3(%rax), %edx movl %edx, (%r14,%rax,4) movq %rcx, %rax cmpq %rcx, %r13 jne .LBB1_13 # %bb.14: leaq 24(%rsp), %rdi movq %r12, %rsi callq hipMalloc movslq 4(%rsp), %rsi shlq $2, %rsi leaq 16(%rsp), %rdi callq hipMalloc movslq 4(%rsp), %rsi shlq $2, %rsi leaq 8(%rsp), %rdi callq hipMalloc movq 24(%rsp), %rdi movslq 4(%rsp), %rdx shlq $2, %rdx movq %rbx, %rsi movl $1, %ecx callq hipMemcpy movq 16(%rsp), %rdi movslq 4(%rsp), %rdx shlq $2, %rdx movq %r14, %rsi movl $1, %ecx callq hipMemcpy movl 4(%rsp), %edx movabsq $4294967296, %rdi # imm = 0x100000000 orq %rdi, %rdx orq $1, %rdi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_16 # %bb.15: movq 24(%rsp), %rax movq 16(%rsp), %rcx movq 8(%rsp), %rdx movl 4(%rsp), %esi movq %rax, 104(%rsp) movq %rcx, 96(%rsp) movq %rdx, 88(%rsp) movl %esi, 36(%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 36(%rsp), %rax movq %rax, 136(%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 $_Z3addPiS_S_i, %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_16: movq 8(%rsp), %rsi movslq 4(%rsp), %rdx shlq $2, %rdx movq %r15, %rdi movl $2, %ecx callq hipMemcpy cmpl $0, 4(%rsp) jle .LBB1_19 # %bb.17: # %.lr.ph.preheader xorl %r12d, %r12d .p2align 4, 0x90 .LBB1_18: # %.lr.ph # =>This Inner Loop Header: Depth=1 movl $_ZSt4cout, %edi movl $.L.str.2, %esi movl $4, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl $_ZSt4cout, %edi movl %r12d, %esi callq _ZNSolsEi movq %rax, %r13 movl $.L.str.3, %esi movl $4, %edx movq %rax, %rdi callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl (%r15,%r12,4), %esi movq %r13, %rdi callq _ZNSolsEi movl $.L.str.4, %esi movl $2, %edx movq %rax, %rdi callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l incq %r12 movslq 4(%rsp), %rax cmpq %rax, %r12 jl .LBB1_18 .LBB1_19: # %._crit_edge48 movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r12 testq %r12, %r12 je .LBB1_11 # %bb.20: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i29 cmpb $0, 56(%r12) je .LBB1_22 # %bb.21: movzbl 67(%r12), %eax jmp .LBB1_23 .LBB1_22: movq %r12, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r12), %rax movq %r12, %rdi movl $10, %esi callq *48(%rax) .LBB1_23: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit32 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movq 24(%rsp), %rdi callq hipFree movq 16(%rsp), %rdi callq hipFree movq 8(%rsp), %rdi callq hipFree movq %rbx, %rdi callq free movq %r14, %rdi callq free movq %r15, %rdi callq free movl $_ZSt4cout, %edi movl $.L.str.5, %esi movl $13, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %rbx testq %rbx, %rbx je .LBB1_11 # %bb.24: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i34 cmpb $0, 56(%rbx) je .LBB1_26 # %bb.25: movzbl 67(%rbx), %eax jmp .LBB1_27 .LBB1_26: movq %rbx, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%rbx), %rax movq %rbx, %rdi movl $10, %esi callq *48(%rax) .LBB1_27: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit37 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv 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 .LBB1_11: # %._crit_edge .cfi_def_cfa_offset 192 callq _ZSt16__throw_bad_castv .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 $_Z3addPiS_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 _Z3addPiS_S_i,@object # @_Z3addPiS_S_i .section .rodata,"a",@progbits .globl _Z3addPiS_S_i .p2align 3, 0x0 _Z3addPiS_S_i: .quad _Z18__device_stub__addPiS_S_i .size _Z3addPiS_S_i, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "\350\257\267\350\276\223\345\205\245\346\225\260\347\273\204\347\232\204\351\225\277\345\272\246:" .size .L.str, 26 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz "\344\275\240\350\276\223\345\205\245\347\232\204\346\225\260\347\273\204\351\225\277\345\272\246\344\270\272\344\270\272\346\255\243\346\225\260,\350\257\267\351\207\215\346\226\260\350\276\223\345\205\245:" .size .L.str.1, 54 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "h_c[" .size .L.str.2, 5 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "] = " .size .L.str.3, 5 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz " " .size .L.str.4, 3 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "\350\277\220\350\241\214\347\273\223\346\235\237!" .size .L.str.5, 14 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z3addPiS_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__addPiS_S_i .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z3addPiS_S_i .addrsig_sym _ZSt4cout .addrsig_sym _ZSt3cin .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 : _Z3addPiS_S_i .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_TID.X ; /* 0x0000000000067919 */ /* 0x000e220000002100 */ /*0020*/ HFMA2.MMA R7, -RZ, RZ, 0, 2.384185791015625e-07 ; /* 0x00000004ff077435 */ /* 0x000fe200000001ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd20000000a00 */ /*0040*/ IMAD.WIDE R2, R6, R7, c[0x0][0x160] ; /* 0x0000580006027625 */ /* 0x001fc800078e0207 */ /*0050*/ IMAD.WIDE R4, R6.reuse, R7.reuse, c[0x0][0x168] ; /* 0x00005a0006047625 */ /* 0x0c0fe400078e0207 */ /*0060*/ LDG.E R2, [R2.64] ; /* 0x0000000402027981 */ /* 0x000ea8000c1e1900 */ /*0070*/ LDG.E R5, [R4.64] ; /* 0x0000000404057981 */ /* 0x000ea2000c1e1900 */ /*0080*/ IMAD.WIDE R6, R6, R7, c[0x0][0x170] ; /* 0x00005c0006067625 */ /* 0x000fe200078e0207 */ /*0090*/ IADD3 R9, R2, R5, RZ ; /* 0x0000000502097210 */ /* 0x004fca0007ffe0ff */ /*00a0*/ STG.E [R6.64], R9 ; /* 0x0000000906007986 */ /* 0x000fe2000c101904 */ /*00b0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00c0*/ BRA 0xc0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*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 _Z3addPiS_S_i .globl _Z3addPiS_S_i .p2align 8 .type _Z3addPiS_S_i,@function _Z3addPiS_S_i: s_load_b128 s[4:7], s[0:1], 0x0 v_lshlrev_b32_e32 v0, 2, v0 s_load_b64 s[0:1], s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_clause 0x1 global_load_b32 v1, v0, s[4:5] global_load_b32 v2, v0, s[6:7] s_waitcnt vmcnt(0) v_add_nc_u32_e32 v1, v2, 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 _Z3addPiS_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 _Z3addPiS_S_i, .Lfunc_end0-_Z3addPiS_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: _Z3addPiS_S_i .private_segment_fixed_size: 0 .sgpr_count: 8 .sgpr_spill_count: 0 .symbol: _Z3addPiS_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_00019623_00000000-6_add.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__Z3addPiS_S_iPiS_S_i .type _Z27__device_stub__Z3addPiS_S_iPiS_S_i, @function _Z27__device_stub__Z3addPiS_S_iPiS_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 _Z3addPiS_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__Z3addPiS_S_iPiS_S_i, .-_Z27__device_stub__Z3addPiS_S_iPiS_S_i .globl _Z3addPiS_S_i .type _Z3addPiS_S_i, @function _Z3addPiS_S_i: .LFB3695: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z27__device_stub__Z3addPiS_S_iPiS_S_i addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE3695: .size _Z3addPiS_S_i, .-_Z3addPiS_S_i .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "\350\257\267\350\276\223\345\205\245\346\225\260\347\273\204\347\232\204\351\225\277\345\272\246:" .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC1: .string "\344\275\240\350\276\223\345\205\245\347\232\204\346\225\260\347\273\204\351\225\277\345\272\246\344\270\272\344\270\272\346\255\243\346\225\260,\350\257\267\351\207\215\346\226\260\350\276\223\345\205\245:" .section .rodata.str1.1 .LC2: .string "h_c[" .LC3: .string "] = " .LC4: .string " " .LC5: .string "\350\277\220\350\241\214\347\273\223\346\235\237!" .text .globl main .type main, @function main: .LFB3669: .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 $72, %rsp .cfi_def_cfa_offset 128 movq %fs:40, %rax movq %rax, 56(%rsp) xorl %eax, %eax movl $0, 4(%rsp) leaq .LC0(%rip), %r12 leaq _ZSt4cout(%rip), %rbx leaq _ZSt3cin(%rip), %r13 leaq .LC1(%rip), %r14 jmp .L21 .L33: movq 56(%rsp), %rax subq %fs:40, %rax jne .L31 call _ZSt16__throw_bad_castv@PLT .L31: call __stack_chk_fail@PLT .L14: movq %rbp, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq 0(%rbp), %rax movl $10, %esi movq %rbp, %rdi call *48(%rax) movl %eax, %esi jmp .L15 .L34: movq 56(%rsp), %rax subq %fs:40, %rax jne .L32 call _ZSt16__throw_bad_castv@PLT .L32: call __stack_chk_fail@PLT .L19: movq %rbp, %rdi call _ZNKSt5ctypeIcE13_M_widen_initEv@PLT movq 0(%rbp), %rax movl $10, %esi movq %rbp, %rdi call *48(%rax) movl %eax, %esi .L20: movsbl %sil, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT .L21: movl $25, %edx movq %r12, %rsi movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movq (%rbx), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %rbp testq %rbp, %rbp je .L33 cmpb $0, 56(%rbp) je .L14 movzbl 67(%rbp), %esi .L15: movsbl %sil, %esi movq %rbx, %rdi call _ZNSo3putEc@PLT movq %rax, %rdi call _ZNSo5flushEv@PLT leaq 4(%rsp), %rsi movq %r13, %rdi call _ZNSirsERi@PLT movl 4(%rsp), %eax testl %eax, %eax jg .L16 movl $53, %edx movq %r14, %rsi movq %rbx, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movq (%rbx), %rax movq -24(%rax), %rax movq 240(%rbx,%rax), %rbp testq %rbp, %rbp je .L34 cmpb $0, 56(%rbp) je .L19 movzbl 67(%rbp), %esi jmp .L20 .L35: movl 4(%rsp), %ecx movq 24(%rsp), %rdx movq 16(%rsp), %rsi movq 8(%rsp), %rdi call _Z27__device_stub__Z3addPiS_S_iPiS_S_i jmp .L23 .L16: movslq %eax, %rbx leaq 0(,%rbx,4), %rbp movq %rbp, %rdi call malloc@PLT movq %rax, %r15 movq %rbp, %rdi call malloc@PLT movq %rax, %r14 movq %rbp, %rdi call malloc@PLT movq %rax, %r13 movl $1, %eax .L22: movl %eax, -4(%r15,%rax,4) leal 2(%rax), %edx movl %edx, -4(%r14,%rax,4) movq %rax, %rdx addq $1, %rax cmpq %rdx, %rbx jne .L22 leaq 8(%rsp), %rdi movq %rbp, %rsi call cudaMalloc@PLT movslq 4(%rsp), %rsi salq $2, %rsi leaq 16(%rsp), %rdi call cudaMalloc@PLT movslq 4(%rsp), %rsi salq $2, %rsi leaq 24(%rsp), %rdi call cudaMalloc@PLT movslq 4(%rsp), %rdx salq $2, %rdx movl $1, %ecx movq %r15, %rsi movq 8(%rsp), %rdi call cudaMemcpy@PLT movslq 4(%rsp), %rdx salq $2, %rdx movl $1, %ecx movq %r14, %rsi movq 16(%rsp), %rdi call cudaMemcpy@PLT movl 4(%rsp), %eax movl %eax, 44(%rsp) movl $1, 48(%rsp) movl $1, 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 .L35 .L23: movslq 4(%rsp), %rdx salq $2, %rdx movl $2, %ecx movq 24(%rsp), %rsi movq %r13, %rdi call cudaMemcpy@PLT cmpl $0, 4(%rsp) jle .L24 movl $0, %ebx leaq _ZSt4cout(%rip), %r12 .L25: movl $4, %edx leaq .LC2(%rip), %rsi movq %r12, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movl %ebx, %esi movq %r12, %rdi call _ZNSolsEi@PLT movq %rax, %rbp movl $4, %edx leaq .LC3(%rip), %rsi movq %rax, %rdi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT movl 0(%r13,%rbx,4), %esi movq %rbp, %rdi call _ZNSolsEi@PLT movq %rax, %rdi movl $2, %edx leaq .LC4(%rip), %rsi call _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l@PLT addq $1, %rbx cmpl %ebx, 4(%rsp) jg .L25 .L24: leaq _ZSt4cout(%rip), %rbx movq %rbx, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT movq 8(%rsp), %rdi call cudaFree@PLT movq 16(%rsp), %rdi call cudaFree@PLT movq 24(%rsp), %rdi call cudaFree@PLT movq %r15, %rdi call free@PLT movq %r14, %rdi call free@PLT movq %r13, %rdi call free@PLT leaq .LC5(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT movq 56(%rsp), %rax subq %fs:40, %rax jne .L36 movl $0, %eax addq $72, %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 .L36: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE3669: .size main, .-main .section .rodata.str1.1 .LC6: .string "_Z3addPiS_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 .LC6(%rip), %rdx movq %rdx, %rcx leaq _Z3addPiS_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 .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.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl _Z18__device_stub__addPiS_S_i # -- Begin function _Z18__device_stub__addPiS_S_i .p2align 4, 0x90 .type _Z18__device_stub__addPiS_S_i,@function _Z18__device_stub__addPiS_S_i: # @_Z18__device_stub__addPiS_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 $_Z3addPiS_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__addPiS_S_i, .Lfunc_end0-_Z18__device_stub__addPiS_S_i .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 %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 $0, 4(%rsp) movl $_ZSt4cout, %edi movl $.L.str, %esi movl $25, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB1_11 # %bb.1: leaq 4(%rsp), %rbx jmp .LBB1_2 .p2align 4, 0x90 .LBB1_9: # in Loop: Header=BB1_2 Depth=1 movq %r14, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) .LBB1_10: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit27 # in Loop: Header=BB1_2 Depth=1 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movl $_ZSt4cout, %edi movl $.L.str, %esi movl $25, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB1_11 .LBB1_2: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i # =>This Inner Loop Header: Depth=1 cmpb $0, 56(%r14) je .LBB1_4 # %bb.3: # in Loop: Header=BB1_2 Depth=1 movzbl 67(%r14), %eax jmp .LBB1_5 .p2align 4, 0x90 .LBB1_4: # in Loop: Header=BB1_2 Depth=1 movq %r14, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax movq %r14, %rdi movl $10, %esi callq *48(%rax) .LBB1_5: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit # in Loop: Header=BB1_2 Depth=1 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movl $_ZSt3cin, %edi movq %rbx, %rsi callq _ZNSirsERi movl 4(%rsp), %r13d testl %r13d, %r13d jg .LBB1_12 # %bb.6: # in Loop: Header=BB1_2 Depth=1 movl $_ZSt4cout, %edi movl $.L.str.1, %esi movl $53, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB1_11 # %bb.7: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i24 # in Loop: Header=BB1_2 Depth=1 cmpb $0, 56(%r14) je .LBB1_9 # %bb.8: # in Loop: Header=BB1_2 Depth=1 movzbl 67(%r14), %eax jmp .LBB1_10 .LBB1_12: # %.critedge leaq (,%r13,4), %r12 movq %r12, %rdi callq malloc movq %rax, %rbx movq %r12, %rdi callq malloc movq %rax, %r14 movq %r12, %rdi callq malloc movq %rax, %r15 xorl %eax, %eax .p2align 4, 0x90 .LBB1_13: # =>This Inner Loop Header: Depth=1 leaq 1(%rax), %rcx movl %ecx, (%rbx,%rax,4) leal 3(%rax), %edx movl %edx, (%r14,%rax,4) movq %rcx, %rax cmpq %rcx, %r13 jne .LBB1_13 # %bb.14: leaq 24(%rsp), %rdi movq %r12, %rsi callq hipMalloc movslq 4(%rsp), %rsi shlq $2, %rsi leaq 16(%rsp), %rdi callq hipMalloc movslq 4(%rsp), %rsi shlq $2, %rsi leaq 8(%rsp), %rdi callq hipMalloc movq 24(%rsp), %rdi movslq 4(%rsp), %rdx shlq $2, %rdx movq %rbx, %rsi movl $1, %ecx callq hipMemcpy movq 16(%rsp), %rdi movslq 4(%rsp), %rdx shlq $2, %rdx movq %r14, %rsi movl $1, %ecx callq hipMemcpy movl 4(%rsp), %edx movabsq $4294967296, %rdi # imm = 0x100000000 orq %rdi, %rdx orq $1, %rdi movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_16 # %bb.15: movq 24(%rsp), %rax movq 16(%rsp), %rcx movq 8(%rsp), %rdx movl 4(%rsp), %esi movq %rax, 104(%rsp) movq %rcx, 96(%rsp) movq %rdx, 88(%rsp) movl %esi, 36(%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 36(%rsp), %rax movq %rax, 136(%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 $_Z3addPiS_S_i, %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_16: movq 8(%rsp), %rsi movslq 4(%rsp), %rdx shlq $2, %rdx movq %r15, %rdi movl $2, %ecx callq hipMemcpy cmpl $0, 4(%rsp) jle .LBB1_19 # %bb.17: # %.lr.ph.preheader xorl %r12d, %r12d .p2align 4, 0x90 .LBB1_18: # %.lr.ph # =>This Inner Loop Header: Depth=1 movl $_ZSt4cout, %edi movl $.L.str.2, %esi movl $4, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl $_ZSt4cout, %edi movl %r12d, %esi callq _ZNSolsEi movq %rax, %r13 movl $.L.str.3, %esi movl $4, %edx movq %rax, %rdi callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl (%r15,%r12,4), %esi movq %r13, %rdi callq _ZNSolsEi movl $.L.str.4, %esi movl $2, %edx movq %rax, %rdi callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l incq %r12 movslq 4(%rsp), %rax cmpq %rax, %r12 jl .LBB1_18 .LBB1_19: # %._crit_edge48 movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r12 testq %r12, %r12 je .LBB1_11 # %bb.20: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i29 cmpb $0, 56(%r12) je .LBB1_22 # %bb.21: movzbl 67(%r12), %eax jmp .LBB1_23 .LBB1_22: movq %r12, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r12), %rax movq %r12, %rdi movl $10, %esi callq *48(%rax) .LBB1_23: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit32 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv movq 24(%rsp), %rdi callq hipFree movq 16(%rsp), %rdi callq hipFree movq 8(%rsp), %rdi callq hipFree movq %rbx, %rdi callq free movq %r14, %rdi callq free movq %r15, %rdi callq free movl $_ZSt4cout, %edi movl $.L.str.5, %esi movl $13, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %rbx testq %rbx, %rbx je .LBB1_11 # %bb.24: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i34 cmpb $0, 56(%rbx) je .LBB1_26 # %bb.25: movzbl 67(%rbx), %eax jmp .LBB1_27 .LBB1_26: movq %rbx, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%rbx), %rax movq %rbx, %rdi movl $10, %esi callq *48(%rax) .LBB1_27: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit37 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc movq %rax, %rdi callq _ZNSo5flushEv 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 .LBB1_11: # %._crit_edge .cfi_def_cfa_offset 192 callq _ZSt16__throw_bad_castv .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 $_Z3addPiS_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 _Z3addPiS_S_i,@object # @_Z3addPiS_S_i .section .rodata,"a",@progbits .globl _Z3addPiS_S_i .p2align 3, 0x0 _Z3addPiS_S_i: .quad _Z18__device_stub__addPiS_S_i .size _Z3addPiS_S_i, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "\350\257\267\350\276\223\345\205\245\346\225\260\347\273\204\347\232\204\351\225\277\345\272\246:" .size .L.str, 26 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz "\344\275\240\350\276\223\345\205\245\347\232\204\346\225\260\347\273\204\351\225\277\345\272\246\344\270\272\344\270\272\346\255\243\346\225\260,\350\257\267\351\207\215\346\226\260\350\276\223\345\205\245:" .size .L.str.1, 54 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "h_c[" .size .L.str.2, 5 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "] = " .size .L.str.3, 5 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz " " .size .L.str.4, 3 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "\350\277\220\350\241\214\347\273\223\346\235\237!" .size .L.str.5, 14 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z3addPiS_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__addPiS_S_i .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z3addPiS_S_i .addrsig_sym _ZSt4cout .addrsig_sym _ZSt3cin .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 <cassert> #include <cuda.h> #include <cuda_runtime.h> #include <iostream> #include <numeric> #include <vector> using namespace std; #define TILE_SIZE 64 #define MAX_MASK_WIDTH 5 __constant__ float c_M[MAX_MASK_WIDTH]; __global__ void convolution1(const float* N, const float* M, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width/2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j]*M[j]; } } P[i] = Pvalue; } __global__ void convolution2(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width / 2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j] * c_M[j]; } } P[i] = Pvalue; } __global__ void convolution3(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; __shared__ float Nds[TILE_SIZE + MAX_MASK_WIDTH - 1]; // load N from global memory into shared memory int n = mask_width/2; if (threadIdx.x >= blockDim.x - n) { int halo_index_left = (blockIdx.x - 1)*blockDim.x + threadIdx.x; Nds[threadIdx.x - (blockDim.x - n)] = (halo_index_left < 0) ? 0 : N[halo_index_left]; } Nds[n + threadIdx.x] = N[i]; if (threadIdx.x < n) { int halo_index_right = (blockIdx.x + 1)*blockDim.x + threadIdx.x; Nds[n + blockDim.x + threadIdx.x] = (halo_index_right >= width) ? 0 : N[halo_index_right]; } __syncthreads(); float Pvalue = 0.0f; for (int j = 0; j < mask_width; ++j) { Pvalue += Nds[threadIdx.x + j]*c_M[j]; } P[i] = Pvalue; } int main(int argc, char* argv[]) { // Query GPU properties cudaDeviceProp dev_prop; cudaGetDeviceProperties(&dev_prop, 0); cout << "---------------------------------------------" << endl; cout << " GPU PROPERTIES " << endl; cout << "---------------------------------------------" << endl; cout << "Device Name: " << dev_prop.name << endl; cout << "Memory Clock Rate: " << dev_prop.memoryClockRate/1.0e6 << " GHz" << endl; cout << "Memory Bandwidth: " << 2.0*dev_prop.memoryClockRate*(dev_prop.memoryBusWidth/8)/1.0e6 << " GB/s" << endl; cout << "Number of SM: " << dev_prop.multiProcessorCount << endl; cout << "Max Threads per SM: " << dev_prop.maxThreadsPerMultiProcessor << endl; cout << "Registers per Block: " << dev_prop.regsPerBlock << endl; cout << "Shared Memory per Block: " << dev_prop.sharedMemPerBlock << " B" << endl; cout << "Total Global Memory per Block: " << dev_prop.totalGlobalMem/1.0e9 << " GB" << endl; cout << endl; int size = atoi(argv[1]); // creating vector on host side vector<float> h_N(size, 1.0f); std::iota(h_N.begin(), h_N.end(), 0.0f); // Copy vector on device side float* d_N; cudaMalloc((void**)&d_N, size*sizeof(float)); cudaMemcpy((void*)d_N, (void*)h_N.data(), size*sizeof(float), cudaMemcpyHostToDevice); // Create mask and send to devide vector<float> h_M = { 1.0f, 1.0f, 2.0f, 1.0f, 1.0f }; int mask_width = h_M.size(); assert(mask_width < MAX_MASK_WIDTH); cudaMemcpyToSymbol(c_M, (void*)h_M.data(), mask_width*sizeof(float)); // Allocate space for solution on device float* d_P; cudaMalloc((void**)&d_P, size*sizeof(float)); // call Kernel int blockDim = TILE_SIZE; int gridDim = ceil(size/(float)blockDim); int version = atoi(argv[2]); if(version == 1) convolution2<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else if(version == 2) convolution3<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else cout << "Wrong inputs!" << endl; // Recover vector from device to host vector<float> h_P(size); cudaMemcpy((void*)h_P.data(), (void*)d_P, size*sizeof(float), cudaMemcpyDeviceToHost); // Finalize storage cudaFree(d_N); cudaFree(d_P); cout << "Closing..." << endl; return 0; }
.file "tmpxft_0014b803_00000000-6_main.cudafe1.cpp" .text #APP .globl _ZSt21ios_base_library_initv #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB4074: .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 .LFE4074: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z40__device_stub__Z12convolution1PKfS0_PfiiPKfS0_Pfii .type _Z40__device_stub__Z12convolution1PKfS0_PfiiPKfS0_Pfii, @function _Z40__device_stub__Z12convolution1PKfS0_PfiiPKfS0_Pfii: .LFB4096: .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) movl %r8d, (%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) movq %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 _Z12convolution1PKfS0_Pfii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE4096: .size _Z40__device_stub__Z12convolution1PKfS0_PfiiPKfS0_Pfii, .-_Z40__device_stub__Z12convolution1PKfS0_PfiiPKfS0_Pfii .globl _Z12convolution1PKfS0_Pfii .type _Z12convolution1PKfS0_Pfii, @function _Z12convolution1PKfS0_Pfii: .LFB4097: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z40__device_stub__Z12convolution1PKfS0_PfiiPKfS0_Pfii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE4097: .size _Z12convolution1PKfS0_Pfii, .-_Z12convolution1PKfS0_Pfii .globl _Z37__device_stub__Z12convolution2PKfPfiiPKfPfii .type _Z37__device_stub__Z12convolution2PKfPfiiPKfPfii, @function _Z37__device_stub__Z12convolution2PKfPfiiPKfPfii: .LFB4098: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%rsp) movl %ecx, 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 .L15 .L11: movq 136(%rsp), %rax subq %fs:40, %rax jne .L16 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L15: .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 _Z12convolution2PKfPfii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L11 .L16: call __stack_chk_fail@PLT .cfi_endproc .LFE4098: .size _Z37__device_stub__Z12convolution2PKfPfiiPKfPfii, .-_Z37__device_stub__Z12convolution2PKfPfiiPKfPfii .globl _Z12convolution2PKfPfii .type _Z12convolution2PKfPfii, @function _Z12convolution2PKfPfii: .LFB4099: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z37__device_stub__Z12convolution2PKfPfiiPKfPfii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE4099: .size _Z12convolution2PKfPfii, .-_Z12convolution2PKfPfii .globl _Z37__device_stub__Z12convolution3PKfPfiiPKfPfii .type _Z37__device_stub__Z12convolution3PKfPfiiPKfPfii, @function _Z37__device_stub__Z12convolution3PKfPfiiPKfPfii: .LFB4100: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%rsp) movl %ecx, 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 .L23 .L19: movq 136(%rsp), %rax subq %fs:40, %rax jne .L24 addq $152, %rsp .cfi_remember_state .cfi_def_cfa_offset 8 ret .L23: .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 _Z12convolution3PKfPfii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L19 .L24: call __stack_chk_fail@PLT .cfi_endproc .LFE4100: .size _Z37__device_stub__Z12convolution3PKfPfiiPKfPfii, .-_Z37__device_stub__Z12convolution3PKfPfiiPKfPfii .globl _Z12convolution3PKfPfii .type _Z12convolution3PKfPfii, @function _Z12convolution3PKfPfii: .LFB4101: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z37__device_stub__Z12convolution3PKfPfiiPKfPfii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE4101: .size _Z12convolution3PKfPfii, .-_Z12convolution3PKfPfii .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z12convolution3PKfPfii" .LC1: .string "_Z12convolution2PKfPfii" .LC2: .string "_Z12convolution1PKfS0_Pfii" .LC3: .string "c_M" .text .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB4103: .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 .LC0(%rip), %rdx movq %rdx, %rcx leaq _Z12convolution3PKfPfii(%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 .LC1(%rip), %rdx movq %rdx, %rcx leaq _Z12convolution2PKfPfii(%rip), %rsi movq %rbx, %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 .LC2(%rip), %rdx movq %rdx, %rcx leaq _Z12convolution1PKfS0_Pfii(%rip), %rsi movq %rbx, %rdi call __cudaRegisterFunction@PLT addq $32, %rsp .cfi_def_cfa_offset 16 pushq $0 .cfi_def_cfa_offset 24 pushq $1 .cfi_def_cfa_offset 32 movl $20, %r9d movl $0, %r8d leaq .LC3(%rip), %rdx movq %rdx, %rcx leaq _ZL3c_M(%rip), %rsi movq %rbx, %rdi call __cudaRegisterVar@PLT addq $16, %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 .LFE4103: .size _ZL24__sti____cudaRegisterAllv, .-_ZL24__sti____cudaRegisterAllv .section .init_array,"aw" .align 8 .quad _ZL24__sti____cudaRegisterAllv .section .text._ZNSt6vectorIfSaIfEED2Ev,"axG",@progbits,_ZNSt6vectorIfSaIfEED5Ev,comdat .align 2 .weak _ZNSt6vectorIfSaIfEED2Ev .type _ZNSt6vectorIfSaIfEED2Ev, @function _ZNSt6vectorIfSaIfEED2Ev: .LFB4418: .cfi_startproc endbr64 movq (%rdi), %rax testq %rax, %rax je .L32 subq $8, %rsp .cfi_def_cfa_offset 16 movq 16(%rdi), %rsi subq %rax, %rsi movq %rax, %rdi call _ZdlPvm@PLT addq $8, %rsp .cfi_def_cfa_offset 8 ret .L32: ret .cfi_endproc .LFE4418: .size _ZNSt6vectorIfSaIfEED2Ev, .-_ZNSt6vectorIfSaIfEED2Ev .weak _ZNSt6vectorIfSaIfEED1Ev .set _ZNSt6vectorIfSaIfEED1Ev,_ZNSt6vectorIfSaIfEED2Ev .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC5: .string "---------------------------------------------" .align 8 .LC6: .string " GPU PROPERTIES " .section .rodata.str1.1 .LC7: .string "Device Name: " .LC8: .string "Memory Clock Rate: " .LC10: .string " GHz" .LC11: .string "Memory Bandwidth: " .LC12: .string " GB/s" .LC13: .string "Number of SM: " .LC14: .string "Max Threads per SM: " .LC15: .string "Registers per Block: " .LC16: .string "Shared Memory per Block: " .LC17: .string " B" .section .rodata.str1.8 .align 8 .LC18: .string "Total Global Memory per Block: " .section .rodata.str1.1 .LC20: .string " GB" .section .rodata.str1.8 .align 8 .LC21: .string "cannot create std::vector larger than max_size()" .section .rodata.str1.1 .LC27: .string "Wrong inputs!" .LC28: .string "Closing..." .text .globl main .type main, @function main: .LFB4071: .cfi_startproc .cfi_personality 0x9b,DW.ref.__gxx_personality_v0 .cfi_lsda 0x1b,.LLSDA4071 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 $1208, %rsp .cfi_def_cfa_offset 1264 movq %rsi, %r14 movq %fs:40, %rax movq %rax, 1192(%rsp) xorl %eax, %eax leaq 160(%rsp), %rbp movl $0, %esi movq %rbp, %rdi .LEHB0: call cudaGetDeviceProperties_v2@PLT leaq .LC5(%rip), %r12 movq %r12, %rsi leaq _ZSt4cout(%rip), %rbx movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq .LC6(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT movq %r12, %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq .LC7(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi movq %rbp, %rsi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq .LC8(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi pxor %xmm0, %xmm0 cvtsi2sdl 768(%rsp), %xmm0 divsd .LC9(%rip), %xmm0 call _ZNSo9_M_insertIdEERSoT_@PLT movq %rax, %rdi leaq .LC10(%rip), %rsi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq .LC11(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi pxor %xmm0, %xmm0 cvtsi2sdl 768(%rsp), %xmm0 addsd %xmm0, %xmm0 movl 772(%rsp), %eax leal 7(%rax), %edx testl %eax, %eax cmovns %eax, %edx sarl $3, %edx pxor %xmm1, %xmm1 cvtsi2sdl %edx, %xmm1 mulsd %xmm1, %xmm0 divsd .LC9(%rip), %xmm0 call _ZNSo9_M_insertIdEERSoT_@PLT movq %rax, %rdi leaq .LC12(%rip), %rsi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq .LC13(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi movl 548(%rsp), %esi call _ZNSolsEi@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq .LC14(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi movl 784(%rsp), %esi call _ZNSolsEi@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq .LC15(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi movl 464(%rsp), %esi call _ZNSolsEi@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq .LC16(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi movq 456(%rsp), %rsi call _ZNSo9_M_insertImEERSoT_@PLT movq %rax, %rdi leaq .LC17(%rip), %rsi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq .LC18(%rip), %rsi movq %rbx, %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi movq 448(%rsp), %rdx testq %rdx, %rdx js .L36 pxor %xmm0, %xmm0 cvtsi2sdq %rdx, %xmm0 .L37: divsd .LC19(%rip), %xmm0 call _ZNSo9_M_insertIdEERSoT_@PLT movq %rax, %rdi leaq .LC20(%rip), %rsi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT leaq _ZSt4cout(%rip), %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT movq 8(%r14), %rdi movl $10, %edx movl $0, %esi call __isoc23_strtol@PLT movq %rax, %rbp movl %eax, %r15d movslq %eax, %r13 movq %r13, %rax shrq $61, %rax jne .L66 testq %r13, %r13 je .L40 leaq 0(,%r13,4), %rbx movq %rbx, %rdi call _Znwm@PLT .LEHE0: movq %rax, %r12 movq %rax, 32(%rsp) leaq (%rax,%rbx), %rdx movq %rdx, 48(%rsp) movss .LC22(%rip), %xmm0 .L41: movss %xmm0, (%rax) addq $4, %rax cmpq %rdx, %rax jne .L41 movq %rdx, 40(%rsp) movq %r12, %rax pxor %xmm0, %xmm0 movss .LC22(%rip), %xmm1 .L42: movss %xmm0, (%rax) addss %xmm1, %xmm0 addq $4, %rax cmpq %rdx, %rax jne .L42 .L56: movq %rsp, %rdi movq %rbx, %rsi .LEHB1: call cudaMalloc@PLT .LEHE1: jmp .L67 .L36: movq %rdx, %rax shrq %rax andl $1, %edx orq %rdx, %rax pxor %xmm0, %xmm0 cvtsi2sdq %rax, %xmm0 addsd %xmm0, %xmm0 jmp .L37 .L66: movq 1192(%rsp), %rax subq %fs:40, %rax jne .L68 leaq .LC21(%rip), %rdi .LEHB2: call _ZSt20__throw_length_errorPKc@PLT .LEHE2: .L68: call __stack_chk_fail@PLT .L67: movl $1, %ecx movq %rbx, %rdx movq %r12, %rsi movq (%rsp), %rdi .LEHB3: call cudaMemcpy@PLT movss .LC22(%rip), %xmm0 movss %xmm0, 128(%rsp) movss %xmm0, 132(%rsp) movl $0x40000000, 136(%rsp) movss %xmm0, 140(%rsp) movss %xmm0, 144(%rsp) movq $0, 72(%rsp) movq $0, 80(%rsp) movl $20, %edi call _Znwm@PLT .LEHE3: movq %rax, %rsi movq %rax, 64(%rsp) leaq 20(%rax), %rax movq %rax, 80(%rsp) movdqa 128(%rsp), %xmm5 movups %xmm5, (%rsi) movl 144(%rsp), %edx movl %edx, 16(%rsi) movq %rax, 72(%rsp) movl $1, %r8d movl $0, %ecx movl $20, %edx leaq _ZL3c_M(%rip), %rdi .LEHB4: call cudaMemcpyToSymbol@PLT leaq 8(%rsp), %rdi movq %rbx, %rsi call cudaMalloc@PLT pxor %xmm0, %xmm0 cvtsi2ssl %ebp, %xmm0 mulss .LC24(%rip), %xmm0 movaps %xmm0, %xmm3 movss .LC29(%rip), %xmm2 movaps %xmm0, %xmm1 andps %xmm2, %xmm1 movss .LC25(%rip), %xmm4 ucomiss %xmm1, %xmm4 jbe .L43 cvttss2sil %xmm0, %eax pxor %xmm1, %xmm1 cvtsi2ssl %eax, %xmm1 cmpnless %xmm1, %xmm3 movss .LC22(%rip), %xmm4 andps %xmm4, %xmm3 addss %xmm1, %xmm3 andnps %xmm0, %xmm2 orps %xmm2, %xmm3 .L43: cvttss2sil %xmm3, %ebp movq 16(%r14), %rdi movl $10, %edx movl $0, %esi call __isoc23_strtol@PLT cmpl $1, %eax je .L69 cmpl $2, %eax je .L70 leaq .LC27(%rip), %rsi leaq _ZSt4cout(%rip), %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT jmp .L71 .L69: movl $64, 96(%rsp) movl $1, 100(%rsp) movl $1, 104(%rsp) movl %ebp, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $0, %r9d movl $0, %r8d movq 96(%rsp), %rdx movl $1, %ecx movq 20(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax jne .L46 movl %r15d, %ecx movl $5, %edx movq 8(%rsp), %rsi movq (%rsp), %rdi call _Z37__device_stub__Z12convolution2PKfPfiiPKfPfii jmp .L46 .L70: movl $64, 96(%rsp) movl $1, 100(%rsp) movl $1, 104(%rsp) movl %ebp, 20(%rsp) movl $1, 24(%rsp) movl $1, 28(%rsp) movl $0, %r9d movl $0, %r8d movq 96(%rsp), %rdx movl $1, %ecx movq 20(%rsp), %rdi movl $1, %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L72 .L46: movq $0, 104(%rsp) movq $0, 112(%rsp) testq %r13, %r13 je .L49 movq %rbx, %rdi call _Znwm@PLT jmp .L73 .L72: movl %r15d, %ecx movl $5, %edx movq 8(%rsp), %rsi movq (%rsp), %rdi call _Z37__device_stub__Z12convolution3PKfPfiiPKfPfii jmp .L46 .L71: movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT .LEHE4: jmp .L46 .L73: movq %rax, %rdi movq %rax, 96(%rsp) leaq (%rax,%rbx), %rax movq %rax, 112(%rsp) movl $0x00000000, (%rdi) leaq 4(%rdi), %rdx cmpq $1, %r13 je .L50 cmpq %rdx, %rax je .L50 .L51: movl $0x00000000, (%rdx) addq $4, %rdx cmpq %rdx, %rax jne .L51 movq %rax, %rdx jmp .L50 .L75: movq (%rsp), %rdi .LEHB5: call cudaFree@PLT movq 8(%rsp), %rdi call cudaFree@PLT leaq .LC28(%rip), %rsi leaq _ZSt4cout(%rip), %rdi call _ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc@PLT movq %rax, %rdi call _ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_@PLT .LEHE5: leaq 96(%rsp), %rdi call _ZNSt6vectorIfSaIfEED1Ev leaq 64(%rsp), %rdi call _ZNSt6vectorIfSaIfEED1Ev leaq 32(%rsp), %rdi call _ZNSt6vectorIfSaIfEED1Ev movq 1192(%rsp), %rax subq %fs:40, %rax jne .L74 movl $0, %eax addq $1208, %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 .L60: .cfi_restore_state endbr64 movq %rax, %rbx leaq 96(%rsp), %rdi call _ZNSt6vectorIfSaIfEED1Ev .L53: leaq 64(%rsp), %rdi call _ZNSt6vectorIfSaIfEED1Ev .L54: leaq 32(%rsp), %rdi call _ZNSt6vectorIfSaIfEED1Ev movq 1192(%rsp), %rax subq %fs:40, %rax je .L55 call __stack_chk_fail@PLT .L59: endbr64 movq %rax, %rbx jmp .L53 .L58: endbr64 movq %rax, %rbx jmp .L54 .L55: movq %rbx, %rdi .LEHB6: call _Unwind_Resume@PLT .LEHE6: .L49: movq $0, 96(%rsp) movq %rbx, 112(%rsp) movl $0, %edi movl $0, %edx .L50: movq %rdx, 104(%rsp) movl $2, %ecx movq %rbx, %rdx movq 8(%rsp), %rsi .LEHB7: call cudaMemcpy@PLT .LEHE7: jmp .L75 .L40: movq $0, 32(%rsp) movq $0, 48(%rsp) movq $0, 40(%rsp) movq %r13, %rbx movl $0, %r12d jmp .L56 .L74: call __stack_chk_fail@PLT .cfi_endproc .LFE4071: .globl __gxx_personality_v0 .section .gcc_except_table,"a",@progbits .LLSDA4071: .byte 0xff .byte 0xff .byte 0x1 .uleb128 .LLSDACSE4071-.LLSDACSB4071 .LLSDACSB4071: .uleb128 .LEHB0-.LFB4071 .uleb128 .LEHE0-.LEHB0 .uleb128 0 .uleb128 0 .uleb128 .LEHB1-.LFB4071 .uleb128 .LEHE1-.LEHB1 .uleb128 .L58-.LFB4071 .uleb128 0 .uleb128 .LEHB2-.LFB4071 .uleb128 .LEHE2-.LEHB2 .uleb128 0 .uleb128 0 .uleb128 .LEHB3-.LFB4071 .uleb128 .LEHE3-.LEHB3 .uleb128 .L58-.LFB4071 .uleb128 0 .uleb128 .LEHB4-.LFB4071 .uleb128 .LEHE4-.LEHB4 .uleb128 .L59-.LFB4071 .uleb128 0 .uleb128 .LEHB5-.LFB4071 .uleb128 .LEHE5-.LEHB5 .uleb128 .L60-.LFB4071 .uleb128 0 .uleb128 .LEHB6-.LFB4071 .uleb128 .LEHE6-.LEHB6 .uleb128 0 .uleb128 0 .uleb128 .LEHB7-.LFB4071 .uleb128 .LEHE7-.LEHB7 .uleb128 .L60-.LFB4071 .uleb128 0 .LLSDACSE4071: .text .size main, .-main .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 _ZL3c_M .comm _ZL3c_M,20,16 .section .rodata.cst8,"aM",@progbits,8 .align 8 .LC9: .long 0 .long 1093567616 .align 8 .LC19: .long 0 .long 1104006501 .section .rodata.cst4,"aM",@progbits,4 .align 4 .LC22: .long 1065353216 .align 4 .LC24: .long 1015021568 .align 4 .LC25: .long 1258291200 .align 4 .LC29: .long 2147483647 .hidden DW.ref.__gxx_personality_v0 .weak DW.ref.__gxx_personality_v0 .section .data.rel.local.DW.ref.__gxx_personality_v0,"awG",@progbits,DW.ref.__gxx_personality_v0,comdat .align 8 .type DW.ref.__gxx_personality_v0, @object .size DW.ref.__gxx_personality_v0, 8 DW.ref.__gxx_personality_v0: .quad __gxx_personality_v0 .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 <cassert> #include <cuda.h> #include <cuda_runtime.h> #include <iostream> #include <numeric> #include <vector> using namespace std; #define TILE_SIZE 64 #define MAX_MASK_WIDTH 5 __constant__ float c_M[MAX_MASK_WIDTH]; __global__ void convolution1(const float* N, const float* M, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width/2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j]*M[j]; } } P[i] = Pvalue; } __global__ void convolution2(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width / 2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j] * c_M[j]; } } P[i] = Pvalue; } __global__ void convolution3(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; __shared__ float Nds[TILE_SIZE + MAX_MASK_WIDTH - 1]; // load N from global memory into shared memory int n = mask_width/2; if (threadIdx.x >= blockDim.x - n) { int halo_index_left = (blockIdx.x - 1)*blockDim.x + threadIdx.x; Nds[threadIdx.x - (blockDim.x - n)] = (halo_index_left < 0) ? 0 : N[halo_index_left]; } Nds[n + threadIdx.x] = N[i]; if (threadIdx.x < n) { int halo_index_right = (blockIdx.x + 1)*blockDim.x + threadIdx.x; Nds[n + blockDim.x + threadIdx.x] = (halo_index_right >= width) ? 0 : N[halo_index_right]; } __syncthreads(); float Pvalue = 0.0f; for (int j = 0; j < mask_width; ++j) { Pvalue += Nds[threadIdx.x + j]*c_M[j]; } P[i] = Pvalue; } int main(int argc, char* argv[]) { // Query GPU properties cudaDeviceProp dev_prop; cudaGetDeviceProperties(&dev_prop, 0); cout << "---------------------------------------------" << endl; cout << " GPU PROPERTIES " << endl; cout << "---------------------------------------------" << endl; cout << "Device Name: " << dev_prop.name << endl; cout << "Memory Clock Rate: " << dev_prop.memoryClockRate/1.0e6 << " GHz" << endl; cout << "Memory Bandwidth: " << 2.0*dev_prop.memoryClockRate*(dev_prop.memoryBusWidth/8)/1.0e6 << " GB/s" << endl; cout << "Number of SM: " << dev_prop.multiProcessorCount << endl; cout << "Max Threads per SM: " << dev_prop.maxThreadsPerMultiProcessor << endl; cout << "Registers per Block: " << dev_prop.regsPerBlock << endl; cout << "Shared Memory per Block: " << dev_prop.sharedMemPerBlock << " B" << endl; cout << "Total Global Memory per Block: " << dev_prop.totalGlobalMem/1.0e9 << " GB" << endl; cout << endl; int size = atoi(argv[1]); // creating vector on host side vector<float> h_N(size, 1.0f); std::iota(h_N.begin(), h_N.end(), 0.0f); // Copy vector on device side float* d_N; cudaMalloc((void**)&d_N, size*sizeof(float)); cudaMemcpy((void*)d_N, (void*)h_N.data(), size*sizeof(float), cudaMemcpyHostToDevice); // Create mask and send to devide vector<float> h_M = { 1.0f, 1.0f, 2.0f, 1.0f, 1.0f }; int mask_width = h_M.size(); assert(mask_width < MAX_MASK_WIDTH); cudaMemcpyToSymbol(c_M, (void*)h_M.data(), mask_width*sizeof(float)); // Allocate space for solution on device float* d_P; cudaMalloc((void**)&d_P, size*sizeof(float)); // call Kernel int blockDim = TILE_SIZE; int gridDim = ceil(size/(float)blockDim); int version = atoi(argv[2]); if(version == 1) convolution2<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else if(version == 2) convolution3<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else cout << "Wrong inputs!" << endl; // Recover vector from device to host vector<float> h_P(size); cudaMemcpy((void*)h_P.data(), (void*)d_P, size*sizeof(float), cudaMemcpyDeviceToHost); // Finalize storage cudaFree(d_N); cudaFree(d_P); cout << "Closing..." << endl; return 0; }
#include <cassert> #include <hip/hip_runtime.h> #include <iostream> #include <numeric> #include <vector> using namespace std; #define TILE_SIZE 64 #define MAX_MASK_WIDTH 5 __constant__ float c_M[MAX_MASK_WIDTH]; __global__ void convolution1(const float* N, const float* M, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width/2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j]*M[j]; } } P[i] = Pvalue; } __global__ void convolution2(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width / 2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j] * c_M[j]; } } P[i] = Pvalue; } __global__ void convolution3(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; __shared__ float Nds[TILE_SIZE + MAX_MASK_WIDTH - 1]; // load N from global memory into shared memory int n = mask_width/2; if (threadIdx.x >= blockDim.x - n) { int halo_index_left = (blockIdx.x - 1)*blockDim.x + threadIdx.x; Nds[threadIdx.x - (blockDim.x - n)] = (halo_index_left < 0) ? 0 : N[halo_index_left]; } Nds[n + threadIdx.x] = N[i]; if (threadIdx.x < n) { int halo_index_right = (blockIdx.x + 1)*blockDim.x + threadIdx.x; Nds[n + blockDim.x + threadIdx.x] = (halo_index_right >= width) ? 0 : N[halo_index_right]; } __syncthreads(); float Pvalue = 0.0f; for (int j = 0; j < mask_width; ++j) { Pvalue += Nds[threadIdx.x + j]*c_M[j]; } P[i] = Pvalue; } int main(int argc, char* argv[]) { // Query GPU properties hipDeviceProp_t dev_prop; hipGetDeviceProperties(&dev_prop, 0); cout << "---------------------------------------------" << endl; cout << " GPU PROPERTIES " << endl; cout << "---------------------------------------------" << endl; cout << "Device Name: " << dev_prop.name << endl; cout << "Memory Clock Rate: " << dev_prop.memoryClockRate/1.0e6 << " GHz" << endl; cout << "Memory Bandwidth: " << 2.0*dev_prop.memoryClockRate*(dev_prop.memoryBusWidth/8)/1.0e6 << " GB/s" << endl; cout << "Number of SM: " << dev_prop.multiProcessorCount << endl; cout << "Max Threads per SM: " << dev_prop.maxThreadsPerMultiProcessor << endl; cout << "Registers per Block: " << dev_prop.regsPerBlock << endl; cout << "Shared Memory per Block: " << dev_prop.sharedMemPerBlock << " B" << endl; cout << "Total Global Memory per Block: " << dev_prop.totalGlobalMem/1.0e9 << " GB" << endl; cout << endl; int size = atoi(argv[1]); // creating vector on host side vector<float> h_N(size, 1.0f); std::iota(h_N.begin(), h_N.end(), 0.0f); // Copy vector on device side float* d_N; hipMalloc((void**)&d_N, size*sizeof(float)); hipMemcpy((void*)d_N, (void*)h_N.data(), size*sizeof(float), hipMemcpyHostToDevice); // Create mask and send to devide vector<float> h_M = { 1.0f, 1.0f, 2.0f, 1.0f, 1.0f }; int mask_width = h_M.size(); assert(mask_width < MAX_MASK_WIDTH); hipMemcpyToSymbol(HIP_SYMBOL(c_M), (void*)h_M.data(), mask_width*sizeof(float)); // Allocate space for solution on device float* d_P; hipMalloc((void**)&d_P, size*sizeof(float)); // call Kernel int blockDim = TILE_SIZE; int gridDim = ceil(size/(float)blockDim); int version = atoi(argv[2]); if(version == 1) convolution2<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else if(version == 2) convolution3<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else cout << "Wrong inputs!" << endl; // Recover vector from device to host vector<float> h_P(size); hipMemcpy((void*)h_P.data(), (void*)d_P, size*sizeof(float), hipMemcpyDeviceToHost); // Finalize storage hipFree(d_N); hipFree(d_P); cout << "Closing..." << endl; return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include <cassert> #include <hip/hip_runtime.h> #include <iostream> #include <numeric> #include <vector> using namespace std; #define TILE_SIZE 64 #define MAX_MASK_WIDTH 5 __constant__ float c_M[MAX_MASK_WIDTH]; __global__ void convolution1(const float* N, const float* M, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width/2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j]*M[j]; } } P[i] = Pvalue; } __global__ void convolution2(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width / 2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j] * c_M[j]; } } P[i] = Pvalue; } __global__ void convolution3(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; __shared__ float Nds[TILE_SIZE + MAX_MASK_WIDTH - 1]; // load N from global memory into shared memory int n = mask_width/2; if (threadIdx.x >= blockDim.x - n) { int halo_index_left = (blockIdx.x - 1)*blockDim.x + threadIdx.x; Nds[threadIdx.x - (blockDim.x - n)] = (halo_index_left < 0) ? 0 : N[halo_index_left]; } Nds[n + threadIdx.x] = N[i]; if (threadIdx.x < n) { int halo_index_right = (blockIdx.x + 1)*blockDim.x + threadIdx.x; Nds[n + blockDim.x + threadIdx.x] = (halo_index_right >= width) ? 0 : N[halo_index_right]; } __syncthreads(); float Pvalue = 0.0f; for (int j = 0; j < mask_width; ++j) { Pvalue += Nds[threadIdx.x + j]*c_M[j]; } P[i] = Pvalue; } int main(int argc, char* argv[]) { // Query GPU properties hipDeviceProp_t dev_prop; hipGetDeviceProperties(&dev_prop, 0); cout << "---------------------------------------------" << endl; cout << " GPU PROPERTIES " << endl; cout << "---------------------------------------------" << endl; cout << "Device Name: " << dev_prop.name << endl; cout << "Memory Clock Rate: " << dev_prop.memoryClockRate/1.0e6 << " GHz" << endl; cout << "Memory Bandwidth: " << 2.0*dev_prop.memoryClockRate*(dev_prop.memoryBusWidth/8)/1.0e6 << " GB/s" << endl; cout << "Number of SM: " << dev_prop.multiProcessorCount << endl; cout << "Max Threads per SM: " << dev_prop.maxThreadsPerMultiProcessor << endl; cout << "Registers per Block: " << dev_prop.regsPerBlock << endl; cout << "Shared Memory per Block: " << dev_prop.sharedMemPerBlock << " B" << endl; cout << "Total Global Memory per Block: " << dev_prop.totalGlobalMem/1.0e9 << " GB" << endl; cout << endl; int size = atoi(argv[1]); // creating vector on host side vector<float> h_N(size, 1.0f); std::iota(h_N.begin(), h_N.end(), 0.0f); // Copy vector on device side float* d_N; hipMalloc((void**)&d_N, size*sizeof(float)); hipMemcpy((void*)d_N, (void*)h_N.data(), size*sizeof(float), hipMemcpyHostToDevice); // Create mask and send to devide vector<float> h_M = { 1.0f, 1.0f, 2.0f, 1.0f, 1.0f }; int mask_width = h_M.size(); assert(mask_width < MAX_MASK_WIDTH); hipMemcpyToSymbol(HIP_SYMBOL(c_M), (void*)h_M.data(), mask_width*sizeof(float)); // Allocate space for solution on device float* d_P; hipMalloc((void**)&d_P, size*sizeof(float)); // call Kernel int blockDim = TILE_SIZE; int gridDim = ceil(size/(float)blockDim); int version = atoi(argv[2]); if(version == 1) convolution2<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else if(version == 2) convolution3<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else cout << "Wrong inputs!" << endl; // Recover vector from device to host vector<float> h_P(size); hipMemcpy((void*)h_P.data(), (void*)d_P, size*sizeof(float), hipMemcpyDeviceToHost); // Finalize storage hipFree(d_N); hipFree(d_P); cout << "Closing..." << endl; return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z12convolution1PKfS0_Pfii .globl _Z12convolution1PKfS0_Pfii .p2align 8 .type _Z12convolution1PKfS0_Pfii,@function _Z12convolution1PKfS0_Pfii: s_clause 0x1 s_load_b32 s2, s[0:1], 0x2c s_load_b32 s3, s[0:1], 0x18 s_waitcnt lgkmcnt(0) s_and_b32 s2, s2, 0xffff s_cmp_lt_i32 s3, 1 v_mad_u64_u32 v[1:2], null, s15, s2, v[0:1] s_cbranch_scc1 .LBB0_5 s_clause 0x1 s_load_b128 s[4:7], s[0:1], 0x0 s_load_b32 s8, s[0:1], 0x1c s_lshr_b32 s2, s3, 31 v_mov_b32_e32 v3, 0 s_add_i32 s2, s3, s2 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_ashr_i32 s2, s2, 1 v_subrev_nc_u32_e32 v2, s2, v1 s_delay_alu instid0(VALU_DEP_2) v_mov_b32_e32 v0, v3 s_branch .LBB0_3 .p2align 6 .LBB0_2: s_or_b32 exec_lo, exec_lo, s2 s_add_i32 s3, s3, -1 v_add_nc_u32_e32 v2, 1, v2 s_add_u32 s6, s6, 4 s_addc_u32 s7, s7, 0 s_cmp_eq_u32 s3, 0 s_cbranch_scc1 .LBB0_6 .LBB0_3: s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_cmp_lt_i32_e32 vcc_lo, -1, v2 s_waitcnt lgkmcnt(0) v_cmp_gt_i32_e64 s2, s8, v2 s_and_b32 s9, vcc_lo, s2 s_delay_alu instid0(SALU_CYCLE_1) s_and_saveexec_b32 s2, s9 s_cbranch_execz .LBB0_2 v_lshlrev_b64 v[4:5], 2, v[2:3] s_load_b32 s9, s[6:7], 0x0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v4, vcc_lo, s4, v4 v_add_co_ci_u32_e32 v5, vcc_lo, s5, v5, vcc_lo global_load_b32 v4, v[4:5], off s_waitcnt vmcnt(0) lgkmcnt(0) v_fmac_f32_e32 v0, s9, v4 s_branch .LBB0_2 .LBB0_5: v_mov_b32_e32 v0, 0 .LBB0_6: s_load_b64 s[0:1], s[0:1], 0x10 v_mov_b32_e32 v2, 0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[1:2], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v1, vcc_lo, s0, v1 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v2, vcc_lo, s1, v2, vcc_lo global_store_b32 v[1:2], v0, off s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z12convolution1PKfS0_Pfii .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 288 .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 _Z12convolution1PKfS0_Pfii, .Lfunc_end0-_Z12convolution1PKfS0_Pfii .section .AMDGPU.csdata,"",@progbits .text .protected _Z12convolution2PKfPfii .globl _Z12convolution2PKfPfii .p2align 8 .type _Z12convolution2PKfPfii,@function _Z12convolution2PKfPfii: 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_cmp_lt_i32 s3, 1 v_mad_u64_u32 v[1:2], null, s15, s2, v[0:1] s_cbranch_scc1 .LBB1_5 s_clause 0x1 s_load_b64 s[4:5], s[0:1], 0x0 s_load_b32 s8, s[0:1], 0x14 s_lshr_b32 s2, s3, 31 v_mov_b32_e32 v3, 0 s_add_i32 s2, s3, s2 s_delay_alu instid0(SALU_CYCLE_1) s_ashr_i32 s2, s2, 1 s_getpc_b64 s[6:7] s_add_u32 s6, s6, c_M@rel32@lo+4 s_addc_u32 s7, s7, c_M@rel32@hi+12 v_subrev_nc_u32_e32 v2, s2, v1 v_mov_b32_e32 v0, v3 s_branch .LBB1_3 .p2align 6 .LBB1_2: s_or_b32 exec_lo, exec_lo, s2 s_add_i32 s3, s3, -1 v_add_nc_u32_e32 v2, 1, v2 s_add_u32 s6, s6, 4 s_addc_u32 s7, s7, 0 s_cmp_eq_u32 s3, 0 s_cbranch_scc1 .LBB1_6 .LBB1_3: s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_cmp_lt_i32_e32 vcc_lo, -1, v2 s_waitcnt lgkmcnt(0) v_cmp_gt_i32_e64 s2, s8, v2 s_and_b32 s9, vcc_lo, s2 s_delay_alu instid0(SALU_CYCLE_1) s_and_saveexec_b32 s2, s9 s_cbranch_execz .LBB1_2 v_lshlrev_b64 v[4:5], 2, v[2:3] s_load_b32 s9, s[6:7], 0x0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2) v_add_co_u32 v4, vcc_lo, s4, v4 v_add_co_ci_u32_e32 v5, vcc_lo, s5, v5, vcc_lo global_load_b32 v4, v[4:5], off s_waitcnt vmcnt(0) lgkmcnt(0) v_fmac_f32_e32 v0, s9, v4 s_branch .LBB1_2 .LBB1_5: v_mov_b32_e32 v0, 0 .LBB1_6: s_load_b64 s[0:1], s[0:1], 0x8 v_mov_b32_e32 v2, 0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[1:2], 2, v[1:2] s_waitcnt lgkmcnt(0) v_add_co_u32 v1, vcc_lo, s0, v1 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v2, vcc_lo, s1, v2, vcc_lo global_store_b32 v[1:2], v0, off s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z12convolution2PKfPfii .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 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_end1: .size _Z12convolution2PKfPfii, .Lfunc_end1-_Z12convolution2PKfPfii .section .AMDGPU.csdata,"",@progbits .text .protected _Z12convolution3PKfPfii .globl _Z12convolution3PKfPfii .p2align 8 .type _Z12convolution3PKfPfii,@function _Z12convolution3PKfPfii: s_clause 0x2 s_load_b32 s4, s[0:1], 0x10 s_load_b32 s5, s[0:1], 0x24 s_load_b64 s[2:3], s[0:1], 0x0 s_waitcnt lgkmcnt(0) s_lshr_b32 s6, s4, 31 s_and_b32 s5, s5, 0xffff s_add_i32 s6, s4, s6 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_ashr_i32 s6, s6, 1 s_sub_i32 s7, s5, s6 s_delay_alu instid0(SALU_CYCLE_1) v_cmp_le_u32_e32 vcc_lo, s7, v0 s_and_saveexec_b32 s7, vcc_lo s_cbranch_execz .LBB2_4 s_add_i32 s8, s15, -1 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_2) v_mad_u64_u32 v[1:2], null, s8, s5, v[0:1] v_mov_b32_e32 v2, 0 s_mov_b32 s8, exec_lo v_cmpx_lt_i32_e32 -1, v1 s_cbranch_execz .LBB2_3 v_mov_b32_e32 v2, 0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b64 v[1:2], 2, v[1:2] v_add_co_u32 v1, vcc_lo, s2, v1 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v2, vcc_lo, s3, v2, vcc_lo global_load_b32 v2, v[1:2], off .LBB2_3: s_or_b32 exec_lo, exec_lo, s8 v_add_nc_u32_e32 v1, s6, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_subrev_nc_u32_e32 v1, s5, v1 v_lshlrev_b32_e32 v1, 2, v1 s_waitcnt vmcnt(0) ds_store_b32 v1, v2 .LBB2_4: s_or_b32 exec_lo, exec_lo, s7 v_mad_u64_u32 v[1:2], null, s15, s5, v[0:1] v_mov_b32_e32 v2, 0 s_mov_b32 s7, exec_lo s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b64 v[3:4], 2, v[1:2] v_add_co_u32 v3, vcc_lo, s2, v3 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v4, vcc_lo, s3, v4, vcc_lo global_load_b32 v3, v[3:4], off v_add_lshl_u32 v4, s6, v0, 2 s_waitcnt vmcnt(0) ds_store_b32 v4, v3 v_cmpx_gt_u32_e64 s6, v0 s_cbranch_execz .LBB2_8 s_load_b32 s8, s[0:1], 0x14 s_add_i32 s15, s15, 1 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(VALU_DEP_2) v_mad_u64_u32 v[3:4], null, s15, s5, v[0:1] v_mov_b32_e32 v4, v2 s_waitcnt lgkmcnt(0) v_cmp_gt_i32_e32 vcc_lo, s8, v3 s_and_saveexec_b32 s8, vcc_lo s_cbranch_execz .LBB2_7 v_ashrrev_i32_e32 v4, 31, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b64 v[3:4], 2, v[3:4] v_add_co_u32 v3, vcc_lo, s2, v3 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v4, vcc_lo, s3, v4, vcc_lo global_load_b32 v4, v[3:4], off .LBB2_7: s_or_b32 exec_lo, exec_lo, s8 s_add_i32 s5, s5, s6 s_delay_alu instid0(SALU_CYCLE_1) v_add_lshl_u32 v3, s5, v0, 2 s_waitcnt vmcnt(0) ds_store_b32 v3, v4 .LBB2_8: s_or_b32 exec_lo, exec_lo, s7 s_cmp_lt_i32 s4, 1 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_cbranch_scc1 .LBB2_11 v_dual_mov_b32 v0, 0 :: v_dual_lshlrev_b32 v3, 2, v0 s_getpc_b64 s[2:3] s_add_u32 s2, s2, c_M@rel32@lo+4 s_addc_u32 s3, s3, c_M@rel32@hi+12 .LBB2_10: ds_load_b32 v4, v3 s_load_b32 s5, s[2:3], 0x0 s_add_i32 s4, s4, -1 v_add_nc_u32_e32 v3, 4, v3 s_add_u32 s2, s2, 4 s_addc_u32 s3, s3, 0 s_cmp_eq_u32 s4, 0 s_waitcnt lgkmcnt(0) v_fmac_f32_e32 v0, s5, v4 s_cbranch_scc0 .LBB2_10 s_branch .LBB2_12 .LBB2_11: v_mov_b32_e32 v0, 0 .LBB2_12: s_load_b64 s[0:1], s[0:1], 0x8 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_store_b32 v[1:2], v0, off s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z12convolution3PKfPfii .amdhsa_group_segment_fixed_size 272 .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 5 .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_end2: .size _Z12convolution3PKfPfii, .Lfunc_end2-_Z12convolution3PKfPfii .section .AMDGPU.csdata,"",@progbits .text .p2alignl 7, 3214868480 .fill 96, 4, 3214868480 .protected c_M .type c_M,@object .section .bss,"aw",@nobits .globl c_M .p2align 4, 0x0 c_M: .zero 20 .size c_M, 20 .type __hip_cuid_,@object .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 c_M .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 - .offset: 28 .size: 4 .value_kind: by_value - .offset: 32 .size: 4 .value_kind: hidden_block_count_x - .offset: 36 .size: 4 .value_kind: hidden_block_count_y - .offset: 40 .size: 4 .value_kind: hidden_block_count_z - .offset: 44 .size: 2 .value_kind: hidden_group_size_x - .offset: 46 .size: 2 .value_kind: hidden_group_size_y - .offset: 48 .size: 2 .value_kind: hidden_group_size_z - .offset: 50 .size: 2 .value_kind: hidden_remainder_x - .offset: 52 .size: 2 .value_kind: hidden_remainder_y - .offset: 54 .size: 2 .value_kind: hidden_remainder_z - .offset: 72 .size: 8 .value_kind: hidden_global_offset_x - .offset: 80 .size: 8 .value_kind: hidden_global_offset_y - .offset: 88 .size: 8 .value_kind: hidden_global_offset_z - .offset: 96 .size: 2 .value_kind: hidden_grid_dims .group_segment_fixed_size: 0 .kernarg_segment_align: 8 .kernarg_segment_size: 288 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z12convolution1PKfS0_Pfii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z12convolution1PKfS0_Pfii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 6 .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 - .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: _Z12convolution2PKfPfii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z12convolution2PKfPfii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 6 .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 - .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: 272 .kernarg_segment_align: 8 .kernarg_segment_size: 280 .language: OpenCL C .language_version: - 2 - 0 .max_flat_workgroup_size: 1024 .name: _Z12convolution3PKfPfii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z12convolution3PKfPfii.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 5 .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 <cassert> #include <hip/hip_runtime.h> #include <iostream> #include <numeric> #include <vector> using namespace std; #define TILE_SIZE 64 #define MAX_MASK_WIDTH 5 __constant__ float c_M[MAX_MASK_WIDTH]; __global__ void convolution1(const float* N, const float* M, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width/2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j]*M[j]; } } P[i] = Pvalue; } __global__ void convolution2(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; float Pvalue = 0.0f; int N_start_point = i - (mask_width / 2); for (int j = 0; j < mask_width; ++j) { if (N_start_point + j >= 0 && N_start_point + j < width) { Pvalue += N[N_start_point + j] * c_M[j]; } } P[i] = Pvalue; } __global__ void convolution3(const float* N, float* P, int mask_width, int width) { unsigned int i = blockIdx.x*blockDim.x + threadIdx.x; __shared__ float Nds[TILE_SIZE + MAX_MASK_WIDTH - 1]; // load N from global memory into shared memory int n = mask_width/2; if (threadIdx.x >= blockDim.x - n) { int halo_index_left = (blockIdx.x - 1)*blockDim.x + threadIdx.x; Nds[threadIdx.x - (blockDim.x - n)] = (halo_index_left < 0) ? 0 : N[halo_index_left]; } Nds[n + threadIdx.x] = N[i]; if (threadIdx.x < n) { int halo_index_right = (blockIdx.x + 1)*blockDim.x + threadIdx.x; Nds[n + blockDim.x + threadIdx.x] = (halo_index_right >= width) ? 0 : N[halo_index_right]; } __syncthreads(); float Pvalue = 0.0f; for (int j = 0; j < mask_width; ++j) { Pvalue += Nds[threadIdx.x + j]*c_M[j]; } P[i] = Pvalue; } int main(int argc, char* argv[]) { // Query GPU properties hipDeviceProp_t dev_prop; hipGetDeviceProperties(&dev_prop, 0); cout << "---------------------------------------------" << endl; cout << " GPU PROPERTIES " << endl; cout << "---------------------------------------------" << endl; cout << "Device Name: " << dev_prop.name << endl; cout << "Memory Clock Rate: " << dev_prop.memoryClockRate/1.0e6 << " GHz" << endl; cout << "Memory Bandwidth: " << 2.0*dev_prop.memoryClockRate*(dev_prop.memoryBusWidth/8)/1.0e6 << " GB/s" << endl; cout << "Number of SM: " << dev_prop.multiProcessorCount << endl; cout << "Max Threads per SM: " << dev_prop.maxThreadsPerMultiProcessor << endl; cout << "Registers per Block: " << dev_prop.regsPerBlock << endl; cout << "Shared Memory per Block: " << dev_prop.sharedMemPerBlock << " B" << endl; cout << "Total Global Memory per Block: " << dev_prop.totalGlobalMem/1.0e9 << " GB" << endl; cout << endl; int size = atoi(argv[1]); // creating vector on host side vector<float> h_N(size, 1.0f); std::iota(h_N.begin(), h_N.end(), 0.0f); // Copy vector on device side float* d_N; hipMalloc((void**)&d_N, size*sizeof(float)); hipMemcpy((void*)d_N, (void*)h_N.data(), size*sizeof(float), hipMemcpyHostToDevice); // Create mask and send to devide vector<float> h_M = { 1.0f, 1.0f, 2.0f, 1.0f, 1.0f }; int mask_width = h_M.size(); assert(mask_width < MAX_MASK_WIDTH); hipMemcpyToSymbol(HIP_SYMBOL(c_M), (void*)h_M.data(), mask_width*sizeof(float)); // Allocate space for solution on device float* d_P; hipMalloc((void**)&d_P, size*sizeof(float)); // call Kernel int blockDim = TILE_SIZE; int gridDim = ceil(size/(float)blockDim); int version = atoi(argv[2]); if(version == 1) convolution2<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else if(version == 2) convolution3<<<gridDim, blockDim>>>(d_N, d_P, mask_width, size); else cout << "Wrong inputs!" << endl; // Recover vector from device to host vector<float> h_P(size); hipMemcpy((void*)h_P.data(), (void*)d_P, size*sizeof(float), hipMemcpyDeviceToHost); // Finalize storage hipFree(d_N); hipFree(d_P); cout << "Closing..." << endl; return 0; }
.text .file "main.hip" # Start of file scope inline assembly .globl _ZSt21ios_base_library_initv # End of file scope inline assembly .globl _Z27__device_stub__convolution1PKfS0_Pfii # -- Begin function _Z27__device_stub__convolution1PKfS0_Pfii .p2align 4, 0x90 .type _Z27__device_stub__convolution1PKfS0_Pfii,@function _Z27__device_stub__convolution1PKfS0_Pfii: # @_Z27__device_stub__convolution1PKfS0_Pfii .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) movl %r8d, (%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) movq %rsp, %rax movq %rax, 112(%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 $_Z12convolution1PKfS0_Pfii, %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 _Z27__device_stub__convolution1PKfS0_Pfii, .Lfunc_end0-_Z27__device_stub__convolution1PKfS0_Pfii .cfi_endproc # -- End function .globl _Z27__device_stub__convolution2PKfPfii # -- Begin function _Z27__device_stub__convolution2PKfPfii .p2align 4, 0x90 .type _Z27__device_stub__convolution2PKfPfii,@function _Z27__device_stub__convolution2PKfPfii: # @_Z27__device_stub__convolution2PKfPfii .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%rsp) movl %ecx, 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 $_Z12convolution2PKfPfii, %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_end1: .size _Z27__device_stub__convolution2PKfPfii, .Lfunc_end1-_Z27__device_stub__convolution2PKfPfii .cfi_endproc # -- End function .globl _Z27__device_stub__convolution3PKfPfii # -- Begin function _Z27__device_stub__convolution3PKfPfii .p2align 4, 0x90 .type _Z27__device_stub__convolution3PKfPfii,@function _Z27__device_stub__convolution3PKfPfii: # @_Z27__device_stub__convolution3PKfPfii .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%rsp) movl %ecx, 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 $_Z12convolution3PKfPfii, %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_end2: .size _Z27__device_stub__convolution3PKfPfii, .Lfunc_end2-_Z27__device_stub__convolution3PKfPfii .cfi_endproc # -- End function .section .rodata.cst8,"aM",@progbits,8 .p2align 3, 0x0 # -- Begin function main .LCPI3_0: .quad 0x412e848000000000 # double 1.0E+6 .LCPI3_3: .quad 0x41cdcd6500000000 # double 1.0E+9 .section .rodata.cst16,"aM",@progbits,16 .p2align 4, 0x0 .LCPI3_1: .long 1127219200 # 0x43300000 .long 1160773632 # 0x45300000 .long 0 # 0x0 .long 0 # 0x0 .LCPI3_2: .quad 0x4330000000000000 # double 4503599627370496 .quad 0x4530000000000000 # double 1.9342813113834067E+25 .section .rodata.cst4,"aM",@progbits,4 .p2align 2, 0x0 .LCPI3_4: .long 0x3f800000 # float 1 .LCPI3_5: .long 0x3c800000 # float 0.015625 .text .globl main .p2align 4, 0x90 .type main,@function main: # @main .Lfunc_begin0: .cfi_startproc .cfi_personality 3, __gxx_personality_v0 .cfi_lsda 3, .Lexception0 # %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 $1608, %rsp # imm = 0x648 .cfi_def_cfa_offset 1664 .cfi_offset %rbx, -56 .cfi_offset %r12, -48 .cfi_offset %r13, -40 .cfi_offset %r14, -32 .cfi_offset %r15, -24 .cfi_offset %rbp, -16 movq %rsi, %r12 .cfi_escape 0x2e, 0x00 leaq 136(%rsp), %rdi xorl %esi, %esi callq hipGetDevicePropertiesR0600 .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str, %esi movl $45, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB3_119 # %bb.1: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i cmpb $0, 56(%r14) je .LBB3_3 # %bb.2: movzbl 67(%r14), %eax jmp .LBB3_4 .LBB3_3: .cfi_escape 0x2e, 0x00 movq %r14, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax .cfi_escape 0x2e, 0x00 movq %r14, %rdi movl $10, %esi callq *48(%rax) .LBB3_4: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit .cfi_escape 0x2e, 0x00 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.1, %esi movl $45, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB3_120 # %bb.5: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i91 cmpb $0, 56(%r14) je .LBB3_7 # %bb.6: movzbl 67(%r14), %eax jmp .LBB3_8 .LBB3_7: .cfi_escape 0x2e, 0x00 movq %r14, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax .cfi_escape 0x2e, 0x00 movq %r14, %rdi movl $10, %esi callq *48(%rax) .LBB3_8: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit94 .cfi_escape 0x2e, 0x00 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str, %esi movl $45, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB3_121 # %bb.9: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i96 cmpb $0, 56(%r14) je .LBB3_11 # %bb.10: movzbl 67(%r14), %eax jmp .LBB3_12 .LBB3_11: .cfi_escape 0x2e, 0x00 movq %r14, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax .cfi_escape 0x2e, 0x00 movq %r14, %rdi movl $10, %esi callq *48(%rax) .LBB3_12: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit99 .cfi_escape 0x2e, 0x00 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.2, %esi movl $13, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l .cfi_escape 0x2e, 0x00 leaq 136(%rsp), %r14 movq %r14, %rdi callq strlen .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movq %r14, %rsi movq %rax, %rdx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB3_122 # %bb.13: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i101 cmpb $0, 56(%r14) je .LBB3_15 # %bb.14: movzbl 67(%r14), %eax jmp .LBB3_16 .LBB3_15: .cfi_escape 0x2e, 0x00 movq %r14, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax .cfi_escape 0x2e, 0x00 movq %r14, %rdi movl $10, %esi callq *48(%rax) .LBB3_16: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit104 .cfi_escape 0x2e, 0x00 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.3, %esi movl $19, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l cvtsi2sdl 744(%rsp), %xmm0 divsd .LCPI3_0(%rip), %xmm0 .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi callq _ZNSo9_M_insertIdEERSoT_ movq %rax, %r14 .cfi_escape 0x2e, 0x00 movl $.L.str.4, %esi movl $4, %edx movq %rax, %rdi callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq (%r14), %rax movq -24(%rax), %rax movq 240(%r14,%rax), %r15 testq %r15, %r15 je .LBB3_123 # %bb.17: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i106 cmpb $0, 56(%r15) je .LBB3_19 # %bb.18: movzbl 67(%r15), %eax jmp .LBB3_20 .LBB3_19: .cfi_escape 0x2e, 0x00 movq %r15, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r15), %rax .cfi_escape 0x2e, 0x00 movq %r15, %rdi movl $10, %esi callq *48(%rax) .LBB3_20: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit109 .cfi_escape 0x2e, 0x00 movsbl %al, %esi movq %r14, %rdi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.5, %esi movl $18, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l cvtsi2sdl 744(%rsp), %xmm1 movl 748(%rsp), %eax leal 7(%rax), %ecx testl %eax, %eax cmovnsl %eax, %ecx sarl $3, %ecx xorps %xmm0, %xmm0 cvtsi2sd %ecx, %xmm0 addsd %xmm1, %xmm1 mulsd %xmm1, %xmm0 divsd .LCPI3_0(%rip), %xmm0 .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi callq _ZNSo9_M_insertIdEERSoT_ movq %rax, %r14 .cfi_escape 0x2e, 0x00 movl $.L.str.6, %esi movl $5, %edx movq %rax, %rdi callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq (%r14), %rax movq -24(%rax), %rax movq 240(%r14,%rax), %r15 testq %r15, %r15 je .LBB3_124 # %bb.21: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i111 cmpb $0, 56(%r15) je .LBB3_23 # %bb.22: movzbl 67(%r15), %eax jmp .LBB3_24 .LBB3_23: .cfi_escape 0x2e, 0x00 movq %r15, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r15), %rax .cfi_escape 0x2e, 0x00 movq %r15, %rdi movl $10, %esi callq *48(%rax) .LBB3_24: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit114 .cfi_escape 0x2e, 0x00 movsbl %al, %esi movq %r14, %rdi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.7, %esi movl $14, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl 524(%rsp), %esi .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi callq _ZNSolsEi movq (%rax), %rcx movq -24(%rcx), %rcx movq 240(%rax,%rcx), %r14 testq %r14, %r14 je .LBB3_125 # %bb.25: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i116 cmpb $0, 56(%r14) je .LBB3_27 # %bb.26: movzbl 67(%r14), %ecx jmp .LBB3_28 .LBB3_27: .cfi_escape 0x2e, 0x00 movq %r14, %rdi movq %rax, %rbx callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax .cfi_escape 0x2e, 0x00 movq %r14, %rdi movl $10, %esi callq *48(%rax) movl %eax, %ecx movq %rbx, %rax .LBB3_28: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit119 .cfi_escape 0x2e, 0x00 movsbl %cl, %esi movq %rax, %rdi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.8, %esi movl $20, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl 760(%rsp), %esi .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi callq _ZNSolsEi movq (%rax), %rcx movq -24(%rcx), %rcx movq 240(%rax,%rcx), %r14 testq %r14, %r14 je .LBB3_126 # %bb.29: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i121 cmpb $0, 56(%r14) je .LBB3_31 # %bb.30: movzbl 67(%r14), %ecx jmp .LBB3_32 .LBB3_31: .cfi_escape 0x2e, 0x00 movq %r14, %rdi movq %rax, %rbx callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax .cfi_escape 0x2e, 0x00 movq %r14, %rdi movl $10, %esi callq *48(%rax) movl %eax, %ecx movq %rbx, %rax .LBB3_32: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit124 .cfi_escape 0x2e, 0x00 movsbl %cl, %esi movq %rax, %rdi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.9, %esi movl $21, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movl 440(%rsp), %esi .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi callq _ZNSolsEi movq (%rax), %rcx movq -24(%rcx), %rcx movq 240(%rax,%rcx), %r14 testq %r14, %r14 je .LBB3_127 # %bb.33: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i126 cmpb $0, 56(%r14) je .LBB3_35 # %bb.34: movzbl 67(%r14), %ecx jmp .LBB3_36 .LBB3_35: .cfi_escape 0x2e, 0x00 movq %r14, %rdi movq %rax, %rbx callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax .cfi_escape 0x2e, 0x00 movq %r14, %rdi movl $10, %esi callq *48(%rax) movl %eax, %ecx movq %rbx, %rax .LBB3_36: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit129 .cfi_escape 0x2e, 0x00 movsbl %cl, %esi movq %rax, %rdi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.10, %esi movl $25, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq 432(%rsp), %rsi .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi callq _ZNSo9_M_insertImEERSoT_ movq %rax, %r14 .cfi_escape 0x2e, 0x00 movl $.L.str.11, %esi movl $2, %edx movq %rax, %rdi callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq (%r14), %rax movq -24(%rax), %rax movq 240(%r14,%rax), %r15 testq %r15, %r15 je .LBB3_128 # %bb.37: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i131 cmpb $0, 56(%r15) je .LBB3_39 # %bb.38: movzbl 67(%r15), %eax jmp .LBB3_40 .LBB3_39: .cfi_escape 0x2e, 0x00 movq %r15, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r15), %rax .cfi_escape 0x2e, 0x00 movq %r15, %rdi movl $10, %esi callq *48(%rax) .LBB3_40: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit134 .cfi_escape 0x2e, 0x00 movsbl %al, %esi movq %r14, %rdi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.12, %esi movl $31, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movsd 424(%rsp), %xmm1 # xmm1 = mem[0],zero unpcklps .LCPI3_1(%rip), %xmm1 # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1] subpd .LCPI3_2(%rip), %xmm1 movapd %xmm1, %xmm0 unpckhpd %xmm1, %xmm0 # xmm0 = xmm0[1],xmm1[1] addsd %xmm1, %xmm0 divsd .LCPI3_3(%rip), %xmm0 .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi callq _ZNSo9_M_insertIdEERSoT_ movq %rax, %r14 .cfi_escape 0x2e, 0x00 movl $.L.str.13, %esi movl $3, %edx movq %rax, %rdi callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l movq (%r14), %rax movq -24(%rax), %rax movq 240(%r14,%rax), %r15 testq %r15, %r15 je .LBB3_129 # %bb.41: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i136 cmpb $0, 56(%r15) je .LBB3_43 # %bb.42: movzbl 67(%r15), %eax jmp .LBB3_44 .LBB3_43: .cfi_escape 0x2e, 0x00 movq %r15, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r15), %rax .cfi_escape 0x2e, 0x00 movq %r15, %rdi movl $10, %esi callq *48(%rax) .LBB3_44: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit139 .cfi_escape 0x2e, 0x00 movsbl %al, %esi movq %r14, %rdi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r14 testq %r14, %r14 je .LBB3_130 # %bb.45: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i141 cmpb $0, 56(%r14) je .LBB3_47 # %bb.46: movzbl 67(%r14), %eax jmp .LBB3_48 .LBB3_47: .cfi_escape 0x2e, 0x00 movq %r14, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv movq (%r14), %rax .cfi_escape 0x2e, 0x00 movq %r14, %rdi movl $10, %esi callq *48(%rax) .LBB3_48: # %_ZSt4endlIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_.exit144 .cfi_escape 0x2e, 0x00 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv movq 8(%r12), %rdi .cfi_escape 0x2e, 0x00 xorl %esi, %esi movl $10, %edx callq __isoc23_strtol movq %rax, %r13 movslq %r13d, %r15 movq %r15, %rax shrq $61, %rax jne .LBB3_131 # %bb.49: # %_ZNSt6vectorIfSaIfEE17_S_check_init_lenEmRKS0_.exit.i movq %r13, %rbp shlq $32, %rbp je .LBB3_50 # %bb.51: # %_ZNSt16allocator_traitsISaIfEE8allocateERS0_m.exit.i.i.i.i movq %rbp, %rdi sarq $30, %rdi .cfi_escape 0x2e, 0x00 callq _Znwm movq %rax, %r14 jmp .LBB3_52 .LBB3_50: xorl %r14d, %r14d .LBB3_52: # %_ZNSt12_Vector_baseIfSaIfEEC2EmRKS0_.exit.i movq %r14, %rax testq %rbp, %rbp je .LBB3_55 # %bb.53: # %.lr.ph.i.i.i.i.i.i.i.i.i.preheader leaq (%r14,%r15,4), %rax shlq $2, %r15 xorl %ecx, %ecx .p2align 4, 0x90 .LBB3_54: # %.lr.ph.i.i.i.i.i.i.i.i.i # =>This Inner Loop Header: Depth=1 movl $1065353216, (%r14,%rcx) # imm = 0x3F800000 addq $4, %rcx cmpq %rcx, %r15 jne .LBB3_54 .LBB3_55: # %_ZNSt6vectorIfSaIfEEC2EmRKfRKS0_.exit movq %r12, %r15 cmpq %rax, %r14 je .LBB3_58 # %bb.56: # %.lr.ph.i.preheader xorpd %xmm0, %xmm0 movss .LCPI3_4(%rip), %xmm1 # xmm1 = mem[0],zero,zero,zero movq %r14, %rcx .p2align 4, 0x90 .LBB3_57: # %.lr.ph.i # =>This Inner Loop Header: Depth=1 movss %xmm0, (%rcx) addss %xmm1, %xmm0 addq $4, %rcx cmpq %rax, %rcx jne .LBB3_57 .LBB3_58: # %_ZSt4iotaIN9__gnu_cxx17__normal_iteratorIPfSt6vectorIfSaIfEEEEfEvT_S7_T0_.exit movq %rbp, %r12 sarq $30, %r12 .Ltmp0: .cfi_escape 0x2e, 0x00 leaq 16(%rsp), %rdi movq %r12, %rsi callq hipMalloc .Ltmp1: # %bb.59: movq 16(%rsp), %rdi .Ltmp2: .cfi_escape 0x2e, 0x00 movq %r14, %rsi movq %r12, %rdx movl $1, %ecx callq hipMemcpy .Ltmp3: # %bb.60: .Ltmp5: .cfi_escape 0x2e, 0x00 movl $20, %edi callq _Znwm .Ltmp6: # %bb.61: # %_ZNSt6vectorIfSaIfEEC2ESt16initializer_listIfERKS0_.exit movq %rax, %rbx movups .Lconstinit(%rip), %xmm0 movups %xmm0, (%rax) movl $1065353216, 16(%rax) # imm = 0x3F800000 .Ltmp8: .cfi_escape 0x2e, 0x00 movl $c_M, %edi movl $20, %edx movq %rax, %rsi xorl %ecx, %ecx movl $1, %r8d movq %rax, 88(%rsp) # 8-byte Spill callq hipMemcpyToSymbol .Ltmp9: # %bb.62: # %_Z17hipMemcpyToSymbolIA5_fE10hipError_tRKT_PKvmm13hipMemcpyKind.exit .Ltmp11: .cfi_escape 0x2e, 0x00 leaq 8(%rsp), %rdi movq %r12, %rsi callq hipMalloc .Ltmp12: # %bb.63: xorps %xmm0, %xmm0 cvtsi2ss %r13d, %xmm0 mulss .LCPI3_5(%rip), %xmm0 .cfi_escape 0x2e, 0x00 callq ceilf@PLT cvttss2si %xmm0, %ebx movq 16(%r15), %rdi .cfi_escape 0x2e, 0x00 xorl %esi, %esi movl $10, %edx callq __isoc23_strtol cmpl $2, %eax je .LBB3_73 # %bb.64: cmpl $1, %eax jne .LBB3_77 # %bb.65: movl %ebx, %eax movabsq $4294967296, %rdi # imm = 0x100000000 orq %rax, %rdi .Ltmp20: .cfi_escape 0x2e, 0x00 movabsq $4294967360, %rdx # imm = 0x100000040 movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration .Ltmp21: # %bb.66: testl %eax, %eax jne .LBB3_87 # %bb.67: movq 16(%rsp), %rax movq 8(%rsp), %rcx movq %rax, 80(%rsp) movq %rcx, 72(%rsp) movl $5, 4(%rsp) movl %r13d, (%rsp) leaq 80(%rsp), %rax movq %rax, 96(%rsp) leaq 72(%rsp), %rax movq %rax, 104(%rsp) leaq 4(%rsp), %rax movq %rax, 112(%rsp) movq %rsp, %rax movq %rax, 120(%rsp) .Ltmp22: .cfi_escape 0x2e, 0x00 leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration .Ltmp23: # %bb.68: # %.noexc51 movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d .Ltmp24: .cfi_escape 0x2e, 0x10 leaq 96(%rsp), %r9 movl $_Z12convolution2PKfPfii, %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 .Ltmp25: jmp .LBB3_87 .LBB3_73: movl %ebx, %eax movabsq $4294967296, %rdi # imm = 0x100000000 orq %rax, %rdi .Ltmp14: .cfi_escape 0x2e, 0x00 movabsq $4294967360, %rdx # imm = 0x100000040 movl $1, %esi movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration .Ltmp15: # %bb.74: testl %eax, %eax jne .LBB3_87 # %bb.75: movq 16(%rsp), %rax movq 8(%rsp), %rcx movq %rax, 80(%rsp) movq %rcx, 72(%rsp) movl $5, 4(%rsp) movl %r13d, (%rsp) leaq 80(%rsp), %rax movq %rax, 96(%rsp) leaq 72(%rsp), %rax movq %rax, 104(%rsp) leaq 4(%rsp), %rax movq %rax, 112(%rsp) movq %rsp, %rax movq %rax, 120(%rsp) .Ltmp16: .cfi_escape 0x2e, 0x00 leaq 56(%rsp), %rdi leaq 40(%rsp), %rsi leaq 32(%rsp), %rdx leaq 24(%rsp), %rcx callq __hipPopCallConfiguration .Ltmp17: # %bb.76: # %.noexc59 movq 56(%rsp), %rsi movl 64(%rsp), %edx movq 40(%rsp), %rcx movl 48(%rsp), %r8d .Ltmp18: .cfi_escape 0x2e, 0x10 leaq 96(%rsp), %r9 movl $_Z12convolution3PKfPfii, %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 .Ltmp19: jmp .LBB3_87 .LBB3_77: .Ltmp26: .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.14, %esi movl $13, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l .Ltmp27: # %bb.78: # %_ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc.exit movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %rbx testq %rbx, %rbx je .LBB3_79 # %bb.81: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i146 cmpb $0, 56(%rbx) je .LBB3_83 # %bb.82: movzbl 67(%rbx), %eax jmp .LBB3_85 .LBB3_83: .Ltmp28: .cfi_escape 0x2e, 0x00 movq %rbx, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv .Ltmp29: # %bb.84: # %.noexc150 movq (%rbx), %rax .Ltmp30: .cfi_escape 0x2e, 0x00 movq %rbx, %rdi movl $10, %esi callq *48(%rax) .Ltmp31: .LBB3_85: # %_ZNKSt9basic_iosIcSt11char_traitsIcEE5widenEc.exit.i .Ltmp32: .cfi_escape 0x2e, 0x00 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc .Ltmp33: # %bb.86: # %.noexc152 .Ltmp34: .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .Ltmp35: .LBB3_87: # %_ZNSt6vectorIfSaIfEE17_S_check_init_lenEmRKS0_.exit.i63 testq %rbp, %rbp je .LBB3_88 # %bb.89: # %_ZNSt16allocator_traitsISaIfEE8allocateERS0_m.exit.i.i.i.i65 .Ltmp36: .cfi_escape 0x2e, 0x00 movq %r12, %rdi callq _Znwm .Ltmp37: # %bb.90: movq %rax, %rbx testq %rbp, %rbp jne .LBB3_92 jmp .LBB3_94 .LBB3_88: xorl %ebx, %ebx testq %rbp, %rbp je .LBB3_94 .LBB3_92: movl $0, (%rbx) movabsq $4294967296, %rax # imm = 0x100000000 cmpq %rax, %rbp je .LBB3_94 # %bb.93: # %_ZSt6fill_nIPfmfET_S1_T0_RKT1_.exit.loopexit.i.i.i.i.i leaq 4(%rbx), %rdi leaq -4(%r12), %rdx .cfi_escape 0x2e, 0x00 xorl %esi, %esi callq memset@PLT .LBB3_94: # %_ZNSt6vectorIfSaIfEEC2EmRKS0_.exit movq 8(%rsp), %rsi .Ltmp39: .cfi_escape 0x2e, 0x00 movq %rbx, %rdi movq %r12, %rdx movl $2, %ecx callq hipMemcpy .Ltmp40: # %bb.95: movq 16(%rsp), %rdi .Ltmp41: .cfi_escape 0x2e, 0x00 callq hipFree .Ltmp42: # %bb.96: movq 8(%rsp), %rdi .Ltmp43: .cfi_escape 0x2e, 0x00 callq hipFree .Ltmp44: # %bb.97: .Ltmp45: .cfi_escape 0x2e, 0x00 movl $_ZSt4cout, %edi movl $.L.str.15, %esi movl $10, %edx callq _ZSt16__ostream_insertIcSt11char_traitsIcEERSt13basic_ostreamIT_T0_ES6_PKS3_l .Ltmp46: # %bb.98: # %_ZStlsISt11char_traitsIcEERSt13basic_ostreamIcT_ES5_PKc.exit70 movq _ZSt4cout(%rip), %rax movq -24(%rax), %rax movq _ZSt4cout+240(%rax), %r12 testq %r12, %r12 je .LBB3_99 # %bb.101: # %_ZSt13__check_facetISt5ctypeIcEERKT_PS3_.exit.i.i156 cmpb $0, 56(%r12) je .LBB3_103 # %bb.102: movzbl 67(%r12), %eax jmp .LBB3_105 .LBB3_103: .Ltmp47: .cfi_escape 0x2e, 0x00 movq %r12, %rdi callq _ZNKSt5ctypeIcE13_M_widen_initEv .Ltmp48: # %bb.104: # %.noexc161 movq (%r12), %rax .Ltmp49: .cfi_escape 0x2e, 0x00 movq %r12, %rdi movl $10, %esi callq *48(%rax) .Ltmp50: .LBB3_105: # %_ZNKSt9basic_iosIcSt11char_traitsIcEE5widenEc.exit.i158 .Ltmp51: .cfi_escape 0x2e, 0x00 movsbl %al, %esi movl $_ZSt4cout, %edi callq _ZNSo3putEc .Ltmp52: # %bb.106: # %.noexc163 .Ltmp53: .cfi_escape 0x2e, 0x00 movq %rax, %rdi callq _ZNSo5flushEv .Ltmp54: # %bb.107: # %_ZNSolsEPFRSoS_E.exit72 testq %rbx, %rbx je .LBB3_109 # %bb.108: .cfi_escape 0x2e, 0x00 movq %rbx, %rdi callq _ZdlPv .LBB3_109: # %_ZNSt6vectorIfSaIfEED2Ev.exit .cfi_escape 0x2e, 0x00 movq 88(%rsp), %rdi # 8-byte Reload callq _ZdlPv testq %r14, %r14 je .LBB3_111 # %bb.110: .cfi_escape 0x2e, 0x00 movq %r14, %rdi callq _ZdlPv .LBB3_111: # %_ZNSt6vectorIfSaIfEED2Ev.exit79 xorl %eax, %eax addq $1608, %rsp # imm = 0x648 .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 .LBB3_119: .cfi_def_cfa_offset 1664 .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_120: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_121: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_122: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_123: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_124: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_125: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_126: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_127: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_128: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_129: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_130: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .LBB3_131: # %.noexc .cfi_escape 0x2e, 0x00 movl $.L.str.16, %edi callq _ZSt20__throw_length_errorPKc .LBB3_99: .Ltmp55: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .Ltmp56: # %bb.100: # %.noexc160 .LBB3_79: .Ltmp58: .cfi_escape 0x2e, 0x00 callq _ZSt16__throw_bad_castv .Ltmp59: # %bb.80: # %.noexc149 .LBB3_112: .Ltmp38: movq %rax, %r12 jmp .LBB3_115 .LBB3_71: .Ltmp13: movq %rax, %r12 jmp .LBB3_115 .LBB3_70: .Ltmp10: movq %rax, %r12 jmp .LBB3_115 .LBB3_132: # %_ZNSt12_Vector_baseIfSaIfEED2Ev.exit.i .Ltmp7: movq %rax, %r12 jmp .LBB3_116 .LBB3_69: .Ltmp4: movq %rax, %r12 jmp .LBB3_116 .LBB3_72: .Ltmp60: movq %rax, %r12 jmp .LBB3_115 .LBB3_113: .Ltmp57: movq %rax, %r12 testq %rbx, %rbx je .LBB3_115 # %bb.114: .cfi_escape 0x2e, 0x00 movq %rbx, %rdi callq _ZdlPv .LBB3_115: # %_ZNSt6vectorIfSaIfEED2Ev.exit85 .cfi_escape 0x2e, 0x00 movq 88(%rsp), %rdi # 8-byte Reload callq _ZdlPv .LBB3_116: testq %r14, %r14 je .LBB3_118 # %bb.117: .cfi_escape 0x2e, 0x00 movq %r14, %rdi callq _ZdlPv .LBB3_118: # %_ZNSt6vectorIfSaIfEED2Ev.exit88 .cfi_escape 0x2e, 0x00 movq %r12, %rdi callq _Unwind_Resume@PLT .Lfunc_end3: .size main, .Lfunc_end3-main .cfi_endproc .section .gcc_except_table,"a",@progbits .p2align 2, 0x0 GCC_except_table3: .Lexception0: .byte 255 # @LPStart Encoding = omit .byte 255 # @TType Encoding = omit .byte 1 # Call site Encoding = uleb128 .uleb128 .Lcst_end0-.Lcst_begin0 .Lcst_begin0: .uleb128 .Lfunc_begin0-.Lfunc_begin0 # >> Call Site 1 << .uleb128 .Ltmp0-.Lfunc_begin0 # Call between .Lfunc_begin0 and .Ltmp0 .byte 0 # has no landing pad .byte 0 # On action: cleanup .uleb128 .Ltmp0-.Lfunc_begin0 # >> Call Site 2 << .uleb128 .Ltmp3-.Ltmp0 # Call between .Ltmp0 and .Ltmp3 .uleb128 .Ltmp4-.Lfunc_begin0 # jumps to .Ltmp4 .byte 0 # On action: cleanup .uleb128 .Ltmp5-.Lfunc_begin0 # >> Call Site 3 << .uleb128 .Ltmp6-.Ltmp5 # Call between .Ltmp5 and .Ltmp6 .uleb128 .Ltmp7-.Lfunc_begin0 # jumps to .Ltmp7 .byte 0 # On action: cleanup .uleb128 .Ltmp8-.Lfunc_begin0 # >> Call Site 4 << .uleb128 .Ltmp9-.Ltmp8 # Call between .Ltmp8 and .Ltmp9 .uleb128 .Ltmp10-.Lfunc_begin0 # jumps to .Ltmp10 .byte 0 # On action: cleanup .uleb128 .Ltmp11-.Lfunc_begin0 # >> Call Site 5 << .uleb128 .Ltmp12-.Ltmp11 # Call between .Ltmp11 and .Ltmp12 .uleb128 .Ltmp13-.Lfunc_begin0 # jumps to .Ltmp13 .byte 0 # On action: cleanup .uleb128 .Ltmp12-.Lfunc_begin0 # >> Call Site 6 << .uleb128 .Ltmp20-.Ltmp12 # Call between .Ltmp12 and .Ltmp20 .byte 0 # has no landing pad .byte 0 # On action: cleanup .uleb128 .Ltmp20-.Lfunc_begin0 # >> Call Site 7 << .uleb128 .Ltmp35-.Ltmp20 # Call between .Ltmp20 and .Ltmp35 .uleb128 .Ltmp60-.Lfunc_begin0 # jumps to .Ltmp60 .byte 0 # On action: cleanup .uleb128 .Ltmp36-.Lfunc_begin0 # >> Call Site 8 << .uleb128 .Ltmp37-.Ltmp36 # Call between .Ltmp36 and .Ltmp37 .uleb128 .Ltmp38-.Lfunc_begin0 # jumps to .Ltmp38 .byte 0 # On action: cleanup .uleb128 .Ltmp37-.Lfunc_begin0 # >> Call Site 9 << .uleb128 .Ltmp39-.Ltmp37 # Call between .Ltmp37 and .Ltmp39 .byte 0 # has no landing pad .byte 0 # On action: cleanup .uleb128 .Ltmp39-.Lfunc_begin0 # >> Call Site 10 << .uleb128 .Ltmp54-.Ltmp39 # Call between .Ltmp39 and .Ltmp54 .uleb128 .Ltmp57-.Lfunc_begin0 # jumps to .Ltmp57 .byte 0 # On action: cleanup .uleb128 .Ltmp54-.Lfunc_begin0 # >> Call Site 11 << .uleb128 .Ltmp55-.Ltmp54 # Call between .Ltmp54 and .Ltmp55 .byte 0 # has no landing pad .byte 0 # On action: cleanup .uleb128 .Ltmp55-.Lfunc_begin0 # >> Call Site 12 << .uleb128 .Ltmp56-.Ltmp55 # Call between .Ltmp55 and .Ltmp56 .uleb128 .Ltmp57-.Lfunc_begin0 # jumps to .Ltmp57 .byte 0 # On action: cleanup .uleb128 .Ltmp58-.Lfunc_begin0 # >> Call Site 13 << .uleb128 .Ltmp59-.Ltmp58 # Call between .Ltmp58 and .Ltmp59 .uleb128 .Ltmp60-.Lfunc_begin0 # jumps to .Ltmp60 .byte 0 # On action: cleanup .uleb128 .Ltmp59-.Lfunc_begin0 # >> Call Site 14 << .uleb128 .Lfunc_end3-.Ltmp59 # Call between .Ltmp59 and .Lfunc_end3 .byte 0 # has no landing pad .byte 0 # On action: cleanup .Lcst_end0: .p2align 2, 0x0 # -- End function .text .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 $_Z12convolution1PKfS0_Pfii, %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 $_Z12convolution2PKfPfii, %esi movl $.L__unnamed_2, %edx movl $.L__unnamed_2, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction xorps %xmm0, %xmm0 movups %xmm0, 16(%rsp) movups %xmm0, (%rsp) movl $_Z12convolution3PKfPfii, %esi movl $.L__unnamed_3, %edx movl $.L__unnamed_3, %ecx movq %rbx, %rdi movl $-1, %r8d xorl %r9d, %r9d callq __hipRegisterFunction movl $0, 8(%rsp) movl $1, (%rsp) movl $c_M, %esi movl $.L__unnamed_4, %edx movl $.L__unnamed_4, %ecx movl $20, %r9d movq %rbx, %rdi xorl %r8d, %r8d callq __hipRegisterVar 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 c_M,@object # @c_M .local c_M .comm c_M,20,16 .type _Z12convolution1PKfS0_Pfii,@object # @_Z12convolution1PKfS0_Pfii .section .rodata,"a",@progbits .globl _Z12convolution1PKfS0_Pfii .p2align 3, 0x0 _Z12convolution1PKfS0_Pfii: .quad _Z27__device_stub__convolution1PKfS0_Pfii .size _Z12convolution1PKfS0_Pfii, 8 .type _Z12convolution2PKfPfii,@object # @_Z12convolution2PKfPfii .globl _Z12convolution2PKfPfii .p2align 3, 0x0 _Z12convolution2PKfPfii: .quad _Z27__device_stub__convolution2PKfPfii .size _Z12convolution2PKfPfii, 8 .type _Z12convolution3PKfPfii,@object # @_Z12convolution3PKfPfii .globl _Z12convolution3PKfPfii .p2align 3, 0x0 _Z12convolution3PKfPfii: .quad _Z27__device_stub__convolution3PKfPfii .size _Z12convolution3PKfPfii, 8 .type .L.str,@object # @.str .section .rodata.str1.1,"aMS",@progbits,1 .L.str: .asciz "---------------------------------------------" .size .L.str, 46 .type .L.str.1,@object # @.str.1 .L.str.1: .asciz " GPU PROPERTIES " .size .L.str.1, 46 .type .L.str.2,@object # @.str.2 .L.str.2: .asciz "Device Name: " .size .L.str.2, 14 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz "Memory Clock Rate: " .size .L.str.3, 20 .type .L.str.4,@object # @.str.4 .L.str.4: .asciz " GHz" .size .L.str.4, 5 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "Memory Bandwidth: " .size .L.str.5, 19 .type .L.str.6,@object # @.str.6 .L.str.6: .asciz " GB/s" .size .L.str.6, 6 .type .L.str.7,@object # @.str.7 .L.str.7: .asciz "Number of SM: " .size .L.str.7, 15 .type .L.str.8,@object # @.str.8 .L.str.8: .asciz "Max Threads per SM: " .size .L.str.8, 21 .type .L.str.9,@object # @.str.9 .L.str.9: .asciz "Registers per Block: " .size .L.str.9, 22 .type .L.str.10,@object # @.str.10 .L.str.10: .asciz "Shared Memory per Block: " .size .L.str.10, 26 .type .L.str.11,@object # @.str.11 .L.str.11: .asciz " B" .size .L.str.11, 3 .type .L.str.12,@object # @.str.12 .L.str.12: .asciz "Total Global Memory per Block: " .size .L.str.12, 32 .type .L.str.13,@object # @.str.13 .L.str.13: .asciz " GB" .size .L.str.13, 4 .type .Lconstinit,@object # @constinit .section .rodata,"a",@progbits .p2align 2, 0x0 .Lconstinit: .long 0x3f800000 # float 1 .long 0x3f800000 # float 1 .long 0x40000000 # float 2 .long 0x3f800000 # float 1 .long 0x3f800000 # float 1 .size .Lconstinit, 20 .type .L.str.14,@object # @.str.14 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.14: .asciz "Wrong inputs!" .size .L.str.14, 14 .type .L.str.15,@object # @.str.15 .L.str.15: .asciz "Closing..." .size .L.str.15, 11 .type .L.str.16,@object # @.str.16 .L.str.16: .asciz "cannot create std::vector larger than max_size()" .size .L.str.16, 49 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z12convolution1PKfS0_Pfii" .size .L__unnamed_1, 27 .type .L__unnamed_2,@object # @1 .L__unnamed_2: .asciz "_Z12convolution2PKfPfii" .size .L__unnamed_2, 24 .type .L__unnamed_3,@object # @2 .L__unnamed_3: .asciz "_Z12convolution3PKfPfii" .size .L__unnamed_3, 24 .type .L__unnamed_4,@object # @3 .L__unnamed_4: .asciz "c_M" .size .L__unnamed_4, 4 .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 _Z27__device_stub__convolution1PKfS0_Pfii .addrsig_sym _Z27__device_stub__convolution2PKfPfii .addrsig_sym _Z27__device_stub__convolution3PKfPfii .addrsig_sym __gxx_personality_v0 .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Unwind_Resume .addrsig_sym c_M .addrsig_sym _Z12convolution1PKfS0_Pfii .addrsig_sym _Z12convolution2PKfPfii .addrsig_sym _Z12convolution3PKfPfii .addrsig_sym _ZSt4cout .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 g_One_Bgrad(float* _delta, float* bgrad, int rows, int cols, int channels) { extern __shared__ float _sum[]; int channel = blockIdx.x; int col = blockIdx.y; int row = threadIdx.x; float delta = _delta[channel * rows * cols + row * cols + col]; _sum[row] = delta; __syncthreads(); int len = rows; while(len != 1) { __syncthreads(); int skip = (len + 1) >> 1; if(threadIdx.x < (len >> 1)) { _sum[threadIdx.x] += _sum[threadIdx.x + skip]; } len = (len + 1) >> 1; } __syncthreads(); if(threadIdx.x == 0) { bgrad[channel * cols + col] = _sum[0] / rows; } }
code for sm_80 Function : _Z11g_One_BgradPfS_iii .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 R3, SR_CTAID.X ; /* 0x0000000000037919 */ /* 0x000e220000002500 */ /*0020*/ IMAD.MOV.U32 R4, RZ, RZ, 0x4 ; /* 0x00000004ff047424 */ /* 0x000fe200078e00ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe40000000a00 */ /*0040*/ S2R R7, SR_TID.X ; /* 0x0000000000077919 */ /* 0x000e280000002100 */ /*0050*/ S2R R0, SR_CTAID.Y ; /* 0x0000000000007919 */ /* 0x000e620000002600 */ /*0060*/ IMAD R5, R3, c[0x0][0x170], R7 ; /* 0x00005c0003057a24 */ /* 0x001fc800078e0207 */ /*0070*/ IMAD R5, R5, c[0x0][0x174], R0 ; /* 0x00005d0005057a24 */ /* 0x002fc800078e0200 */ /*0080*/ IMAD.WIDE R4, R5, R4, c[0x0][0x160] ; /* 0x0000580005047625 */ /* 0x000fcc00078e0204 */ /*0090*/ LDG.E R4, [R4.64] ; /* 0x0000000404047981 */ /* 0x000ea2000c1e1900 */ /*00a0*/ IMAD.MOV.U32 R2, RZ, RZ, c[0x0][0x170] ; /* 0x00005c00ff027624 */ /* 0x000fe200078e00ff */ /*00b0*/ ISETP.NE.AND P0, PT, R7, RZ, PT ; /* 0x000000ff0700720c */ /* 0x000fc80003f05270 */ /*00c0*/ ISETP.NE.AND P1, PT, R2, 0x1, PT ; /* 0x000000010200780c */ /* 0x000fe20003f25270 */ /*00d0*/ STS [R7.X4], R4 ; /* 0x0000000407007388 */ /* 0x0041e80000004800 */ /*00e0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000ff00000010000 */ /*00f0*/ @!P1 BRA 0x200 ; /* 0x0000010000009947 */ /* 0x000fea0003800000 */ /*0100*/ IMAD.SHL.U32 R2, R7, 0x4, RZ ; /* 0x0000000407027824 */ /* 0x001fe400078e00ff */ /*0110*/ IMAD.MOV.U32 R4, RZ, RZ, c[0x0][0x170] ; /* 0x00005c00ff047624 */ /* 0x000fca00078e00ff */ /*0120*/ SHF.R.S32.HI R5, RZ, 0x1, R4 ; /* 0x00000001ff057819 */ /* 0x000fe20000011404 */ /*0130*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe60000010000 */ /*0140*/ ISETP.GE.U32.AND P1, PT, R7, R5, PT ; /* 0x000000050700720c */ /* 0x000fda0003f26070 */ /*0150*/ @!P1 LEA R5, R4.reuse, 0x2, 0x1 ; /* 0x0000000204059811 */ /* 0x040fe400078e08ff */ /*0160*/ IADD3 R4, R4, 0x1, RZ ; /* 0x0000000104047810 */ /* 0x000fe40007ffe0ff */ /*0170*/ @!P1 LOP3.LUT R5, R5, 0xfffffffc, RZ, 0xc0, !PT ; /* 0xfffffffc05059812 */ /* 0x000fe400078ec0ff */ /*0180*/ SHF.R.S32.HI R4, RZ, 0x1, R4 ; /* 0x00000001ff047819 */ /* 0x000fe20000011404 */ /*0190*/ @!P1 LDS R6, [R7.X4] ; /* 0x0000000007069984 */ /* 0x000fe40000004800 */ /*01a0*/ @!P1 IMAD.IADD R5, R2, 0x1, R5 ; /* 0x0000000102059824 */ /* 0x000fcc00078e0205 */ /*01b0*/ @!P1 LDS R5, [R5] ; /* 0x0000000005059984 */ /* 0x000e240000000800 */ /*01c0*/ @!P1 FADD R6, R6, R5 ; /* 0x0000000506069221 */ /* 0x001fca0000000000 */ /*01d0*/ @!P1 STS [R7.X4], R6 ; /* 0x0000000607009388 */ /* 0x0001e20000004800 */ /*01e0*/ ISETP.NE.AND P1, PT, R4, 0x1, PT ; /* 0x000000010400780c */ /* 0x000fda0003f25270 */ /*01f0*/ @P1 BRA 0x120 ; /* 0xffffff2000001947 */ /* 0x001fea000383ffff */ /*0200*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x001fec0000010000 */ /*0210*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0220*/ LDS R4, [RZ] ; /* 0x00000000ff047984 */ /* 0x000e220000000800 */ /*0230*/ I2F R5, c[0x0][0x170] ; /* 0x00005c0000057b06 */ /* 0x000e700000201400 */ /*0240*/ MUFU.RCP R2, R5 ; /* 0x0000000500027308 */ /* 0x002e640000001000 */ /*0250*/ FFMA R7, -R5, R2, 1 ; /* 0x3f80000005077423 */ /* 0x002fc80000000102 */ /*0260*/ FFMA R7, R2, R7, R2 ; /* 0x0000000702077223 */ /* 0x000fe40000000002 */ /*0270*/ FCHK P0, R4, R5 ; /* 0x0000000504007302 */ /* 0x001e240000000000 */ /*0280*/ FFMA R2, R4, R7, RZ ; /* 0x0000000704027223 */ /* 0x000fc800000000ff */ /*0290*/ FFMA R6, -R5, R2, R4 ; /* 0x0000000205067223 */ /* 0x000fc80000000104 */ /*02a0*/ FFMA R7, R7, R6, R2 ; /* 0x0000000607077223 */ /* 0x000fe20000000002 */ /*02b0*/ @!P0 BRA 0x2e0 ; /* 0x0000002000008947 */ /* 0x001fea0003800000 */ /*02c0*/ MOV R2, 0x2e0 ; /* 0x000002e000027802 */ /* 0x000fe40000000f00 */ /*02d0*/ CALL.REL.NOINC 0x330 ; /* 0x0000005000007944 */ /* 0x000fea0003c00000 */ /*02e0*/ IMAD.MOV.U32 R2, RZ, RZ, 0x4 ; /* 0x00000004ff027424 */ /* 0x000fe400078e00ff */ /*02f0*/ IMAD R3, R3, c[0x0][0x174], R0 ; /* 0x00005d0003037a24 */ /* 0x000fc800078e0200 */ /*0300*/ IMAD.WIDE R2, R3, R2, c[0x0][0x168] ; /* 0x00005a0003027625 */ /* 0x000fca00078e0202 */ /*0310*/ STG.E [R2.64], R7 ; /* 0x0000000702007986 */ /* 0x000fe2000c101904 */ /*0320*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0330*/ SHF.R.U32.HI R7, RZ, 0x17, R5.reuse ; /* 0x00000017ff077819 */ /* 0x100fe20000011605 */ /*0340*/ IMAD.MOV.U32 R8, RZ, RZ, R4.reuse ; /* 0x000000ffff087224 */ /* 0x100fe200078e0004 */ /*0350*/ SHF.R.U32.HI R6, RZ, 0x17, R4 ; /* 0x00000017ff067819 */ /* 0x000fe20000011604 */ /*0360*/ IMAD.MOV.U32 R9, RZ, RZ, R5 ; /* 0x000000ffff097224 */ /* 0x000fe200078e0005 */ /*0370*/ LOP3.LUT R7, R7, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff07077812 */ /* 0x000fe400078ec0ff */ /*0380*/ LOP3.LUT R6, R6, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff06067812 */ /* 0x000fe400078ec0ff */ /*0390*/ IADD3 R12, R7, -0x1, RZ ; /* 0xffffffff070c7810 */ /* 0x000fe40007ffe0ff */ /*03a0*/ IADD3 R11, R6, -0x1, RZ ; /* 0xffffffff060b7810 */ /* 0x000fc40007ffe0ff */ /*03b0*/ ISETP.GT.U32.AND P0, PT, R12, 0xfd, PT ; /* 0x000000fd0c00780c */ /* 0x000fc80003f04070 */ /*03c0*/ ISETP.GT.U32.OR P0, PT, R11, 0xfd, P0 ; /* 0x000000fd0b00780c */ /* 0x000fda0000704470 */ /*03d0*/ @!P0 IMAD.MOV.U32 R10, RZ, RZ, RZ ; /* 0x000000ffff0a8224 */ /* 0x000fe200078e00ff */ /*03e0*/ @!P0 BRA 0x560 ; /* 0x0000017000008947 */ /* 0x000fea0003800000 */ /*03f0*/ FSETP.GTU.FTZ.AND P0, PT, |R4|, +INF , PT ; /* 0x7f8000000400780b */ /* 0x000fe40003f1c200 */ /*0400*/ FSETP.GTU.FTZ.AND P1, PT, |R5|, +INF , PT ; /* 0x7f8000000500780b */ /* 0x000fc80003f3c200 */ /*0410*/ PLOP3.LUT P0, PT, P0, P1, PT, 0xa8, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000703570 */ /*0420*/ @P0 BRA 0x940 ; /* 0x0000051000000947 */ /* 0x000fea0003800000 */ /*0430*/ LOP3.LUT P0, RZ, R9, 0x7fffffff, R8, 0xc8, !PT ; /* 0x7fffffff09ff7812 */ /* 0x000fda000780c808 */ /*0440*/ @!P0 BRA 0x920 ; /* 0x000004d000008947 */ /* 0x000fea0003800000 */ /*0450*/ FSETP.NEU.FTZ.AND P2, PT, |R4|.reuse, +INF , PT ; /* 0x7f8000000400780b */ /* 0x040fe40003f5d200 */ /*0460*/ FSETP.NEU.FTZ.AND P1, PT, |R5|, +INF , PT ; /* 0x7f8000000500780b */ /* 0x000fe40003f3d200 */ /*0470*/ FSETP.NEU.FTZ.AND P0, PT, |R4|, +INF , PT ; /* 0x7f8000000400780b */ /* 0x000fd60003f1d200 */ /*0480*/ @!P1 BRA !P2, 0x920 ; /* 0x0000049000009947 */ /* 0x000fea0005000000 */ /*0490*/ LOP3.LUT P2, RZ, R8, 0x7fffffff, RZ, 0xc0, !PT ; /* 0x7fffffff08ff7812 */ /* 0x000fc8000784c0ff */ /*04a0*/ PLOP3.LUT P1, PT, P1, P2, PT, 0x2a, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000f24572 */ /*04b0*/ @P1 BRA 0x900 ; /* 0x0000044000001947 */ /* 0x000fea0003800000 */ /*04c0*/ LOP3.LUT P1, RZ, R9, 0x7fffffff, RZ, 0xc0, !PT ; /* 0x7fffffff09ff7812 */ /* 0x000fc8000782c0ff */ /*04d0*/ PLOP3.LUT P0, PT, P0, P1, PT, 0x2a, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000702572 */ /*04e0*/ @P0 BRA 0x8d0 ; /* 0x000003e000000947 */ /* 0x000fea0003800000 */ /*04f0*/ ISETP.GE.AND P0, PT, R11, RZ, PT ; /* 0x000000ff0b00720c */ /* 0x000fe40003f06270 */ /*0500*/ ISETP.GE.AND P1, PT, R12, RZ, PT ; /* 0x000000ff0c00720c */ /* 0x000fd60003f26270 */ /*0510*/ @P0 IMAD.MOV.U32 R10, RZ, RZ, RZ ; /* 0x000000ffff0a0224 */ /* 0x000fe400078e00ff */ /*0520*/ @!P0 IMAD.MOV.U32 R10, RZ, RZ, -0x40 ; /* 0xffffffc0ff0a8424 */ /* 0x000fe400078e00ff */ /*0530*/ @!P0 FFMA R8, R4, 1.84467440737095516160e+19, RZ ; /* 0x5f80000004088823 */ /* 0x000fe400000000ff */ /*0540*/ @!P1 FFMA R9, R5, 1.84467440737095516160e+19, RZ ; /* 0x5f80000005099823 */ /* 0x000fe200000000ff */ /*0550*/ @!P1 IADD3 R10, R10, 0x40, RZ ; /* 0x000000400a0a9810 */ /* 0x000fe40007ffe0ff */ /*0560*/ LEA R4, R7, 0xc0800000, 0x17 ; /* 0xc080000007047811 */ /* 0x000fe400078eb8ff */ /*0570*/ IADD3 R6, R6, -0x7f, RZ ; /* 0xffffff8106067810 */ /* 0x000fc60007ffe0ff */ /*0580*/ IMAD.IADD R9, R9, 0x1, -R4 ; /* 0x0000000109097824 */ /* 0x000fe200078e0a04 */ /*0590*/ IADD3 R7, R6.reuse, 0x7f, -R7 ; /* 0x0000007f06077810 */ /* 0x040fe20007ffe807 */ /*05a0*/ IMAD R8, R6, -0x800000, R8 ; /* 0xff80000006087824 */ /* 0x000fe400078e0208 */ /*05b0*/ MUFU.RCP R4, R9 ; /* 0x0000000900047308 */ /* 0x000e220000001000 */ /*05c0*/ FADD.FTZ R5, -R9, -RZ ; /* 0x800000ff09057221 */ /* 0x000fe40000010100 */ /*05d0*/ IMAD.IADD R7, R7, 0x1, R10 ; /* 0x0000000107077824 */ /* 0x000fe400078e020a */ /*05e0*/ FFMA R11, R4, R5, 1 ; /* 0x3f800000040b7423 */ /* 0x001fc80000000005 */ /*05f0*/ FFMA R13, R4, R11, R4 ; /* 0x0000000b040d7223 */ /* 0x000fc80000000004 */ /*0600*/ FFMA R4, R8, R13, RZ ; /* 0x0000000d08047223 */ /* 0x000fc800000000ff */ /*0610*/ FFMA R11, R5, R4, R8 ; /* 0x00000004050b7223 */ /* 0x000fc80000000008 */ /*0620*/ FFMA R12, R13, R11, R4 ; /* 0x0000000b0d0c7223 */ /* 0x000fc80000000004 */ /*0630*/ FFMA R8, R5, R12, R8 ; /* 0x0000000c05087223 */ /* 0x000fc80000000008 */ /*0640*/ FFMA R4, R13, R8, R12 ; /* 0x000000080d047223 */ /* 0x000fca000000000c */ /*0650*/ SHF.R.U32.HI R5, RZ, 0x17, R4 ; /* 0x00000017ff057819 */ /* 0x000fc80000011604 */ /*0660*/ LOP3.LUT R5, R5, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff05057812 */ /* 0x000fca00078ec0ff */ /*0670*/ IMAD.IADD R9, R5, 0x1, R7 ; /* 0x0000000105097824 */ /* 0x000fca00078e0207 */ /*0680*/ IADD3 R5, R9, -0x1, RZ ; /* 0xffffffff09057810 */ /* 0x000fc80007ffe0ff */ /*0690*/ ISETP.GE.U32.AND P0, PT, R5, 0xfe, PT ; /* 0x000000fe0500780c */ /* 0x000fda0003f06070 */ /*06a0*/ @!P0 BRA 0x8b0 ; /* 0x0000020000008947 */ /* 0x000fea0003800000 */ /*06b0*/ ISETP.GT.AND P0, PT, R9, 0xfe, PT ; /* 0x000000fe0900780c */ /* 0x000fda0003f04270 */ /*06c0*/ @P0 BRA 0x880 ; /* 0x000001b000000947 */ /* 0x000fea0003800000 */ /*06d0*/ ISETP.GE.AND P0, PT, R9, 0x1, PT ; /* 0x000000010900780c */ /* 0x000fda0003f06270 */ /*06e0*/ @P0 BRA 0x950 ; /* 0x0000026000000947 */ /* 0x000fea0003800000 */ /*06f0*/ ISETP.GE.AND P0, PT, R9, -0x18, PT ; /* 0xffffffe80900780c */ /* 0x000fe40003f06270 */ /*0700*/ LOP3.LUT R4, R4, 0x80000000, RZ, 0xc0, !PT ; /* 0x8000000004047812 */ /* 0x000fd600078ec0ff */ /*0710*/ @!P0 BRA 0x950 ; /* 0x0000023000008947 */ /* 0x000fea0003800000 */ /*0720*/ FFMA.RZ R5, R13, R8.reuse, R12.reuse ; /* 0x000000080d057223 */ /* 0x180fe2000000c00c */ /*0730*/ ISETP.NE.AND P2, PT, R9, RZ, PT ; /* 0x000000ff0900720c */ /* 0x000fe20003f45270 */ /*0740*/ FFMA.RM R6, R13, R8.reuse, R12.reuse ; /* 0x000000080d067223 */ /* 0x180fe2000000400c */ /*0750*/ ISETP.NE.AND P1, PT, R9, RZ, PT ; /* 0x000000ff0900720c */ /* 0x000fe40003f25270 */ /*0760*/ LOP3.LUT R7, R5, 0x7fffff, RZ, 0xc0, !PT ; /* 0x007fffff05077812 */ /* 0x000fe200078ec0ff */ /*0770*/ FFMA.RP R5, R13, R8, R12 ; /* 0x000000080d057223 */ /* 0x000fe2000000800c */ /*0780*/ IADD3 R8, R9, 0x20, RZ ; /* 0x0000002009087810 */ /* 0x000fe20007ffe0ff */ /*0790*/ IMAD.MOV R9, RZ, RZ, -R9 ; /* 0x000000ffff097224 */ /* 0x000fe200078e0a09 */ /*07a0*/ LOP3.LUT R7, R7, 0x800000, RZ, 0xfc, !PT ; /* 0x0080000007077812 */ /* 0x000fe400078efcff */ /*07b0*/ FSETP.NEU.FTZ.AND P0, PT, R5, R6, PT ; /* 0x000000060500720b */ /* 0x000fc40003f1d000 */ /*07c0*/ SHF.L.U32 R8, R7, R8, RZ ; /* 0x0000000807087219 */ /* 0x000fe400000006ff */ /*07d0*/ SEL R6, R9, RZ, P2 ; /* 0x000000ff09067207 */ /* 0x000fe40001000000 */ /*07e0*/ ISETP.NE.AND P1, PT, R8, RZ, P1 ; /* 0x000000ff0800720c */ /* 0x000fe40000f25270 */ /*07f0*/ SHF.R.U32.HI R6, RZ, R6, R7 ; /* 0x00000006ff067219 */ /* 0x000fe40000011607 */ /*0800*/ PLOP3.LUT P0, PT, P0, P1, PT, 0xa8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe40000703570 */ /*0810*/ SHF.R.U32.HI R8, RZ, 0x1, R6 ; /* 0x00000001ff087819 */ /* 0x000fc40000011606 */ /*0820*/ SEL R5, RZ, 0x1, !P0 ; /* 0x00000001ff057807 */ /* 0x000fc80004000000 */ /*0830*/ LOP3.LUT R5, R5, 0x1, R8, 0xf8, !PT ; /* 0x0000000105057812 */ /* 0x000fc800078ef808 */ /*0840*/ LOP3.LUT R5, R5, R6, RZ, 0xc0, !PT ; /* 0x0000000605057212 */ /* 0x000fca00078ec0ff */ /*0850*/ IMAD.IADD R5, R8, 0x1, R5 ; /* 0x0000000108057824 */ /* 0x000fca00078e0205 */ /*0860*/ LOP3.LUT R4, R5, R4, RZ, 0xfc, !PT ; /* 0x0000000405047212 */ /* 0x000fe200078efcff */ /*0870*/ BRA 0x950 ; /* 0x000000d000007947 */ /* 0x000fea0003800000 */ /*0880*/ LOP3.LUT R4, R4, 0x80000000, RZ, 0xc0, !PT ; /* 0x8000000004047812 */ /* 0x000fc800078ec0ff */ /*0890*/ LOP3.LUT R4, R4, 0x7f800000, RZ, 0xfc, !PT ; /* 0x7f80000004047812 */ /* 0x000fe200078efcff */ /*08a0*/ BRA 0x950 ; /* 0x000000a000007947 */ /* 0x000fea0003800000 */ /*08b0*/ IMAD R4, R7, 0x800000, R4 ; /* 0x0080000007047824 */ /* 0x000fe200078e0204 */ /*08c0*/ BRA 0x950 ; /* 0x0000008000007947 */ /* 0x000fea0003800000 */ /*08d0*/ LOP3.LUT R4, R9, 0x80000000, R8, 0x48, !PT ; /* 0x8000000009047812 */ /* 0x000fc800078e4808 */ /*08e0*/ LOP3.LUT R4, R4, 0x7f800000, RZ, 0xfc, !PT ; /* 0x7f80000004047812 */ /* 0x000fe200078efcff */ /*08f0*/ BRA 0x950 ; /* 0x0000005000007947 */ /* 0x000fea0003800000 */ /*0900*/ LOP3.LUT R4, R9, 0x80000000, R8, 0x48, !PT ; /* 0x8000000009047812 */ /* 0x000fe200078e4808 */ /*0910*/ BRA 0x950 ; /* 0x0000003000007947 */ /* 0x000fea0003800000 */ /*0920*/ MUFU.RSQ R4, -QNAN ; /* 0xffc0000000047908 */ /* 0x000e220000001400 */ /*0930*/ BRA 0x950 ; /* 0x0000001000007947 */ /* 0x000fea0003800000 */ /*0940*/ FADD.FTZ R4, R4, R5 ; /* 0x0000000504047221 */ /* 0x000fc80000010000 */ /*0950*/ IMAD.MOV.U32 R7, RZ, RZ, R4 ; /* 0x000000ffff077224 */ /* 0x001fe400078e0004 */ /*0960*/ IMAD.MOV.U32 R4, RZ, RZ, R2 ; /* 0x000000ffff047224 */ /* 0x000fe400078e0002 */ /*0970*/ IMAD.MOV.U32 R5, RZ, RZ, 0x0 ; /* 0x00000000ff057424 */ /* 0x000fc800078e00ff */ /*0980*/ RET.REL.NODEC R4 0x0 ; /* 0xfffff67004007950 */ /* 0x000fea0003c3ffff */ /*0990*/ BRA 0x990; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*09a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a00*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a10*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a20*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a30*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a40*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a50*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a60*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a70*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include "includes.h" __global__ void g_One_Bgrad(float* _delta, float* bgrad, int rows, int cols, int channels) { extern __shared__ float _sum[]; int channel = blockIdx.x; int col = blockIdx.y; int row = threadIdx.x; float delta = _delta[channel * rows * cols + row * cols + col]; _sum[row] = delta; __syncthreads(); int len = rows; while(len != 1) { __syncthreads(); int skip = (len + 1) >> 1; if(threadIdx.x < (len >> 1)) { _sum[threadIdx.x] += _sum[threadIdx.x + skip]; } len = (len + 1) >> 1; } __syncthreads(); if(threadIdx.x == 0) { bgrad[channel * cols + col] = _sum[0] / rows; } }
.file "tmpxft_00153560_00000000-6_g_One_Bgrad.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 _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii .type _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii, @function _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii: .LFB2051: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%rsp) movl %ecx, 8(%rsp) movl %r8d, 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 12(%rsp), %rax movq %rax, 112(%rsp) leaq 8(%rsp), %rax movq %rax, 120(%rsp) leaq 4(%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 _Z11g_One_BgradPfS_iii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2051: .size _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii, .-_Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii .globl _Z11g_One_BgradPfS_iii .type _Z11g_One_BgradPfS_iii, @function _Z11g_One_BgradPfS_iii: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z11g_One_BgradPfS_iii, .-_Z11g_One_BgradPfS_iii .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z11g_One_BgradPfS_iii" .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 _Z11g_One_BgradPfS_iii(%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 g_One_Bgrad(float* _delta, float* bgrad, int rows, int cols, int channels) { extern __shared__ float _sum[]; int channel = blockIdx.x; int col = blockIdx.y; int row = threadIdx.x; float delta = _delta[channel * rows * cols + row * cols + col]; _sum[row] = delta; __syncthreads(); int len = rows; while(len != 1) { __syncthreads(); int skip = (len + 1) >> 1; if(threadIdx.x < (len >> 1)) { _sum[threadIdx.x] += _sum[threadIdx.x + skip]; } len = (len + 1) >> 1; } __syncthreads(); if(threadIdx.x == 0) { bgrad[channel * cols + col] = _sum[0] / rows; } }
#include <hip/hip_runtime.h> #include "includes.h" __global__ void g_One_Bgrad(float* _delta, float* bgrad, int rows, int cols, int channels) { extern __shared__ float _sum[]; int channel = blockIdx.x; int col = blockIdx.y; int row = threadIdx.x; float delta = _delta[channel * rows * cols + row * cols + col]; _sum[row] = delta; __syncthreads(); int len = rows; while(len != 1) { __syncthreads(); int skip = (len + 1) >> 1; if(threadIdx.x < (len >> 1)) { _sum[threadIdx.x] += _sum[threadIdx.x + skip]; } len = (len + 1) >> 1; } __syncthreads(); if(threadIdx.x == 0) { bgrad[channel * cols + col] = _sum[0] / rows; } }
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 g_One_Bgrad(float* _delta, float* bgrad, int rows, int cols, int channels) { extern __shared__ float _sum[]; int channel = blockIdx.x; int col = blockIdx.y; int row = threadIdx.x; float delta = _delta[channel * rows * cols + row * cols + col]; _sum[row] = delta; __syncthreads(); int len = rows; while(len != 1) { __syncthreads(); int skip = (len + 1) >> 1; if(threadIdx.x < (len >> 1)) { _sum[threadIdx.x] += _sum[threadIdx.x + skip]; } len = (len + 1) >> 1; } __syncthreads(); if(threadIdx.x == 0) { bgrad[channel * cols + col] = _sum[0] / rows; } }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z11g_One_BgradPfS_iii .globl _Z11g_One_BgradPfS_iii .p2align 8 .type _Z11g_One_BgradPfS_iii,@function _Z11g_One_BgradPfS_iii: s_clause 0x1 s_load_b64 s[4:5], s[0:1], 0x10 s_load_b64 s[6:7], s[0:1], 0x0 s_mov_b32 s2, s15 s_waitcnt lgkmcnt(0) v_mad_u64_u32 v[1:2], null, s14, s4, v[0:1] s_cmp_eq_u32 s4, 1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u64_u32 v[2:3], null, v1, s5, s[2:3] v_ashrrev_i32_e32 v3, 31, v2 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b64 v[1:2], 2, v[2:3] v_add_co_u32 v1, vcc_lo, s6, v1 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v2, vcc_lo, s7, v2, vcc_lo global_load_b32 v2, v[1:2], off v_lshl_add_u32 v1, v0, 2, 0 s_waitcnt vmcnt(0) ds_store_b32 v1, v2 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_cbranch_scc1 .LBB0_5 s_mov_b32 s3, s4 s_branch .LBB0_3 .p2align 6 .LBB0_2: s_or_b32 exec_lo, exec_lo, s6 s_cmp_lg_u32 s3, 1 s_cbranch_scc0 .LBB0_5 .LBB0_3: s_ashr_i32 s6, s3, 1 s_add_i32 s3, s3, 1 v_cmp_gt_u32_e32 vcc_lo, s6, v0 s_ashr_i32 s3, s3, 1 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_and_saveexec_b32 s6, vcc_lo s_cbranch_execz .LBB0_2 v_add_nc_u32_e32 v2, s3, v0 s_delay_alu instid0(VALU_DEP_1) v_lshl_add_u32 v2, v2, 2, 0 ds_load_b32 v2, v2 ds_load_b32 v3, v1 s_waitcnt lgkmcnt(0) v_add_f32_e32 v2, v2, v3 ds_store_b32 v1, v2 s_branch .LBB0_2 .LBB0_5: s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_mov_b32 s3, exec_lo v_cmpx_eq_u32_e32 0, v0 s_cbranch_execz .LBB0_7 v_mov_b32_e32 v0, 0 v_cvt_f32_i32_e32 v1, s4 s_load_b64 s[0:1], s[0:1], 0x8 s_mul_i32 s3, s14, s5 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) s_add_i32 s2, s3, s2 ds_load_b32 v0, v0 s_ashr_i32 s3, s2, 31 s_lshl_b64 s[2:3], s[2:3], 2 s_waitcnt lgkmcnt(0) s_add_u32 s0, s0, s2 s_addc_u32 s1, s1, s3 v_div_scale_f32 v2, null, v1, v1, v0 v_div_scale_f32 v5, vcc_lo, v0, v1, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_rcp_f32_e32 v3, v2 s_waitcnt_depctr 0xfff v_fma_f32 v4, -v2, v3, 1.0 v_fmac_f32_e32 v3, v4, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f32_e32 v4, v5, v3 v_fma_f32 v6, -v2, v4, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fmac_f32_e32 v4, v6, v3 v_fma_f32 v2, -v2, v4, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_div_fmas_f32 v2, v2, v3, v4 v_div_fixup_f32 v0, v2, v1, v0 v_mov_b32_e32 v1, 0 global_store_b32 v1, v0, s[0:1] .LBB0_7: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z11g_One_BgradPfS_iii .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 28 .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 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 _Z11g_One_BgradPfS_iii, .Lfunc_end0-_Z11g_One_BgradPfS_iii .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 - .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: _Z11g_One_BgradPfS_iii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z11g_One_BgradPfS_iii.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 <hip/hip_runtime.h> #include "includes.h" __global__ void g_One_Bgrad(float* _delta, float* bgrad, int rows, int cols, int channels) { extern __shared__ float _sum[]; int channel = blockIdx.x; int col = blockIdx.y; int row = threadIdx.x; float delta = _delta[channel * rows * cols + row * cols + col]; _sum[row] = delta; __syncthreads(); int len = rows; while(len != 1) { __syncthreads(); int skip = (len + 1) >> 1; if(threadIdx.x < (len >> 1)) { _sum[threadIdx.x] += _sum[threadIdx.x + skip]; } len = (len + 1) >> 1; } __syncthreads(); if(threadIdx.x == 0) { bgrad[channel * cols + col] = _sum[0] / rows; } }
.text .file "g_One_Bgrad.hip" .globl _Z26__device_stub__g_One_BgradPfS_iii # -- Begin function _Z26__device_stub__g_One_BgradPfS_iii .p2align 4, 0x90 .type _Z26__device_stub__g_One_BgradPfS_iii,@function _Z26__device_stub__g_One_BgradPfS_iii: # @_Z26__device_stub__g_One_BgradPfS_iii .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%rsp) movl %ecx, 8(%rsp) movl %r8d, 4(%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 4(%rsp), %rax movq %rax, 112(%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 $_Z11g_One_BgradPfS_iii, %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 _Z26__device_stub__g_One_BgradPfS_iii, .Lfunc_end0-_Z26__device_stub__g_One_BgradPfS_iii .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 $_Z11g_One_BgradPfS_iii, %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 _Z11g_One_BgradPfS_iii,@object # @_Z11g_One_BgradPfS_iii .section .rodata,"a",@progbits .globl _Z11g_One_BgradPfS_iii .p2align 3, 0x0 _Z11g_One_BgradPfS_iii: .quad _Z26__device_stub__g_One_BgradPfS_iii .size _Z11g_One_BgradPfS_iii, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z11g_One_BgradPfS_iii" .size .L__unnamed_1, 23 .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__g_One_BgradPfS_iii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z11g_One_BgradPfS_iii .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 : _Z11g_One_BgradPfS_iii .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 R3, SR_CTAID.X ; /* 0x0000000000037919 */ /* 0x000e220000002500 */ /*0020*/ IMAD.MOV.U32 R4, RZ, RZ, 0x4 ; /* 0x00000004ff047424 */ /* 0x000fe200078e00ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fe40000000a00 */ /*0040*/ S2R R7, SR_TID.X ; /* 0x0000000000077919 */ /* 0x000e280000002100 */ /*0050*/ S2R R0, SR_CTAID.Y ; /* 0x0000000000007919 */ /* 0x000e620000002600 */ /*0060*/ IMAD R5, R3, c[0x0][0x170], R7 ; /* 0x00005c0003057a24 */ /* 0x001fc800078e0207 */ /*0070*/ IMAD R5, R5, c[0x0][0x174], R0 ; /* 0x00005d0005057a24 */ /* 0x002fc800078e0200 */ /*0080*/ IMAD.WIDE R4, R5, R4, c[0x0][0x160] ; /* 0x0000580005047625 */ /* 0x000fcc00078e0204 */ /*0090*/ LDG.E R4, [R4.64] ; /* 0x0000000404047981 */ /* 0x000ea2000c1e1900 */ /*00a0*/ IMAD.MOV.U32 R2, RZ, RZ, c[0x0][0x170] ; /* 0x00005c00ff027624 */ /* 0x000fe200078e00ff */ /*00b0*/ ISETP.NE.AND P0, PT, R7, RZ, PT ; /* 0x000000ff0700720c */ /* 0x000fc80003f05270 */ /*00c0*/ ISETP.NE.AND P1, PT, R2, 0x1, PT ; /* 0x000000010200780c */ /* 0x000fe20003f25270 */ /*00d0*/ STS [R7.X4], R4 ; /* 0x0000000407007388 */ /* 0x0041e80000004800 */ /*00e0*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000ff00000010000 */ /*00f0*/ @!P1 BRA 0x200 ; /* 0x0000010000009947 */ /* 0x000fea0003800000 */ /*0100*/ IMAD.SHL.U32 R2, R7, 0x4, RZ ; /* 0x0000000407027824 */ /* 0x001fe400078e00ff */ /*0110*/ IMAD.MOV.U32 R4, RZ, RZ, c[0x0][0x170] ; /* 0x00005c00ff047624 */ /* 0x000fca00078e00ff */ /*0120*/ SHF.R.S32.HI R5, RZ, 0x1, R4 ; /* 0x00000001ff057819 */ /* 0x000fe20000011404 */ /*0130*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x000fe60000010000 */ /*0140*/ ISETP.GE.U32.AND P1, PT, R7, R5, PT ; /* 0x000000050700720c */ /* 0x000fda0003f26070 */ /*0150*/ @!P1 LEA R5, R4.reuse, 0x2, 0x1 ; /* 0x0000000204059811 */ /* 0x040fe400078e08ff */ /*0160*/ IADD3 R4, R4, 0x1, RZ ; /* 0x0000000104047810 */ /* 0x000fe40007ffe0ff */ /*0170*/ @!P1 LOP3.LUT R5, R5, 0xfffffffc, RZ, 0xc0, !PT ; /* 0xfffffffc05059812 */ /* 0x000fe400078ec0ff */ /*0180*/ SHF.R.S32.HI R4, RZ, 0x1, R4 ; /* 0x00000001ff047819 */ /* 0x000fe20000011404 */ /*0190*/ @!P1 LDS R6, [R7.X4] ; /* 0x0000000007069984 */ /* 0x000fe40000004800 */ /*01a0*/ @!P1 IMAD.IADD R5, R2, 0x1, R5 ; /* 0x0000000102059824 */ /* 0x000fcc00078e0205 */ /*01b0*/ @!P1 LDS R5, [R5] ; /* 0x0000000005059984 */ /* 0x000e240000000800 */ /*01c0*/ @!P1 FADD R6, R6, R5 ; /* 0x0000000506069221 */ /* 0x001fca0000000000 */ /*01d0*/ @!P1 STS [R7.X4], R6 ; /* 0x0000000607009388 */ /* 0x0001e20000004800 */ /*01e0*/ ISETP.NE.AND P1, PT, R4, 0x1, PT ; /* 0x000000010400780c */ /* 0x000fda0003f25270 */ /*01f0*/ @P1 BRA 0x120 ; /* 0xffffff2000001947 */ /* 0x001fea000383ffff */ /*0200*/ BAR.SYNC.DEFER_BLOCKING 0x0 ; /* 0x0000000000007b1d */ /* 0x001fec0000010000 */ /*0210*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*0220*/ LDS R4, [RZ] ; /* 0x00000000ff047984 */ /* 0x000e220000000800 */ /*0230*/ I2F R5, c[0x0][0x170] ; /* 0x00005c0000057b06 */ /* 0x000e700000201400 */ /*0240*/ MUFU.RCP R2, R5 ; /* 0x0000000500027308 */ /* 0x002e640000001000 */ /*0250*/ FFMA R7, -R5, R2, 1 ; /* 0x3f80000005077423 */ /* 0x002fc80000000102 */ /*0260*/ FFMA R7, R2, R7, R2 ; /* 0x0000000702077223 */ /* 0x000fe40000000002 */ /*0270*/ FCHK P0, R4, R5 ; /* 0x0000000504007302 */ /* 0x001e240000000000 */ /*0280*/ FFMA R2, R4, R7, RZ ; /* 0x0000000704027223 */ /* 0x000fc800000000ff */ /*0290*/ FFMA R6, -R5, R2, R4 ; /* 0x0000000205067223 */ /* 0x000fc80000000104 */ /*02a0*/ FFMA R7, R7, R6, R2 ; /* 0x0000000607077223 */ /* 0x000fe20000000002 */ /*02b0*/ @!P0 BRA 0x2e0 ; /* 0x0000002000008947 */ /* 0x001fea0003800000 */ /*02c0*/ MOV R2, 0x2e0 ; /* 0x000002e000027802 */ /* 0x000fe40000000f00 */ /*02d0*/ CALL.REL.NOINC 0x330 ; /* 0x0000005000007944 */ /* 0x000fea0003c00000 */ /*02e0*/ IMAD.MOV.U32 R2, RZ, RZ, 0x4 ; /* 0x00000004ff027424 */ /* 0x000fe400078e00ff */ /*02f0*/ IMAD R3, R3, c[0x0][0x174], R0 ; /* 0x00005d0003037a24 */ /* 0x000fc800078e0200 */ /*0300*/ IMAD.WIDE R2, R3, R2, c[0x0][0x168] ; /* 0x00005a0003027625 */ /* 0x000fca00078e0202 */ /*0310*/ STG.E [R2.64], R7 ; /* 0x0000000702007986 */ /* 0x000fe2000c101904 */ /*0320*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*0330*/ SHF.R.U32.HI R7, RZ, 0x17, R5.reuse ; /* 0x00000017ff077819 */ /* 0x100fe20000011605 */ /*0340*/ IMAD.MOV.U32 R8, RZ, RZ, R4.reuse ; /* 0x000000ffff087224 */ /* 0x100fe200078e0004 */ /*0350*/ SHF.R.U32.HI R6, RZ, 0x17, R4 ; /* 0x00000017ff067819 */ /* 0x000fe20000011604 */ /*0360*/ IMAD.MOV.U32 R9, RZ, RZ, R5 ; /* 0x000000ffff097224 */ /* 0x000fe200078e0005 */ /*0370*/ LOP3.LUT R7, R7, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff07077812 */ /* 0x000fe400078ec0ff */ /*0380*/ LOP3.LUT R6, R6, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff06067812 */ /* 0x000fe400078ec0ff */ /*0390*/ IADD3 R12, R7, -0x1, RZ ; /* 0xffffffff070c7810 */ /* 0x000fe40007ffe0ff */ /*03a0*/ IADD3 R11, R6, -0x1, RZ ; /* 0xffffffff060b7810 */ /* 0x000fc40007ffe0ff */ /*03b0*/ ISETP.GT.U32.AND P0, PT, R12, 0xfd, PT ; /* 0x000000fd0c00780c */ /* 0x000fc80003f04070 */ /*03c0*/ ISETP.GT.U32.OR P0, PT, R11, 0xfd, P0 ; /* 0x000000fd0b00780c */ /* 0x000fda0000704470 */ /*03d0*/ @!P0 IMAD.MOV.U32 R10, RZ, RZ, RZ ; /* 0x000000ffff0a8224 */ /* 0x000fe200078e00ff */ /*03e0*/ @!P0 BRA 0x560 ; /* 0x0000017000008947 */ /* 0x000fea0003800000 */ /*03f0*/ FSETP.GTU.FTZ.AND P0, PT, |R4|, +INF , PT ; /* 0x7f8000000400780b */ /* 0x000fe40003f1c200 */ /*0400*/ FSETP.GTU.FTZ.AND P1, PT, |R5|, +INF , PT ; /* 0x7f8000000500780b */ /* 0x000fc80003f3c200 */ /*0410*/ PLOP3.LUT P0, PT, P0, P1, PT, 0xa8, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000703570 */ /*0420*/ @P0 BRA 0x940 ; /* 0x0000051000000947 */ /* 0x000fea0003800000 */ /*0430*/ LOP3.LUT P0, RZ, R9, 0x7fffffff, R8, 0xc8, !PT ; /* 0x7fffffff09ff7812 */ /* 0x000fda000780c808 */ /*0440*/ @!P0 BRA 0x920 ; /* 0x000004d000008947 */ /* 0x000fea0003800000 */ /*0450*/ FSETP.NEU.FTZ.AND P2, PT, |R4|.reuse, +INF , PT ; /* 0x7f8000000400780b */ /* 0x040fe40003f5d200 */ /*0460*/ FSETP.NEU.FTZ.AND P1, PT, |R5|, +INF , PT ; /* 0x7f8000000500780b */ /* 0x000fe40003f3d200 */ /*0470*/ FSETP.NEU.FTZ.AND P0, PT, |R4|, +INF , PT ; /* 0x7f8000000400780b */ /* 0x000fd60003f1d200 */ /*0480*/ @!P1 BRA !P2, 0x920 ; /* 0x0000049000009947 */ /* 0x000fea0005000000 */ /*0490*/ LOP3.LUT P2, RZ, R8, 0x7fffffff, RZ, 0xc0, !PT ; /* 0x7fffffff08ff7812 */ /* 0x000fc8000784c0ff */ /*04a0*/ PLOP3.LUT P1, PT, P1, P2, PT, 0x2a, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000f24572 */ /*04b0*/ @P1 BRA 0x900 ; /* 0x0000044000001947 */ /* 0x000fea0003800000 */ /*04c0*/ LOP3.LUT P1, RZ, R9, 0x7fffffff, RZ, 0xc0, !PT ; /* 0x7fffffff09ff7812 */ /* 0x000fc8000782c0ff */ /*04d0*/ PLOP3.LUT P0, PT, P0, P1, PT, 0x2a, 0x0 ; /* 0x000000000000781c */ /* 0x000fda0000702572 */ /*04e0*/ @P0 BRA 0x8d0 ; /* 0x000003e000000947 */ /* 0x000fea0003800000 */ /*04f0*/ ISETP.GE.AND P0, PT, R11, RZ, PT ; /* 0x000000ff0b00720c */ /* 0x000fe40003f06270 */ /*0500*/ ISETP.GE.AND P1, PT, R12, RZ, PT ; /* 0x000000ff0c00720c */ /* 0x000fd60003f26270 */ /*0510*/ @P0 IMAD.MOV.U32 R10, RZ, RZ, RZ ; /* 0x000000ffff0a0224 */ /* 0x000fe400078e00ff */ /*0520*/ @!P0 IMAD.MOV.U32 R10, RZ, RZ, -0x40 ; /* 0xffffffc0ff0a8424 */ /* 0x000fe400078e00ff */ /*0530*/ @!P0 FFMA R8, R4, 1.84467440737095516160e+19, RZ ; /* 0x5f80000004088823 */ /* 0x000fe400000000ff */ /*0540*/ @!P1 FFMA R9, R5, 1.84467440737095516160e+19, RZ ; /* 0x5f80000005099823 */ /* 0x000fe200000000ff */ /*0550*/ @!P1 IADD3 R10, R10, 0x40, RZ ; /* 0x000000400a0a9810 */ /* 0x000fe40007ffe0ff */ /*0560*/ LEA R4, R7, 0xc0800000, 0x17 ; /* 0xc080000007047811 */ /* 0x000fe400078eb8ff */ /*0570*/ IADD3 R6, R6, -0x7f, RZ ; /* 0xffffff8106067810 */ /* 0x000fc60007ffe0ff */ /*0580*/ IMAD.IADD R9, R9, 0x1, -R4 ; /* 0x0000000109097824 */ /* 0x000fe200078e0a04 */ /*0590*/ IADD3 R7, R6.reuse, 0x7f, -R7 ; /* 0x0000007f06077810 */ /* 0x040fe20007ffe807 */ /*05a0*/ IMAD R8, R6, -0x800000, R8 ; /* 0xff80000006087824 */ /* 0x000fe400078e0208 */ /*05b0*/ MUFU.RCP R4, R9 ; /* 0x0000000900047308 */ /* 0x000e220000001000 */ /*05c0*/ FADD.FTZ R5, -R9, -RZ ; /* 0x800000ff09057221 */ /* 0x000fe40000010100 */ /*05d0*/ IMAD.IADD R7, R7, 0x1, R10 ; /* 0x0000000107077824 */ /* 0x000fe400078e020a */ /*05e0*/ FFMA R11, R4, R5, 1 ; /* 0x3f800000040b7423 */ /* 0x001fc80000000005 */ /*05f0*/ FFMA R13, R4, R11, R4 ; /* 0x0000000b040d7223 */ /* 0x000fc80000000004 */ /*0600*/ FFMA R4, R8, R13, RZ ; /* 0x0000000d08047223 */ /* 0x000fc800000000ff */ /*0610*/ FFMA R11, R5, R4, R8 ; /* 0x00000004050b7223 */ /* 0x000fc80000000008 */ /*0620*/ FFMA R12, R13, R11, R4 ; /* 0x0000000b0d0c7223 */ /* 0x000fc80000000004 */ /*0630*/ FFMA R8, R5, R12, R8 ; /* 0x0000000c05087223 */ /* 0x000fc80000000008 */ /*0640*/ FFMA R4, R13, R8, R12 ; /* 0x000000080d047223 */ /* 0x000fca000000000c */ /*0650*/ SHF.R.U32.HI R5, RZ, 0x17, R4 ; /* 0x00000017ff057819 */ /* 0x000fc80000011604 */ /*0660*/ LOP3.LUT R5, R5, 0xff, RZ, 0xc0, !PT ; /* 0x000000ff05057812 */ /* 0x000fca00078ec0ff */ /*0670*/ IMAD.IADD R9, R5, 0x1, R7 ; /* 0x0000000105097824 */ /* 0x000fca00078e0207 */ /*0680*/ IADD3 R5, R9, -0x1, RZ ; /* 0xffffffff09057810 */ /* 0x000fc80007ffe0ff */ /*0690*/ ISETP.GE.U32.AND P0, PT, R5, 0xfe, PT ; /* 0x000000fe0500780c */ /* 0x000fda0003f06070 */ /*06a0*/ @!P0 BRA 0x8b0 ; /* 0x0000020000008947 */ /* 0x000fea0003800000 */ /*06b0*/ ISETP.GT.AND P0, PT, R9, 0xfe, PT ; /* 0x000000fe0900780c */ /* 0x000fda0003f04270 */ /*06c0*/ @P0 BRA 0x880 ; /* 0x000001b000000947 */ /* 0x000fea0003800000 */ /*06d0*/ ISETP.GE.AND P0, PT, R9, 0x1, PT ; /* 0x000000010900780c */ /* 0x000fda0003f06270 */ /*06e0*/ @P0 BRA 0x950 ; /* 0x0000026000000947 */ /* 0x000fea0003800000 */ /*06f0*/ ISETP.GE.AND P0, PT, R9, -0x18, PT ; /* 0xffffffe80900780c */ /* 0x000fe40003f06270 */ /*0700*/ LOP3.LUT R4, R4, 0x80000000, RZ, 0xc0, !PT ; /* 0x8000000004047812 */ /* 0x000fd600078ec0ff */ /*0710*/ @!P0 BRA 0x950 ; /* 0x0000023000008947 */ /* 0x000fea0003800000 */ /*0720*/ FFMA.RZ R5, R13, R8.reuse, R12.reuse ; /* 0x000000080d057223 */ /* 0x180fe2000000c00c */ /*0730*/ ISETP.NE.AND P2, PT, R9, RZ, PT ; /* 0x000000ff0900720c */ /* 0x000fe20003f45270 */ /*0740*/ FFMA.RM R6, R13, R8.reuse, R12.reuse ; /* 0x000000080d067223 */ /* 0x180fe2000000400c */ /*0750*/ ISETP.NE.AND P1, PT, R9, RZ, PT ; /* 0x000000ff0900720c */ /* 0x000fe40003f25270 */ /*0760*/ LOP3.LUT R7, R5, 0x7fffff, RZ, 0xc0, !PT ; /* 0x007fffff05077812 */ /* 0x000fe200078ec0ff */ /*0770*/ FFMA.RP R5, R13, R8, R12 ; /* 0x000000080d057223 */ /* 0x000fe2000000800c */ /*0780*/ IADD3 R8, R9, 0x20, RZ ; /* 0x0000002009087810 */ /* 0x000fe20007ffe0ff */ /*0790*/ IMAD.MOV R9, RZ, RZ, -R9 ; /* 0x000000ffff097224 */ /* 0x000fe200078e0a09 */ /*07a0*/ LOP3.LUT R7, R7, 0x800000, RZ, 0xfc, !PT ; /* 0x0080000007077812 */ /* 0x000fe400078efcff */ /*07b0*/ FSETP.NEU.FTZ.AND P0, PT, R5, R6, PT ; /* 0x000000060500720b */ /* 0x000fc40003f1d000 */ /*07c0*/ SHF.L.U32 R8, R7, R8, RZ ; /* 0x0000000807087219 */ /* 0x000fe400000006ff */ /*07d0*/ SEL R6, R9, RZ, P2 ; /* 0x000000ff09067207 */ /* 0x000fe40001000000 */ /*07e0*/ ISETP.NE.AND P1, PT, R8, RZ, P1 ; /* 0x000000ff0800720c */ /* 0x000fe40000f25270 */ /*07f0*/ SHF.R.U32.HI R6, RZ, R6, R7 ; /* 0x00000006ff067219 */ /* 0x000fe40000011607 */ /*0800*/ PLOP3.LUT P0, PT, P0, P1, PT, 0xa8, 0x0 ; /* 0x000000000000781c */ /* 0x000fe40000703570 */ /*0810*/ SHF.R.U32.HI R8, RZ, 0x1, R6 ; /* 0x00000001ff087819 */ /* 0x000fc40000011606 */ /*0820*/ SEL R5, RZ, 0x1, !P0 ; /* 0x00000001ff057807 */ /* 0x000fc80004000000 */ /*0830*/ LOP3.LUT R5, R5, 0x1, R8, 0xf8, !PT ; /* 0x0000000105057812 */ /* 0x000fc800078ef808 */ /*0840*/ LOP3.LUT R5, R5, R6, RZ, 0xc0, !PT ; /* 0x0000000605057212 */ /* 0x000fca00078ec0ff */ /*0850*/ IMAD.IADD R5, R8, 0x1, R5 ; /* 0x0000000108057824 */ /* 0x000fca00078e0205 */ /*0860*/ LOP3.LUT R4, R5, R4, RZ, 0xfc, !PT ; /* 0x0000000405047212 */ /* 0x000fe200078efcff */ /*0870*/ BRA 0x950 ; /* 0x000000d000007947 */ /* 0x000fea0003800000 */ /*0880*/ LOP3.LUT R4, R4, 0x80000000, RZ, 0xc0, !PT ; /* 0x8000000004047812 */ /* 0x000fc800078ec0ff */ /*0890*/ LOP3.LUT R4, R4, 0x7f800000, RZ, 0xfc, !PT ; /* 0x7f80000004047812 */ /* 0x000fe200078efcff */ /*08a0*/ BRA 0x950 ; /* 0x000000a000007947 */ /* 0x000fea0003800000 */ /*08b0*/ IMAD R4, R7, 0x800000, R4 ; /* 0x0080000007047824 */ /* 0x000fe200078e0204 */ /*08c0*/ BRA 0x950 ; /* 0x0000008000007947 */ /* 0x000fea0003800000 */ /*08d0*/ LOP3.LUT R4, R9, 0x80000000, R8, 0x48, !PT ; /* 0x8000000009047812 */ /* 0x000fc800078e4808 */ /*08e0*/ LOP3.LUT R4, R4, 0x7f800000, RZ, 0xfc, !PT ; /* 0x7f80000004047812 */ /* 0x000fe200078efcff */ /*08f0*/ BRA 0x950 ; /* 0x0000005000007947 */ /* 0x000fea0003800000 */ /*0900*/ LOP3.LUT R4, R9, 0x80000000, R8, 0x48, !PT ; /* 0x8000000009047812 */ /* 0x000fe200078e4808 */ /*0910*/ BRA 0x950 ; /* 0x0000003000007947 */ /* 0x000fea0003800000 */ /*0920*/ MUFU.RSQ R4, -QNAN ; /* 0xffc0000000047908 */ /* 0x000e220000001400 */ /*0930*/ BRA 0x950 ; /* 0x0000001000007947 */ /* 0x000fea0003800000 */ /*0940*/ FADD.FTZ R4, R4, R5 ; /* 0x0000000504047221 */ /* 0x000fc80000010000 */ /*0950*/ IMAD.MOV.U32 R7, RZ, RZ, R4 ; /* 0x000000ffff077224 */ /* 0x001fe400078e0004 */ /*0960*/ IMAD.MOV.U32 R4, RZ, RZ, R2 ; /* 0x000000ffff047224 */ /* 0x000fe400078e0002 */ /*0970*/ IMAD.MOV.U32 R5, RZ, RZ, 0x0 ; /* 0x00000000ff057424 */ /* 0x000fc800078e00ff */ /*0980*/ RET.REL.NODEC R4 0x0 ; /* 0xfffff67004007950 */ /* 0x000fea0003c3ffff */ /*0990*/ BRA 0x990; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*09a0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09b0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09c0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09d0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09e0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*09f0*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a00*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a10*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a20*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a30*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a40*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a50*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a60*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0a70*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z11g_One_BgradPfS_iii .globl _Z11g_One_BgradPfS_iii .p2align 8 .type _Z11g_One_BgradPfS_iii,@function _Z11g_One_BgradPfS_iii: s_clause 0x1 s_load_b64 s[4:5], s[0:1], 0x10 s_load_b64 s[6:7], s[0:1], 0x0 s_mov_b32 s2, s15 s_waitcnt lgkmcnt(0) v_mad_u64_u32 v[1:2], null, s14, s4, v[0:1] s_cmp_eq_u32 s4, 1 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mad_u64_u32 v[2:3], null, v1, s5, s[2:3] v_ashrrev_i32_e32 v3, 31, v2 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b64 v[1:2], 2, v[2:3] v_add_co_u32 v1, vcc_lo, s6, v1 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v2, vcc_lo, s7, v2, vcc_lo global_load_b32 v2, v[1:2], off v_lshl_add_u32 v1, v0, 2, 0 s_waitcnt vmcnt(0) ds_store_b32 v1, v2 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_cbranch_scc1 .LBB0_5 s_mov_b32 s3, s4 s_branch .LBB0_3 .p2align 6 .LBB0_2: s_or_b32 exec_lo, exec_lo, s6 s_cmp_lg_u32 s3, 1 s_cbranch_scc0 .LBB0_5 .LBB0_3: s_ashr_i32 s6, s3, 1 s_add_i32 s3, s3, 1 v_cmp_gt_u32_e32 vcc_lo, s6, v0 s_ashr_i32 s3, s3, 1 s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_and_saveexec_b32 s6, vcc_lo s_cbranch_execz .LBB0_2 v_add_nc_u32_e32 v2, s3, v0 s_delay_alu instid0(VALU_DEP_1) v_lshl_add_u32 v2, v2, 2, 0 ds_load_b32 v2, v2 ds_load_b32 v3, v1 s_waitcnt lgkmcnt(0) v_add_f32_e32 v2, v2, v3 ds_store_b32 v1, v2 s_branch .LBB0_2 .LBB0_5: s_waitcnt lgkmcnt(0) s_barrier buffer_gl0_inv s_mov_b32 s3, exec_lo v_cmpx_eq_u32_e32 0, v0 s_cbranch_execz .LBB0_7 v_mov_b32_e32 v0, 0 v_cvt_f32_i32_e32 v1, s4 s_load_b64 s[0:1], s[0:1], 0x8 s_mul_i32 s3, s14, s5 s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1) s_add_i32 s2, s3, s2 ds_load_b32 v0, v0 s_ashr_i32 s3, s2, 31 s_lshl_b64 s[2:3], s[2:3], 2 s_waitcnt lgkmcnt(0) s_add_u32 s0, s0, s2 s_addc_u32 s1, s1, s3 v_div_scale_f32 v2, null, v1, v1, v0 v_div_scale_f32 v5, vcc_lo, v0, v1, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_rcp_f32_e32 v3, v2 s_waitcnt_depctr 0xfff v_fma_f32 v4, -v2, v3, 1.0 v_fmac_f32_e32 v3, v4, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_mul_f32_e32 v4, v5, v3 v_fma_f32 v6, -v2, v4, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_fmac_f32_e32 v4, v6, v3 v_fma_f32 v2, -v2, v4, v5 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_div_fmas_f32 v2, v2, v3, v4 v_div_fixup_f32 v0, v2, v1, v0 v_mov_b32_e32 v1, 0 global_store_b32 v1, v0, s[0:1] .LBB0_7: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z11g_One_BgradPfS_iii .amdhsa_group_segment_fixed_size 0 .amdhsa_private_segment_fixed_size 0 .amdhsa_kernarg_size 28 .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 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 _Z11g_One_BgradPfS_iii, .Lfunc_end0-_Z11g_One_BgradPfS_iii .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 - .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: _Z11g_One_BgradPfS_iii .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z11g_One_BgradPfS_iii.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_00153560_00000000-6_g_One_Bgrad.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 _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii .type _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii, @function _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii: .LFB2051: .cfi_startproc endbr64 subq $152, %rsp .cfi_def_cfa_offset 160 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%rsp) movl %ecx, 8(%rsp) movl %r8d, 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 12(%rsp), %rax movq %rax, 112(%rsp) leaq 8(%rsp), %rax movq %rax, 120(%rsp) leaq 4(%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 _Z11g_One_BgradPfS_iii(%rip), %rdi call cudaLaunchKernel@PLT addq $16, %rsp .cfi_def_cfa_offset 160 jmp .L3 .L8: call __stack_chk_fail@PLT .cfi_endproc .LFE2051: .size _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii, .-_Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii .globl _Z11g_One_BgradPfS_iii .type _Z11g_One_BgradPfS_iii, @function _Z11g_One_BgradPfS_iii: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z36__device_stub__Z11g_One_BgradPfS_iiiPfS_iii addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z11g_One_BgradPfS_iii, .-_Z11g_One_BgradPfS_iii .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z11g_One_BgradPfS_iii" .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 _Z11g_One_BgradPfS_iii(%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 "g_One_Bgrad.hip" .globl _Z26__device_stub__g_One_BgradPfS_iii # -- Begin function _Z26__device_stub__g_One_BgradPfS_iii .p2align 4, 0x90 .type _Z26__device_stub__g_One_BgradPfS_iii,@function _Z26__device_stub__g_One_BgradPfS_iii: # @_Z26__device_stub__g_One_BgradPfS_iii .cfi_startproc # %bb.0: subq $120, %rsp .cfi_def_cfa_offset 128 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%rsp) movl %ecx, 8(%rsp) movl %r8d, 4(%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 4(%rsp), %rax movq %rax, 112(%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 $_Z11g_One_BgradPfS_iii, %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 _Z26__device_stub__g_One_BgradPfS_iii, .Lfunc_end0-_Z26__device_stub__g_One_BgradPfS_iii .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 $_Z11g_One_BgradPfS_iii, %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 _Z11g_One_BgradPfS_iii,@object # @_Z11g_One_BgradPfS_iii .section .rodata,"a",@progbits .globl _Z11g_One_BgradPfS_iii .p2align 3, 0x0 _Z11g_One_BgradPfS_iii: .quad _Z26__device_stub__g_One_BgradPfS_iii .size _Z11g_One_BgradPfS_iii, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z11g_One_BgradPfS_iii" .size .L__unnamed_1, 23 .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__g_One_BgradPfS_iii .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z11g_One_BgradPfS_iii .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.
#ifndef _GENSPARSEMAT_ #define _GENSPARSEMAT_ void GenSparseMat(float *conVec, int rows, int clms, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : input vector / flattened matrix sparseVec : sparse vector idxVec : every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : number of non-zero elements in ith row */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < rows; ++i) { nPost = 0; for(j = 0; j < clms; ++j) { if(conVec[i + clms * j]) { /* i --> j */ sparseVec[counter] = j; counter += 1; nPost += 1; } } nPostNeurons[i] = nPost; } idxVec[0] = 0; for(i = 1; i < rows; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } void GenSparseFeedForwardMat(float *conVec, int nff, int nNeurons, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : nff*nNeurons-by-1 - input vector / flattened matrix sparseVec : - sparse vector idxVec : nff-by-1 - every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : nff-by-1 - number of 2/3 neurons reciving connections from */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < nff; ++i) { nPost = 0; for(j = 0; j < nNeurons; ++j) { if(conVec[i + j * nff]) { sparseVec[counter] = j; counter += 1; nPost += 1; } nPostNeurons[i] = nPost; } } idxVec[0] = 0; for(i = 1; i < nff; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } #endif
code for sm_80
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#ifndef _GENSPARSEMAT_ #define _GENSPARSEMAT_ void GenSparseMat(float *conVec, int rows, int clms, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : input vector / flattened matrix sparseVec : sparse vector idxVec : every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : number of non-zero elements in ith row */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < rows; ++i) { nPost = 0; for(j = 0; j < clms; ++j) { if(conVec[i + clms * j]) { /* i --> j */ sparseVec[counter] = j; counter += 1; nPost += 1; } } nPostNeurons[i] = nPost; } idxVec[0] = 0; for(i = 1; i < rows; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } void GenSparseFeedForwardMat(float *conVec, int nff, int nNeurons, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : nff*nNeurons-by-1 - input vector / flattened matrix sparseVec : - sparse vector idxVec : nff-by-1 - every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : nff-by-1 - number of 2/3 neurons reciving connections from */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < nff; ++i) { nPost = 0; for(j = 0; j < nNeurons; ++j) { if(conVec[i + j * nff]) { sparseVec[counter] = j; counter += 1; nPost += 1; } nPostNeurons[i] = nPost; } } idxVec[0] = 0; for(i = 1; i < nff; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } #endif
.file "tmpxft_000afbb4_00000000-6_GenSparseMat.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2031: .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 .LFE2031: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z12GenSparseMatPfiiPiS0_S0_ .type _Z12GenSparseMatPfiiPiS0_S0_, @function _Z12GenSparseMatPfiiPiS0_S0_: .LFB2027: .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 movq %r8, %rbx movq %r9, %rbp movslq %esi, %r9 testq %r9, %r9 je .L4 movq %rdi, %r12 movq %rcx, %r11 movslq %edx, %rdi leaq 0(,%rdi,4), %r10 movl $0, %ecx movl $0, %r8d pxor %xmm0, %xmm0 jmp .L5 .L14: movl %edx, (%r11,%rcx,4) addq $1, %rcx addq $1, %rsi .L6: addq $1, %rdx addq %r10, %rax cmpq %rdi, %rdx je .L10 .L8: ucomiss (%rax), %xmm0 jp .L14 je .L6 jmp .L14 .L10: movl %esi, 0(%rbp,%r8,4) addq $1, %r8 addq $4, %r12 cmpq %r9, %r8 je .L9 .L5: movq %r12, %rax movl $0, %esi movl $0, %edx testq %rdi, %rdi jne .L8 movq %rdi, %rsi jmp .L10 .L9: movl $0, (%rbx) cmpq $1, %r9 jbe .L3 movl $1, %eax .L12: movl -4(%rbp,%rax,4), %edx addl -4(%rbx,%rax,4), %edx movl %edx, (%rbx,%rax,4) addq $1, %rax cmpq %r9, %rax jne .L12 .L3: popq %rbx .cfi_remember_state .cfi_def_cfa_offset 24 popq %rbp .cfi_def_cfa_offset 16 popq %r12 .cfi_def_cfa_offset 8 ret .L4: .cfi_restore_state movl $0, (%r8) jmp .L3 .cfi_endproc .LFE2027: .size _Z12GenSparseMatPfiiPiS0_S0_, .-_Z12GenSparseMatPfiiPiS0_S0_ .globl _Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .type _Z23GenSparseFeedForwardMatPfiiPiS0_S0_, @function _Z23GenSparseFeedForwardMatPfiiPiS0_S0_: .LFB2028: .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 movq %r8, %rbp movslq %esi, %r8 testq %r8, %r8 je .L20 movq %rdi, %r13 movl %edx, %r12d movq %rcx, %rbx movq %r9, %rdi leaq 0(,%r8,4), %r11 movl $0, %esi movl $0, %r14d movslq %edx, %r10 pxor %xmm0, %xmm0 jmp .L21 .L30: movl %edx, (%rbx,%rsi,4) addq $1, %rsi addq $1, %rcx .L22: movl %ecx, (%rdi) addq $1, %rdx addq %r11, %rax cmpq %r10, %rdx je .L26 .L24: ucomiss (%rax), %xmm0 jp .L30 je .L22 jmp .L30 .L26: addq $1, %r14 addq $4, %rdi addq $4, %r13 cmpq %r8, %r14 je .L25 .L21: movq %r13, %rax movl $0, %ecx movl $0, %edx testl %r12d, %r12d jne .L24 jmp .L26 .L25: movl $0, 0(%rbp) cmpq $1, %r8 jbe .L19 movl $1, %eax .L28: movl -4(%r9,%rax,4), %edx addl -4(%rbp,%rax,4), %edx movl %edx, 0(%rbp,%rax,4) addq $1, %rax cmpq %r8, %rax jne .L28 .L19: popq %rbx .cfi_remember_state .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 .L20: .cfi_restore_state movl $0, 0(%rbp) jmp .L19 .cfi_endproc .LFE2028: .size _Z23GenSparseFeedForwardMatPfiiPiS0_S0_, .-_Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .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) 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.
#ifndef _GENSPARSEMAT_ #define _GENSPARSEMAT_ void GenSparseMat(float *conVec, int rows, int clms, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : input vector / flattened matrix sparseVec : sparse vector idxVec : every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : number of non-zero elements in ith row */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < rows; ++i) { nPost = 0; for(j = 0; j < clms; ++j) { if(conVec[i + clms * j]) { /* i --> j */ sparseVec[counter] = j; counter += 1; nPost += 1; } } nPostNeurons[i] = nPost; } idxVec[0] = 0; for(i = 1; i < rows; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } void GenSparseFeedForwardMat(float *conVec, int nff, int nNeurons, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : nff*nNeurons-by-1 - input vector / flattened matrix sparseVec : - sparse vector idxVec : nff-by-1 - every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : nff-by-1 - number of 2/3 neurons reciving connections from */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < nff; ++i) { nPost = 0; for(j = 0; j < nNeurons; ++j) { if(conVec[i + j * nff]) { sparseVec[counter] = j; counter += 1; nPost += 1; } nPostNeurons[i] = nPost; } } idxVec[0] = 0; for(i = 1; i < nff; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } #endif
#ifndef _GENSPARSEMAT_ #define _GENSPARSEMAT_ void GenSparseMat(float *conVec, int rows, int clms, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : input vector / flattened matrix sparseVec : sparse vector idxVec : every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : number of non-zero elements in ith row */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < rows; ++i) { nPost = 0; for(j = 0; j < clms; ++j) { if(conVec[i + clms * j]) { /* i --> j */ sparseVec[counter] = j; counter += 1; nPost += 1; } } nPostNeurons[i] = nPost; } idxVec[0] = 0; for(i = 1; i < rows; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } void GenSparseFeedForwardMat(float *conVec, int nff, int nNeurons, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : nff*nNeurons-by-1 - input vector / flattened matrix sparseVec : - sparse vector idxVec : nff-by-1 - every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : nff-by-1 - number of 2/3 neurons reciving connections from */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < nff; ++i) { nPost = 0; for(j = 0; j < nNeurons; ++j) { if(conVec[i + j * nff]) { sparseVec[counter] = j; counter += 1; nPost += 1; } nPostNeurons[i] = nPost; } } idxVec[0] = 0; for(i = 1; i < nff; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } #endif
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#ifndef _GENSPARSEMAT_ #define _GENSPARSEMAT_ void GenSparseMat(float *conVec, int rows, int clms, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : input vector / flattened matrix sparseVec : sparse vector idxVec : every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : number of non-zero elements in ith row */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < rows; ++i) { nPost = 0; for(j = 0; j < clms; ++j) { if(conVec[i + clms * j]) { /* i --> j */ sparseVec[counter] = j; counter += 1; nPost += 1; } } nPostNeurons[i] = nPost; } idxVec[0] = 0; for(i = 1; i < rows; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } void GenSparseFeedForwardMat(float *conVec, int nff, int nNeurons, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : nff*nNeurons-by-1 - input vector / flattened matrix sparseVec : - sparse vector idxVec : nff-by-1 - every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : nff-by-1 - number of 2/3 neurons reciving connections from */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < nff; ++i) { nPost = 0; for(j = 0; j < nNeurons; ++j) { if(conVec[i + j * nff]) { sparseVec[counter] = j; counter += 1; nPost += 1; } nPostNeurons[i] = nPost; } } idxVec[0] = 0; for(i = 1; i < nff; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } #endif
.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.
#ifndef _GENSPARSEMAT_ #define _GENSPARSEMAT_ void GenSparseMat(float *conVec, int rows, int clms, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : input vector / flattened matrix sparseVec : sparse vector idxVec : every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : number of non-zero elements in ith row */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < rows; ++i) { nPost = 0; for(j = 0; j < clms; ++j) { if(conVec[i + clms * j]) { /* i --> j */ sparseVec[counter] = j; counter += 1; nPost += 1; } } nPostNeurons[i] = nPost; } idxVec[0] = 0; for(i = 1; i < rows; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } void GenSparseFeedForwardMat(float *conVec, int nff, int nNeurons, int* sparseVec, int* idxVec, int* nPostNeurons ) { /* generate sparse representation conVec : nff*nNeurons-by-1 - input vector / flattened matrix sparseVec : - sparse vector idxVec : nff-by-1 - every element is the starting index in sparseVec for ith row in matrix conVec nPostNeurons : nff-by-1 - number of 2/3 neurons reciving connections from */ unsigned long long int i, j, counter = 0, nPost; for(i = 0; i < nff; ++i) { nPost = 0; for(j = 0; j < nNeurons; ++j) { if(conVec[i + j * nff]) { sparseVec[counter] = j; counter += 1; nPost += 1; } nPostNeurons[i] = nPost; } } idxVec[0] = 0; for(i = 1; i < nff; ++i) { idxVec[i] = idxVec[i-1] + nPostNeurons[i-1]; } } #endif
.text .file "GenSparseMat.hip" .globl _Z12GenSparseMatPfiiPiS0_S0_ # -- Begin function _Z12GenSparseMatPfiiPiS0_S0_ .p2align 4, 0x90 .type _Z12GenSparseMatPfiiPiS0_S0_,@function _Z12GenSparseMatPfiiPiS0_S0_: # @_Z12GenSparseMatPfiiPiS0_S0_ .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 .cfi_offset %rbx, -48 .cfi_offset %r12, -40 .cfi_offset %r13, -32 .cfi_offset %r14, -24 .cfi_offset %r15, -16 movslq %esi, %rax testl %eax, %eax je .LBB0_9 # %bb.1: # %.preheader.lr.ph movslq %edx, %r10 leaq (,%r10,4), %r11 xorl %ebx, %ebx xorps %xmm0, %xmm0 xorl %r14d, %r14d jmp .LBB0_2 .p2align 4, 0x90 .LBB0_3: # in Loop: Header=BB0_2 Depth=1 xorl %r15d, %r15d .LBB0_8: # %._crit_edge # in Loop: Header=BB0_2 Depth=1 movl %r15d, (%r9,%r14,4) incq %r14 addq $4, %rdi cmpq %rax, %r14 je .LBB0_9 .LBB0_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB0_5 Depth 2 testl %edx, %edx je .LBB0_3 # %bb.4: # %.lr.ph # in Loop: Header=BB0_2 Depth=1 movq %rdi, %r12 xorl %r15d, %r15d xorl %r13d, %r13d jmp .LBB0_5 .p2align 4, 0x90 .LBB0_7: # in Loop: Header=BB0_5 Depth=2 incq %r13 addq %r11, %r12 cmpq %r13, %r10 je .LBB0_8 .LBB0_5: # Parent Loop BB0_2 Depth=1 # => This Inner Loop Header: Depth=2 movss (%r12), %xmm1 # xmm1 = mem[0],zero,zero,zero ucomiss %xmm0, %xmm1 jne .LBB0_6 jnp .LBB0_7 .LBB0_6: # in Loop: Header=BB0_5 Depth=2 movl %r13d, (%rcx,%rbx,4) incq %rbx incq %r15 jmp .LBB0_7 .LBB0_9: # %._crit_edge40 movl $0, (%r8) cmpl $2, %esi jb .LBB0_12 # %bb.10: # %.lr.ph44.preheader movl (%r8), %ecx decq %rax xorl %edx, %edx .p2align 4, 0x90 .LBB0_11: # %.lr.ph44 # =>This Inner Loop Header: Depth=1 addl (%r9,%rdx,4), %ecx movl %ecx, 4(%r8,%rdx,4) incq %rdx cmpq %rdx, %rax jne .LBB0_11 .LBB0_12: # %._crit_edge45 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_end0: .size _Z12GenSparseMatPfiiPiS0_S0_, .Lfunc_end0-_Z12GenSparseMatPfiiPiS0_S0_ .cfi_endproc # -- End function .globl _Z23GenSparseFeedForwardMatPfiiPiS0_S0_ # -- Begin function _Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .p2align 4, 0x90 .type _Z23GenSparseFeedForwardMatPfiiPiS0_S0_,@function _Z23GenSparseFeedForwardMatPfiiPiS0_S0_: # @_Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .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 .cfi_offset %rbx, -48 .cfi_offset %r12, -40 .cfi_offset %r13, -32 .cfi_offset %r14, -24 .cfi_offset %r15, -16 movslq %esi, %rax testl %eax, %eax je .LBB1_8 # %bb.1: # %.preheader.lr.ph movslq %edx, %r10 leaq (,%rax,4), %r11 xorl %ebx, %ebx xorps %xmm0, %xmm0 xorl %r14d, %r14d jmp .LBB1_2 .p2align 4, 0x90 .LBB1_7: # %._crit_edge # in Loop: Header=BB1_2 Depth=1 incq %r14 addq $4, %rdi cmpq %rax, %r14 je .LBB1_8 .LBB1_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB1_4 Depth 2 testl %edx, %edx je .LBB1_7 # %bb.3: # %.lr.ph # in Loop: Header=BB1_2 Depth=1 movq %rdi, %r15 xorl %r12d, %r12d xorl %r13d, %r13d jmp .LBB1_4 .p2align 4, 0x90 .LBB1_6: # in Loop: Header=BB1_4 Depth=2 movl %r12d, (%r9,%r14,4) incq %r13 addq %r11, %r15 cmpq %r13, %r10 je .LBB1_7 .LBB1_4: # Parent Loop BB1_2 Depth=1 # => This Inner Loop Header: Depth=2 movss (%r15), %xmm1 # xmm1 = mem[0],zero,zero,zero ucomiss %xmm0, %xmm1 jne .LBB1_5 jnp .LBB1_6 .LBB1_5: # in Loop: Header=BB1_4 Depth=2 movl %r13d, (%rcx,%rbx,4) incq %rbx incq %r12 jmp .LBB1_6 .LBB1_8: # %._crit_edge39 movl $0, (%r8) cmpl $2, %esi jb .LBB1_11 # %bb.9: # %.lr.ph43.preheader movl (%r8), %ecx decq %rax xorl %edx, %edx .p2align 4, 0x90 .LBB1_10: # %.lr.ph43 # =>This Inner Loop Header: Depth=1 addl (%r9,%rdx,4), %ecx movl %ecx, 4(%r8,%rdx,4) incq %rdx cmpq %rdx, %rax jne .LBB1_10 .LBB1_11: # %._crit_edge44 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 _Z23GenSparseFeedForwardMatPfiiPiS0_S0_, .Lfunc_end1-_Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .cfi_endproc # -- End function .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_000afbb4_00000000-6_GenSparseMat.cudafe1.cpp" .text #APP #NO_APP .type _ZL26__cudaUnregisterBinaryUtilv, @function _ZL26__cudaUnregisterBinaryUtilv: .LFB2031: .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 .LFE2031: .size _ZL26__cudaUnregisterBinaryUtilv, .-_ZL26__cudaUnregisterBinaryUtilv .globl _Z12GenSparseMatPfiiPiS0_S0_ .type _Z12GenSparseMatPfiiPiS0_S0_, @function _Z12GenSparseMatPfiiPiS0_S0_: .LFB2027: .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 movq %r8, %rbx movq %r9, %rbp movslq %esi, %r9 testq %r9, %r9 je .L4 movq %rdi, %r12 movq %rcx, %r11 movslq %edx, %rdi leaq 0(,%rdi,4), %r10 movl $0, %ecx movl $0, %r8d pxor %xmm0, %xmm0 jmp .L5 .L14: movl %edx, (%r11,%rcx,4) addq $1, %rcx addq $1, %rsi .L6: addq $1, %rdx addq %r10, %rax cmpq %rdi, %rdx je .L10 .L8: ucomiss (%rax), %xmm0 jp .L14 je .L6 jmp .L14 .L10: movl %esi, 0(%rbp,%r8,4) addq $1, %r8 addq $4, %r12 cmpq %r9, %r8 je .L9 .L5: movq %r12, %rax movl $0, %esi movl $0, %edx testq %rdi, %rdi jne .L8 movq %rdi, %rsi jmp .L10 .L9: movl $0, (%rbx) cmpq $1, %r9 jbe .L3 movl $1, %eax .L12: movl -4(%rbp,%rax,4), %edx addl -4(%rbx,%rax,4), %edx movl %edx, (%rbx,%rax,4) addq $1, %rax cmpq %r9, %rax jne .L12 .L3: popq %rbx .cfi_remember_state .cfi_def_cfa_offset 24 popq %rbp .cfi_def_cfa_offset 16 popq %r12 .cfi_def_cfa_offset 8 ret .L4: .cfi_restore_state movl $0, (%r8) jmp .L3 .cfi_endproc .LFE2027: .size _Z12GenSparseMatPfiiPiS0_S0_, .-_Z12GenSparseMatPfiiPiS0_S0_ .globl _Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .type _Z23GenSparseFeedForwardMatPfiiPiS0_S0_, @function _Z23GenSparseFeedForwardMatPfiiPiS0_S0_: .LFB2028: .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 movq %r8, %rbp movslq %esi, %r8 testq %r8, %r8 je .L20 movq %rdi, %r13 movl %edx, %r12d movq %rcx, %rbx movq %r9, %rdi leaq 0(,%r8,4), %r11 movl $0, %esi movl $0, %r14d movslq %edx, %r10 pxor %xmm0, %xmm0 jmp .L21 .L30: movl %edx, (%rbx,%rsi,4) addq $1, %rsi addq $1, %rcx .L22: movl %ecx, (%rdi) addq $1, %rdx addq %r11, %rax cmpq %r10, %rdx je .L26 .L24: ucomiss (%rax), %xmm0 jp .L30 je .L22 jmp .L30 .L26: addq $1, %r14 addq $4, %rdi addq $4, %r13 cmpq %r8, %r14 je .L25 .L21: movq %r13, %rax movl $0, %ecx movl $0, %edx testl %r12d, %r12d jne .L24 jmp .L26 .L25: movl $0, 0(%rbp) cmpq $1, %r8 jbe .L19 movl $1, %eax .L28: movl -4(%r9,%rax,4), %edx addl -4(%rbp,%rax,4), %edx movl %edx, 0(%rbp,%rax,4) addq $1, %rax cmpq %r8, %rax jne .L28 .L19: popq %rbx .cfi_remember_state .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 .L20: .cfi_restore_state movl $0, 0(%rbp) jmp .L19 .cfi_endproc .LFE2028: .size _Z23GenSparseFeedForwardMatPfiiPiS0_S0_, .-_Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .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) 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 "GenSparseMat.hip" .globl _Z12GenSparseMatPfiiPiS0_S0_ # -- Begin function _Z12GenSparseMatPfiiPiS0_S0_ .p2align 4, 0x90 .type _Z12GenSparseMatPfiiPiS0_S0_,@function _Z12GenSparseMatPfiiPiS0_S0_: # @_Z12GenSparseMatPfiiPiS0_S0_ .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 .cfi_offset %rbx, -48 .cfi_offset %r12, -40 .cfi_offset %r13, -32 .cfi_offset %r14, -24 .cfi_offset %r15, -16 movslq %esi, %rax testl %eax, %eax je .LBB0_9 # %bb.1: # %.preheader.lr.ph movslq %edx, %r10 leaq (,%r10,4), %r11 xorl %ebx, %ebx xorps %xmm0, %xmm0 xorl %r14d, %r14d jmp .LBB0_2 .p2align 4, 0x90 .LBB0_3: # in Loop: Header=BB0_2 Depth=1 xorl %r15d, %r15d .LBB0_8: # %._crit_edge # in Loop: Header=BB0_2 Depth=1 movl %r15d, (%r9,%r14,4) incq %r14 addq $4, %rdi cmpq %rax, %r14 je .LBB0_9 .LBB0_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB0_5 Depth 2 testl %edx, %edx je .LBB0_3 # %bb.4: # %.lr.ph # in Loop: Header=BB0_2 Depth=1 movq %rdi, %r12 xorl %r15d, %r15d xorl %r13d, %r13d jmp .LBB0_5 .p2align 4, 0x90 .LBB0_7: # in Loop: Header=BB0_5 Depth=2 incq %r13 addq %r11, %r12 cmpq %r13, %r10 je .LBB0_8 .LBB0_5: # Parent Loop BB0_2 Depth=1 # => This Inner Loop Header: Depth=2 movss (%r12), %xmm1 # xmm1 = mem[0],zero,zero,zero ucomiss %xmm0, %xmm1 jne .LBB0_6 jnp .LBB0_7 .LBB0_6: # in Loop: Header=BB0_5 Depth=2 movl %r13d, (%rcx,%rbx,4) incq %rbx incq %r15 jmp .LBB0_7 .LBB0_9: # %._crit_edge40 movl $0, (%r8) cmpl $2, %esi jb .LBB0_12 # %bb.10: # %.lr.ph44.preheader movl (%r8), %ecx decq %rax xorl %edx, %edx .p2align 4, 0x90 .LBB0_11: # %.lr.ph44 # =>This Inner Loop Header: Depth=1 addl (%r9,%rdx,4), %ecx movl %ecx, 4(%r8,%rdx,4) incq %rdx cmpq %rdx, %rax jne .LBB0_11 .LBB0_12: # %._crit_edge45 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_end0: .size _Z12GenSparseMatPfiiPiS0_S0_, .Lfunc_end0-_Z12GenSparseMatPfiiPiS0_S0_ .cfi_endproc # -- End function .globl _Z23GenSparseFeedForwardMatPfiiPiS0_S0_ # -- Begin function _Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .p2align 4, 0x90 .type _Z23GenSparseFeedForwardMatPfiiPiS0_S0_,@function _Z23GenSparseFeedForwardMatPfiiPiS0_S0_: # @_Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .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 .cfi_offset %rbx, -48 .cfi_offset %r12, -40 .cfi_offset %r13, -32 .cfi_offset %r14, -24 .cfi_offset %r15, -16 movslq %esi, %rax testl %eax, %eax je .LBB1_8 # %bb.1: # %.preheader.lr.ph movslq %edx, %r10 leaq (,%rax,4), %r11 xorl %ebx, %ebx xorps %xmm0, %xmm0 xorl %r14d, %r14d jmp .LBB1_2 .p2align 4, 0x90 .LBB1_7: # %._crit_edge # in Loop: Header=BB1_2 Depth=1 incq %r14 addq $4, %rdi cmpq %rax, %r14 je .LBB1_8 .LBB1_2: # %.preheader # =>This Loop Header: Depth=1 # Child Loop BB1_4 Depth 2 testl %edx, %edx je .LBB1_7 # %bb.3: # %.lr.ph # in Loop: Header=BB1_2 Depth=1 movq %rdi, %r15 xorl %r12d, %r12d xorl %r13d, %r13d jmp .LBB1_4 .p2align 4, 0x90 .LBB1_6: # in Loop: Header=BB1_4 Depth=2 movl %r12d, (%r9,%r14,4) incq %r13 addq %r11, %r15 cmpq %r13, %r10 je .LBB1_7 .LBB1_4: # Parent Loop BB1_2 Depth=1 # => This Inner Loop Header: Depth=2 movss (%r15), %xmm1 # xmm1 = mem[0],zero,zero,zero ucomiss %xmm0, %xmm1 jne .LBB1_5 jnp .LBB1_6 .LBB1_5: # in Loop: Header=BB1_4 Depth=2 movl %r13d, (%rcx,%rbx,4) incq %rbx incq %r12 jmp .LBB1_6 .LBB1_8: # %._crit_edge39 movl $0, (%r8) cmpl $2, %esi jb .LBB1_11 # %bb.9: # %.lr.ph43.preheader movl (%r8), %ecx decq %rax xorl %edx, %edx .p2align 4, 0x90 .LBB1_10: # %.lr.ph43 # =>This Inner Loop Header: Depth=1 addl (%r9,%rdx,4), %ecx movl %ecx, 4(%r8,%rdx,4) incq %rdx cmpq %rdx, %rax jne .LBB1_10 .LBB1_11: # %._crit_edge44 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 _Z23GenSparseFeedForwardMatPfiiPiS0_S0_, .Lfunc_end1-_Z23GenSparseFeedForwardMatPfiiPiS0_S0_ .cfi_endproc # -- End function .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.
/******************************************************************************* GPU OPTIMIZED MONTE CARLO (GOMC) 2.75 Copyright (C) 2022 GOMC Group A copy of the MIT License can be found in License.txt along with this program, also can be found at <https://opensource.org/licenses/MIT>. ********************************************************************************/ #include "CUDAMemoryManager.cuh" #ifdef GOMC_CUDA long long CUDAMemoryManager::totalAllocatedBytes = 0; std::unordered_map<void *, std::pair<unsigned int, std::string> > CUDAMemoryManager::allocatedPointers; cudaError_t CUDAMemoryManager::mallocMemory(void **address, unsigned int size, std::string var_name) { cudaError_t ret = cudaMalloc(address, size); allocatedPointers[*address] = make_pair(size, var_name); totalAllocatedBytes += size; if (size == 0) { std::cout << "Warning! You are trying to allocate " << var_name << " with a size of zero bytes!\n"; } return ret; } cudaError_t CUDAMemoryManager::freeMemory(void *address, std::string var_name) { if(allocatedPointers.find(address) != allocatedPointers.end()) { totalAllocatedBytes -= allocatedPointers[address].first; allocatedPointers.erase(address); } else if (address != nullptr) { std::cout << "Warning! You are trying to free " << var_name << " but it has already been freed\n" << "\tor was never allocated!\n"; } return cudaFree(address); } bool CUDAMemoryManager::isFreed() { bool ret = allocatedPointers.size() == 0; while(allocatedPointers.size() != 0) { auto it = allocatedPointers.begin(); std::cout << "You forgot to free memory " << it->second.second << " with " << it->second.first << " bytes allocated!\n"; std::cout << "I am going to free it for you!\n"; freeMemory(it->first, it->second.second); } return ret; } #endif
code for sm_80
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
/******************************************************************************* GPU OPTIMIZED MONTE CARLO (GOMC) 2.75 Copyright (C) 2022 GOMC Group A copy of the MIT License can be found in License.txt along with this program, also can be found at <https://opensource.org/licenses/MIT>. ********************************************************************************/ #include "CUDAMemoryManager.cuh" #ifdef GOMC_CUDA long long CUDAMemoryManager::totalAllocatedBytes = 0; std::unordered_map<void *, std::pair<unsigned int, std::string> > CUDAMemoryManager::allocatedPointers; cudaError_t CUDAMemoryManager::mallocMemory(void **address, unsigned int size, std::string var_name) { cudaError_t ret = cudaMalloc(address, size); allocatedPointers[*address] = make_pair(size, var_name); totalAllocatedBytes += size; if (size == 0) { std::cout << "Warning! You are trying to allocate " << var_name << " with a size of zero bytes!\n"; } return ret; } cudaError_t CUDAMemoryManager::freeMemory(void *address, std::string var_name) { if(allocatedPointers.find(address) != allocatedPointers.end()) { totalAllocatedBytes -= allocatedPointers[address].first; allocatedPointers.erase(address); } else if (address != nullptr) { std::cout << "Warning! You are trying to free " << var_name << " but it has already been freed\n" << "\tor was never allocated!\n"; } return cudaFree(address); } bool CUDAMemoryManager::isFreed() { bool ret = allocatedPointers.size() == 0; while(allocatedPointers.size() != 0) { auto it = allocatedPointers.begin(); std::cout << "You forgot to free memory " << it->second.second << " with " << it->second.first << " bytes allocated!\n"; std::cout << "I am going to free it for you!\n"; freeMemory(it->first, it->second.second); } return ret; } #endif
.file "tmpxft_00187404_00000000-6_CUDAMemoryManager.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 .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2052: .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 .LFE2052: .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.
/******************************************************************************* GPU OPTIMIZED MONTE CARLO (GOMC) 2.75 Copyright (C) 2022 GOMC Group A copy of the MIT License can be found in License.txt along with this program, also can be found at <https://opensource.org/licenses/MIT>. ********************************************************************************/ #include "CUDAMemoryManager.cuh" #ifdef GOMC_CUDA long long CUDAMemoryManager::totalAllocatedBytes = 0; std::unordered_map<void *, std::pair<unsigned int, std::string> > CUDAMemoryManager::allocatedPointers; cudaError_t CUDAMemoryManager::mallocMemory(void **address, unsigned int size, std::string var_name) { cudaError_t ret = cudaMalloc(address, size); allocatedPointers[*address] = make_pair(size, var_name); totalAllocatedBytes += size; if (size == 0) { std::cout << "Warning! You are trying to allocate " << var_name << " with a size of zero bytes!\n"; } return ret; } cudaError_t CUDAMemoryManager::freeMemory(void *address, std::string var_name) { if(allocatedPointers.find(address) != allocatedPointers.end()) { totalAllocatedBytes -= allocatedPointers[address].first; allocatedPointers.erase(address); } else if (address != nullptr) { std::cout << "Warning! You are trying to free " << var_name << " but it has already been freed\n" << "\tor was never allocated!\n"; } return cudaFree(address); } bool CUDAMemoryManager::isFreed() { bool ret = allocatedPointers.size() == 0; while(allocatedPointers.size() != 0) { auto it = allocatedPointers.begin(); std::cout << "You forgot to free memory " << it->second.second << " with " << it->second.first << " bytes allocated!\n"; std::cout << "I am going to free it for you!\n"; freeMemory(it->first, it->second.second); } return ret; } #endif
/******************************************************************************* GPU OPTIMIZED MONTE CARLO (GOMC) 2.75 Copyright (C) 2022 GOMC Group A copy of the MIT License can be found in License.txt along with this program, also can be found at <https://opensource.org/licenses/MIT>. ********************************************************************************/ #pragma once #ifdef GOMC_CUDA #include <hip/hip_runtime.h> #include <unordered_map> #include <iostream> #define CUMALLOC(address,size) CUDAMemoryManager::mallocMemory(address,size,#address) #define CUFREE(address) CUDAMemoryManager::freeMemory(address,#address) class CUDAMemoryManager { public: static hipError_t mallocMemory(void **address, unsigned int size, std::string var_name); static hipError_t freeMemory(void *address, std::string var_name); static bool isFreed(); private: static long long totalAllocatedBytes; static std::unordered_map<void *, std::pair<unsigned int, std::string> > allocatedPointers; }; #endif
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
/******************************************************************************* GPU OPTIMIZED MONTE CARLO (GOMC) 2.75 Copyright (C) 2022 GOMC Group A copy of the MIT License can be found in License.txt along with this program, also can be found at <https://opensource.org/licenses/MIT>. ********************************************************************************/ #pragma once #ifdef GOMC_CUDA #include <hip/hip_runtime.h> #include <unordered_map> #include <iostream> #define CUMALLOC(address,size) CUDAMemoryManager::mallocMemory(address,size,#address) #define CUFREE(address) CUDAMemoryManager::freeMemory(address,#address) class CUDAMemoryManager { public: static hipError_t mallocMemory(void **address, unsigned int size, std::string var_name); static hipError_t freeMemory(void *address, std::string var_name); static bool isFreed(); private: static long long totalAllocatedBytes; static std::unordered_map<void *, std::pair<unsigned int, std::string> > allocatedPointers; }; #endif
.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.
/******************************************************************************* GPU OPTIMIZED MONTE CARLO (GOMC) 2.75 Copyright (C) 2022 GOMC Group A copy of the MIT License can be found in License.txt along with this program, also can be found at <https://opensource.org/licenses/MIT>. ********************************************************************************/ #pragma once #ifdef GOMC_CUDA #include <hip/hip_runtime.h> #include <unordered_map> #include <iostream> #define CUMALLOC(address,size) CUDAMemoryManager::mallocMemory(address,size,#address) #define CUFREE(address) CUDAMemoryManager::freeMemory(address,#address) class CUDAMemoryManager { public: static hipError_t mallocMemory(void **address, unsigned int size, std::string var_name); static hipError_t freeMemory(void *address, std::string var_name); static bool isFreed(); private: static long long totalAllocatedBytes; static std::unordered_map<void *, std::pair<unsigned int, std::string> > allocatedPointers; }; #endif
.text .file "CUDAMemoryManager.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_00187404_00000000-6_CUDAMemoryManager.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 .type _ZL24__sti____cudaRegisterAllv, @function _ZL24__sti____cudaRegisterAllv: .LFB2052: .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 .LFE2052: .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 "CUDAMemoryManager.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.
#include "includes.h" __global__ void VecAdd(double* A,double* B,double* C) { // extern __shared__ float sdata[]; int i=threadIdx.x; C[i]=A[i]+B[i]; }
code for sm_80 Function : _Z6VecAddPdS_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 R8, SR_TID.X ; /* 0x0000000000087919 */ /* 0x000e220000002100 */ /*0020*/ HFMA2.MMA R9, -RZ, RZ, 0, 4.76837158203125e-07 ; /* 0x00000008ff097435 */ /* 0x000fe200000001ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd20000000a00 */ /*0040*/ IMAD.WIDE R2, R8, R9, c[0x0][0x160] ; /* 0x0000580008027625 */ /* 0x001fc800078e0209 */ /*0050*/ IMAD.WIDE R4, R8.reuse, R9.reuse, c[0x0][0x168] ; /* 0x00005a0008047625 */ /* 0x0c0fe400078e0209 */ /*0060*/ LDG.E.64 R2, [R2.64] ; /* 0x0000000402027981 */ /* 0x000ea8000c1e1b00 */ /*0070*/ LDG.E.64 R4, [R4.64] ; /* 0x0000000404047981 */ /* 0x000ea2000c1e1b00 */ /*0080*/ IMAD.WIDE R8, R8, R9, c[0x0][0x170] ; /* 0x00005c0008087625 */ /* 0x000fe200078e0209 */ /*0090*/ DADD R6, R2, R4 ; /* 0x0000000002067229 */ /* 0x004e0e0000000004 */ /*00a0*/ STG.E.64 [R8.64], R6 ; /* 0x0000000608007986 */ /* 0x001fe2000c101b04 */ /*00b0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00c0*/ BRA 0xc0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*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 VecAdd(double* A,double* B,double* C) { // extern __shared__ float sdata[]; int i=threadIdx.x; C[i]=A[i]+B[i]; }
.file "tmpxft_000f5fc6_00000000-6_VecAdd.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 _Z29__device_stub__Z6VecAddPdS_S_PdS_S_ .type _Z29__device_stub__Z6VecAddPdS_S_PdS_S_, @function _Z29__device_stub__Z6VecAddPdS_S_PdS_S_: .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 _Z6VecAddPdS_S_(%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 _Z29__device_stub__Z6VecAddPdS_S_PdS_S_, .-_Z29__device_stub__Z6VecAddPdS_S_PdS_S_ .globl _Z6VecAddPdS_S_ .type _Z6VecAddPdS_S_, @function _Z6VecAddPdS_S_: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z6VecAddPdS_S_PdS_S_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z6VecAddPdS_S_, .-_Z6VecAddPdS_S_ .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z6VecAddPdS_S_" .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 _Z6VecAddPdS_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 .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 VecAdd(double* A,double* B,double* C) { // extern __shared__ float sdata[]; int i=threadIdx.x; C[i]=A[i]+B[i]; }
#include <hip/hip_runtime.h> #include "includes.h" __global__ void VecAdd(double* A,double* B,double* C) { // extern __shared__ float sdata[]; int i=threadIdx.x; C[i]=A[i]+B[i]; }
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 VecAdd(double* A,double* B,double* C) { // extern __shared__ float sdata[]; int i=threadIdx.x; C[i]=A[i]+B[i]; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z6VecAddPdS_S_ .globl _Z6VecAddPdS_S_ .p2align 8 .type _Z6VecAddPdS_S_,@function _Z6VecAddPdS_S_: s_load_b128 s[4:7], s[0:1], 0x0 v_lshlrev_b32_e32 v4, 3, v0 s_load_b64 s[0:1], s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_clause 0x1 global_load_b64 v[0:1], v4, s[4:5] global_load_b64 v[2:3], v4, s[6:7] s_waitcnt vmcnt(0) v_add_f64 v[0:1], v[0:1], v[2:3] global_store_b64 v4, v[0:1], s[0:1] s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z6VecAddPdS_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 5 .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 _Z6VecAddPdS_S_, .Lfunc_end0-_Z6VecAddPdS_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: _Z6VecAddPdS_S_ .private_segment_fixed_size: 0 .sgpr_count: 8 .sgpr_spill_count: 0 .symbol: _Z6VecAddPdS_S_.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 5 .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 VecAdd(double* A,double* B,double* C) { // extern __shared__ float sdata[]; int i=threadIdx.x; C[i]=A[i]+B[i]; }
.text .file "VecAdd.hip" .globl _Z21__device_stub__VecAddPdS_S_ # -- Begin function _Z21__device_stub__VecAddPdS_S_ .p2align 4, 0x90 .type _Z21__device_stub__VecAddPdS_S_,@function _Z21__device_stub__VecAddPdS_S_: # @_Z21__device_stub__VecAddPdS_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 $_Z6VecAddPdS_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__VecAddPdS_S_, .Lfunc_end0-_Z21__device_stub__VecAddPdS_S_ .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 $_Z6VecAddPdS_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_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 _Z6VecAddPdS_S_,@object # @_Z6VecAddPdS_S_ .section .rodata,"a",@progbits .globl _Z6VecAddPdS_S_ .p2align 3, 0x0 _Z6VecAddPdS_S_: .quad _Z21__device_stub__VecAddPdS_S_ .size _Z6VecAddPdS_S_, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z6VecAddPdS_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 .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__VecAddPdS_S_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z6VecAddPdS_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 : _Z6VecAddPdS_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 R8, SR_TID.X ; /* 0x0000000000087919 */ /* 0x000e220000002100 */ /*0020*/ HFMA2.MMA R9, -RZ, RZ, 0, 4.76837158203125e-07 ; /* 0x00000008ff097435 */ /* 0x000fe200000001ff */ /*0030*/ ULDC.64 UR4, c[0x0][0x118] ; /* 0x0000460000047ab9 */ /* 0x000fd20000000a00 */ /*0040*/ IMAD.WIDE R2, R8, R9, c[0x0][0x160] ; /* 0x0000580008027625 */ /* 0x001fc800078e0209 */ /*0050*/ IMAD.WIDE R4, R8.reuse, R9.reuse, c[0x0][0x168] ; /* 0x00005a0008047625 */ /* 0x0c0fe400078e0209 */ /*0060*/ LDG.E.64 R2, [R2.64] ; /* 0x0000000402027981 */ /* 0x000ea8000c1e1b00 */ /*0070*/ LDG.E.64 R4, [R4.64] ; /* 0x0000000404047981 */ /* 0x000ea2000c1e1b00 */ /*0080*/ IMAD.WIDE R8, R8, R9, c[0x0][0x170] ; /* 0x00005c0008087625 */ /* 0x000fe200078e0209 */ /*0090*/ DADD R6, R2, R4 ; /* 0x0000000002067229 */ /* 0x004e0e0000000004 */ /*00a0*/ STG.E.64 [R8.64], R6 ; /* 0x0000000608007986 */ /* 0x001fe2000c101b04 */ /*00b0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*00c0*/ BRA 0xc0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*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 _Z6VecAddPdS_S_ .globl _Z6VecAddPdS_S_ .p2align 8 .type _Z6VecAddPdS_S_,@function _Z6VecAddPdS_S_: s_load_b128 s[4:7], s[0:1], 0x0 v_lshlrev_b32_e32 v4, 3, v0 s_load_b64 s[0:1], s[0:1], 0x10 s_waitcnt lgkmcnt(0) s_clause 0x1 global_load_b64 v[0:1], v4, s[4:5] global_load_b64 v[2:3], v4, s[6:7] s_waitcnt vmcnt(0) v_add_f64 v[0:1], v[0:1], v[2:3] global_store_b64 v4, v[0:1], s[0:1] s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z6VecAddPdS_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 5 .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 _Z6VecAddPdS_S_, .Lfunc_end0-_Z6VecAddPdS_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: _Z6VecAddPdS_S_ .private_segment_fixed_size: 0 .sgpr_count: 8 .sgpr_spill_count: 0 .symbol: _Z6VecAddPdS_S_.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 5 .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_000f5fc6_00000000-6_VecAdd.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 _Z29__device_stub__Z6VecAddPdS_S_PdS_S_ .type _Z29__device_stub__Z6VecAddPdS_S_PdS_S_, @function _Z29__device_stub__Z6VecAddPdS_S_PdS_S_: .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 _Z6VecAddPdS_S_(%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 _Z29__device_stub__Z6VecAddPdS_S_PdS_S_, .-_Z29__device_stub__Z6VecAddPdS_S_PdS_S_ .globl _Z6VecAddPdS_S_ .type _Z6VecAddPdS_S_, @function _Z6VecAddPdS_S_: .LFB2052: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z6VecAddPdS_S_PdS_S_ addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2052: .size _Z6VecAddPdS_S_, .-_Z6VecAddPdS_S_ .section .rodata.str1.1,"aMS",@progbits,1 .LC0: .string "_Z6VecAddPdS_S_" .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 _Z6VecAddPdS_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 .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 "VecAdd.hip" .globl _Z21__device_stub__VecAddPdS_S_ # -- Begin function _Z21__device_stub__VecAddPdS_S_ .p2align 4, 0x90 .type _Z21__device_stub__VecAddPdS_S_,@function _Z21__device_stub__VecAddPdS_S_: # @_Z21__device_stub__VecAddPdS_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 $_Z6VecAddPdS_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__VecAddPdS_S_, .Lfunc_end0-_Z21__device_stub__VecAddPdS_S_ .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 $_Z6VecAddPdS_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_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 _Z6VecAddPdS_S_,@object # @_Z6VecAddPdS_S_ .section .rodata,"a",@progbits .globl _Z6VecAddPdS_S_ .p2align 3, 0x0 _Z6VecAddPdS_S_: .quad _Z21__device_stub__VecAddPdS_S_ .size _Z6VecAddPdS_S_, 8 .type .L__unnamed_1,@object # @0 .section .rodata.str1.1,"aMS",@progbits,1 .L__unnamed_1: .asciz "_Z6VecAddPdS_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 .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__VecAddPdS_S_ .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z6VecAddPdS_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<stdio.h> #include <cuda.h> #include <sys/time.h> __global__ void compute(int* x,int* y,int n){ int col=threadIdx.x+blockIdx.x*blockDim.x; int row=threadIdx.y+blockIdx.y*blockDim.y; int num=col+row*n; int neighbor=0; //cell in the middle has eight neighbors, //a cell in a corner has only three neighbors, //a cell on an edge has five neighbors. if(col<n && row<n){ //corner 3 //In order to move to corner, //it should move either diagonal or move left/right and move up/down //top left corner if(col==0 && row==0){ neighbor+=x[num+1]; //move right neighbor+=x[num+n]; //move bottom neighbor+=x[num+n+1]; //bottom right } //bottom left corner else if(col==0 && row==n-1){ neighbor+=x[num+1]; //move right neighbor+=x[num-n]; //move up neighbor+=x[num-n+1]; //top right } //bottom right else if(col==n-1 && row==n-1){ neighbor+=x[num-1]; //move left neighbor+=x[num-n]; //move up neighbor+=x[num-n-1]; //top left } //edge 5 //In order to move to edge // it should just move to left/right/up/down (including corner) //top edge else if(row==0 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num+n-1]; //bottom left -- corner neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right -- corner } //bottom edge else if(row==n-1 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left -- corner } //Left edge else if(col==0 && row>0 && row<n-1){ neighbor+=x[num+1]; //right neighbor+=x[num-n]; //Top neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n+1]; //Bottom right-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //cell in the middle has eight neighbors, else{ neighbor+=x[num-1];//left neighbor+=x[num+1]; //right neighbor+=x[num-n-1];//top left neighbor+=x[num-n]; //top neighbor+=x[num-n+1]; //top right neighbor+=x[num+n-1]; //bottom left neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right } //Die : 0 //Live: 1 //A live cell with zero or one live neighbor dies from loneliness. if(x[num]==1 && (neighbor ==0 || neighbor ==1)) y[num]=0; //A live cell with four or more live neighbors dies due to overpopulation. else if(x[num]==1 && neighbor>=4) y[num]=0; //A dead cell with two or three live neighbors becomes alive. else if(x[num]==1 && (neighbor==2 || neighbor==3)) y[num]=1; //Otherwise, a cell's state stays unchanged else y[num] = x[num]; } } int main(void){ int i,j,k; int row= 6; int col= 4; int start[row][col]; int Round[row][col]; dim3 threadsPerBlock(32,32); dim3 numBlocks(row/threadsPerBlock.x,col/threadsPerBlock.x); int* x; int* y; int generation =1;// maximum generation/iteration float milliseconds=0; cudaEvent_t t_start,t_stop; cudaEventCreate(&t_start); cudaEventCreate(&t_stop); //Initilazie the matrix of the x Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) start[i][j]=rand()%2; //Initilazie the matrix of the y Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) Round[i][j]=0; cudaMalloc((void **) &x,sizeof(int)*row*col); cudaMemcpy(x,start,sizeof(int)*row*col,cudaMemcpyHostToDevice); printf("Start\n"); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(start[i][j]){ printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } for(k=0;k<= generation;k++) { cudaEventRecord(t_start); compute<<<numBlocks,threadsPerBlock>>>(y,y,row); cudaEventRecord(t_stop); cudaMalloc((void **) &y,sizeof(int)*row*col); cudaMemcpy(Round,y,sizeof(int)*row*col,cudaMemcpyDeviceToHost); printf("\n Round %d \n",k); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(Round[i][j]) { printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } cudaEventElapsedTime(&milliseconds,t_start,t_stop); printf("Time taken for this computation = %f milliseconds\n\n",milliseconds); } return 0; }
code for sm_80 Function : _Z7computePiS_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*/ S2R R17, SR_CTAID.Y ; /* 0x0000000000117919 */ /* 0x000e280000002600 */ /*0020*/ S2R R0, SR_TID.Y ; /* 0x0000000000007919 */ /* 0x000e280000002200 */ /*0030*/ S2R R10, SR_CTAID.X ; /* 0x00000000000a7919 */ /* 0x000e680000002500 */ /*0040*/ S2R R3, SR_TID.X ; /* 0x0000000000037919 */ /* 0x000e620000002100 */ /*0050*/ IMAD R17, R17, c[0x0][0x4], R0 ; /* 0x0000010011117a24 */ /* 0x001fca00078e0200 */ /*0060*/ ISETP.GE.AND P0, PT, R17, c[0x0][0x170], PT ; /* 0x00005c0011007a0c */ /* 0x000fe20003f06270 */ /*0070*/ IMAD R10, R10, c[0x0][0x0], R3 ; /* 0x000000000a0a7a24 */ /* 0x002fca00078e0203 */ /*0080*/ ISETP.GE.OR P0, PT, R10, c[0x0][0x170], P0 ; /* 0x00005c000a007a0c */ /* 0x000fda0000706670 */ /*0090*/ @P0 EXIT ; /* 0x000000000000094d */ /* 0x000fea0003800000 */ /*00a0*/ LOP3.LUT P0, RZ, R10, R17, RZ, 0xfc, !PT ; /* 0x000000110aff7212 */ /* 0x000fe2000780fcff */ /*00b0*/ IMAD R0, R17, c[0x0][0x170], R10 ; /* 0x00005c0011007a24 */ /* 0x000fe200078e020a */ /*00c0*/ ULDC.64 UR6, c[0x0][0x118] ; /* 0x0000460000067ab9 */ /* 0x000fe20000000a00 */ /*00d0*/ IMAD.MOV.U32 R11, RZ, RZ, 0x4 ; /* 0x00000004ff0b7424 */ /* 0x000fe200078e00ff */ /*00e0*/ BSSY B0, 0x660 ; /* 0x0000057000007945 */ /* 0x000fe40003800000 */ /*00f0*/ SHF.R.S32.HI R9, RZ, 0x1f, R0 ; /* 0x0000001fff097819 */ /* 0x000fe20000011400 */ /*0100*/ IMAD.WIDE R2, R0, R11, c[0x0][0x160] ; /* 0x0000580000027625 */ /* 0x000fcc00078e020b */ /*0110*/ @!P0 BRA 0x610 ; /* 0x000004f000008947 */ /* 0x000fea0003800000 */ /*0120*/ IMAD.MOV.U32 R19, RZ, RZ, c[0x0][0x170] ; /* 0x00005c00ff137624 */ /* 0x000fe200078e00ff */ /*0130*/ ISETP.EQ.AND P1, PT, R10, RZ, PT ; /* 0x000000ff0a00720c */ /* 0x000fc80003f22270 */ /*0140*/ IADD3 R19, R19, -0x1, RZ ; /* 0xffffffff13137810 */ /* 0x000fc80007ffe0ff */ /*0150*/ ISETP.NE.AND P0, PT, R17, R19, PT ; /* 0x000000131100720c */ /* 0x000fda0003f05270 */ /*0160*/ @!P0 BRA P1, 0x5c0 ; /* 0x0000045000008947 */ /* 0x000fea0000800000 */ /*0170*/ ISETP.EQ.AND P2, PT, R10.reuse, R19.reuse, PT ; /* 0x000000130a00720c */ /* 0x0c0fe40003f42270 */ /*0180*/ ISETP.NE.AND P1, PT, R10, R19, PT ; /* 0x000000130a00720c */ /* 0x000fd60003f25270 */ /*0190*/ @!P0 BRA P2, 0x550 ; /* 0x000003b000008947 */ /* 0x000fea0001000000 */ /*01a0*/ ISETP.GT.AND P2, PT, R10.reuse, RZ, PT ; /* 0x000000ff0a00720c */ /* 0x040fe40003f44270 */ /*01b0*/ ISETP.LT.AND P3, PT, R10, R19, PT ; /* 0x000000130a00720c */ /* 0x000fe40003f61270 */ /*01c0*/ ISETP.EQ.AND P2, PT, R17, RZ, P2 ; /* 0x000000ff1100720c */ /* 0x000fe40001742270 */ /*01d0*/ IADD3 R8, R0, c[0x0][0x170], RZ ; /* 0x00005c0000087a10 */ /* 0x000fc80007ffe0ff */ /*01e0*/ IADD3 R4, R8, -0x1, RZ ; /* 0xffffffff08047810 */ /* 0x000fca0007ffe0ff */ /*01f0*/ IMAD.WIDE R4, R4, R11, c[0x0][0x160] ; /* 0x0000580004047625 */ /* 0x000fe400078e020b */ /*0200*/ @P2 BRA P3, 0x4e0 ; /* 0x000002d000002947 */ /* 0x000fea0001800000 */ /*0210*/ ISETP.GT.AND P0, PT, R10.reuse, RZ, !P0 ; /* 0x000000ff0a00720c */ /* 0x040fe20004704270 */ /*0220*/ ULDC UR4, c[0x0][0x170] ; /* 0x00005c0000047ab9 */ /* 0x000fe20000000800 */ /*0230*/ ISETP.LT.AND P2, PT, R10, R19, PT ; /* 0x000000130a00720c */ /* 0x000fe20003f41270 */ /*0240*/ ULOP3.LUT UR4, URZ, UR4, URZ, 0x33, !UPT ; /* 0x000000043f047292 */ /* 0x000fcc000f8e333f */ /*0250*/ IADD3 R6, R0, UR4, RZ ; /* 0x0000000400067c10 */ /* 0x000fe2000fffe0ff */ /*0260*/ IMAD.U32 R13, RZ, RZ, UR4 ; /* 0x00000004ff0d7e24 */ /* 0x000fc8000f8e00ff */ /*0270*/ IMAD.WIDE R6, R6, R11, c[0x0][0x160] ; /* 0x0000580006067625 */ /* 0x000fe200078e020b */ /*0280*/ @P0 BRA P2, 0x470 ; /* 0x000001e000000947 */ /* 0x000fea0001000000 */ /*0290*/ ISETP.GT.AND P0, PT, R17.reuse, RZ, PT ; /* 0x000000ff1100720c */ /* 0x040fe40003f04270 */ /*02a0*/ ISETP.LT.AND P2, PT, R17, R19, PT ; /* 0x000000131100720c */ /* 0x000fe40003f41270 */ /*02b0*/ ISETP.EQ.AND P0, PT, R10, RZ, P0 ; /* 0x000000ff0a00720c */ /* 0x000fe40000702270 */ /*02c0*/ IADD3 R14, R0, -c[0x0][0x170], RZ ; /* 0x80005c00000e7a10 */ /* 0x000fca0007ffe0ff */ /*02d0*/ IMAD.WIDE R14, R14, R11, c[0x0][0x160] ; /* 0x000058000e0e7625 */ /* 0x000fcc00078e020b */ /*02e0*/ @P0 BRA P2, 0x400 ; /* 0x0000011000000947 */ /* 0x000fea0001000000 */ /*02f0*/ ISETP.GT.AND P1, PT, R17.reuse, RZ, !P1 ; /* 0x000000ff1100720c */ /* 0x040fe20004f24270 */ /*0300*/ LDG.E R16, [R2.64+-0x4] ; /* 0xfffffc0602107981 */ /* 0x000ea6000c1e1900 */ /*0310*/ ISETP.LT.AND P1, PT, R17, R19, P1 ; /* 0x000000131100720c */ /* 0x000fe20000f21270 */ /*0320*/ LDG.E R14, [R14.64] ; /* 0x000000060e0e7981 */ /* 0x000ee8000c1e1900 */ /*0330*/ LDG.E R17, [R6.64] ; /* 0x0000000606117981 */ /* 0x000eb0000c1e1900 */ /*0340*/ @!P1 LDG.E R13, [R2.64+0x4] ; /* 0x00000406020d9981 */ /* 0x000ea8000c1e1900 */ /*0350*/ @!P1 LDG.E R10, [R6.64+0x8] ; /* 0x00000806060a9981 */ /* 0x000ee8000c1e1900 */ /*0360*/ @!P1 LDG.E R5, [R4.64] ; /* 0x0000000604059981 */ /* 0x000f22000c1e1900 */ /*0370*/ @!P1 IMAD.MOV.U32 R12, RZ, RZ, R8 ; /* 0x000000ffff0c9224 */ /* 0x000fc400078e0008 */ /*0380*/ @P1 IMAD.MOV.U32 R12, RZ, RZ, R8 ; /* 0x000000ffff0c1224 */ /* 0x000fe200078e0008 */ /*0390*/ @!P1 IADD3 R13, R17, R13, R16 ; /* 0x0000000d110d9210 */ /* 0x004fc80007ffe010 */ /*03a0*/ @!P1 IADD3 R10, R10, R13, R14 ; /* 0x0000000d0a0a9210 */ /* 0x008fe20007ffe00e */ /*03b0*/ @!P1 IMAD.MOV.U32 R13, RZ, RZ, 0x1 ; /* 0x00000001ff0d9424 */ /* 0x000fe400078e00ff */ /*03c0*/ @P1 IMAD.MOV.U32 R13, RZ, RZ, -0x1 ; /* 0xffffffffff0d1424 */ /* 0x000fe400078e00ff */ /*03d0*/ @!P1 IMAD.IADD R10, R10, 0x1, R5 ; /* 0x000000010a0a9824 */ /* 0x010fe200078e0205 */ /*03e0*/ @P1 IADD3 R10, R17, R14, R16 ; /* 0x0000000e110a1210 */ /* 0x000fe20007ffe010 */ /*03f0*/ BRA 0x650 ; /* 0x0000025000007947 */ /* 0x000fea0003800000 */ /*0400*/ LDG.E R10, [R14.64] ; /* 0x000000060e0a7981 */ /* 0x000ea8000c1e1900 */ /*0410*/ LDG.E R6, [R6.64+0x8] ; /* 0x0000080606067981 */ /* 0x000ea8000c1e1900 */ /*0420*/ LDG.E R5, [R2.64+0x4] ; /* 0x0000040602057981 */ /* 0x000ea2000c1e1900 */ /*0430*/ IMAD.MOV.U32 R13, RZ, RZ, 0x1 ; /* 0x00000001ff0d7424 */ /* 0x000fc400078e00ff */ /*0440*/ IMAD.MOV.U32 R12, RZ, RZ, R8 ; /* 0x000000ffff0c7224 */ /* 0x000fe200078e0008 */ /*0450*/ IADD3 R10, R6, R10, R5 ; /* 0x0000000a060a7210 */ /* 0x004fe20007ffe005 */ /*0460*/ BRA 0x650 ; /* 0x000001e000007947 */ /* 0x000fea0003800000 */ /*0470*/ LDG.E R6, [R6.64+0x8] ; /* 0x0000080606067981 */ /* 0x000ea8000c1e1900 */ /*0480*/ LDG.E R10, [R2.64+0x4] ; /* 0x00000406020a7981 */ /* 0x000ea8000c1e1900 */ /*0490*/ LDG.E R5, [R2.64+-0x4] ; /* 0xfffffc0602057981 */ /* 0x000ea2000c1e1900 */ /*04a0*/ IADD3 R12, R0, -c[0x0][0x170], RZ ; /* 0x80005c00000c7a10 */ /* 0x000fe20007ffe0ff */ /*04b0*/ IMAD.MOV.U32 R8, RZ, RZ, R0 ; /* 0x000000ffff087224 */ /* 0x000fe200078e0000 */ /*04c0*/ IADD3 R10, R6, R10, R5 ; /* 0x0000000a060a7210 */ /* 0x004fe20007ffe005 */ /*04d0*/ BRA 0x650 ; /* 0x0000017000007947 */ /* 0x000fea0003800000 */ /*04e0*/ LDG.E R4, [R4.64] ; /* 0x0000000604047981 */ /* 0x000ea8000c1e1900 */ /*04f0*/ LDG.E R10, [R2.64+0x4] ; /* 0x00000406020a7981 */ /* 0x000ea8000c1e1900 */ /*0500*/ LDG.E R7, [R2.64+-0x4] ; /* 0xfffffc0602077981 */ /* 0x000ea2000c1e1900 */ /*0510*/ IMAD.MOV.U32 R13, RZ, RZ, 0x1 ; /* 0x00000001ff0d7424 */ /* 0x000fc400078e00ff */ /*0520*/ IMAD.MOV.U32 R12, RZ, RZ, R8 ; /* 0x000000ffff0c7224 */ /* 0x000fe200078e0008 */ /*0530*/ IADD3 R10, R4, R10, R7 ; /* 0x0000000a040a7210 */ /* 0x004fe20007ffe007 */ /*0540*/ BRA 0x650 ; /* 0x0000010000007947 */ /* 0x000fea0003800000 */ /*0550*/ LDG.E R10, [R2.64+-0x4] ; /* 0xfffffc06020a7981 */ /* 0x000162000c1e1900 */ /*0560*/ ULDC UR4, c[0x0][0x170] ; /* 0x00005c0000047ab9 */ /* 0x000fe20000000800 */ /*0570*/ IADD3 R12, R0, -c[0x0][0x170], RZ ; /* 0x80005c00000c7a10 */ /* 0x000fe20007ffe0ff */ /*0580*/ ULOP3.LUT UR4, URZ, UR4, URZ, 0x33, !UPT ; /* 0x000000043f047292 */ /* 0x000fe2000f8e333f */ /*0590*/ IMAD.MOV.U32 R8, RZ, RZ, R0 ; /* 0x000000ffff087224 */ /* 0x000fca00078e0000 */ /*05a0*/ IMAD.U32 R13, RZ, RZ, UR4 ; /* 0x00000004ff0d7e24 */ /* 0x000fe2000f8e00ff */ /*05b0*/ BRA 0x650 ; /* 0x0000009000007947 */ /* 0x000fea0003800000 */ /*05c0*/ LDG.E R10, [R2.64+0x4] ; /* 0x00000406020a7981 */ /* 0x000162000c1e1900 */ /*05d0*/ IADD3 R8, R0, -c[0x0][0x170], RZ ; /* 0x80005c0000087a10 */ /* 0x000fe20007ffe0ff */ /*05e0*/ IMAD.MOV.U32 R13, RZ, RZ, 0x1 ; /* 0x00000001ff0d7424 */ /* 0x000fc800078e00ff */ /*05f0*/ IMAD.MOV.U32 R12, RZ, RZ, R8 ; /* 0x000000ffff0c7224 */ /* 0x000fe200078e0008 */ /*0600*/ BRA 0x650 ; /* 0x0000004000007947 */ /* 0x000fea0003800000 */ /*0610*/ LDG.E R10, [R2.64+0x4] ; /* 0x00000406020a7981 */ /* 0x000162000c1e1900 */ /*0620*/ IADD3 R8, R0, c[0x0][0x170], RZ ; /* 0x00005c0000087a10 */ /* 0x000fe20007ffe0ff */ /*0630*/ IMAD.MOV.U32 R13, RZ, RZ, 0x1 ; /* 0x00000001ff0d7424 */ /* 0x000fc800078e00ff */ /*0640*/ IMAD.MOV.U32 R12, RZ, RZ, R8 ; /* 0x000000ffff0c7224 */ /* 0x000fe400078e0008 */ /*0650*/ BSYNC B0 ; /* 0x0000000000007941 */ /* 0x001fea0003800000 */ /*0660*/ IMAD.IADD R6, R13, 0x1, R8 ; /* 0x000000010d067824 */ /* 0x000fe200078e0208 */ /*0670*/ LDG.E R3, [R2.64] ; /* 0x0000000602037981 */ /* 0x000ea2000c1e1900 */ /*0680*/ IMAD.WIDE R4, R12, R11, c[0x0][0x160] ; /* 0x000058000c047625 */ /* 0x000fc800078e020b */ /*0690*/ IMAD.WIDE R6, R6, R11, c[0x0][0x160] ; /* 0x0000580006067625 */ /* 0x000fe400078e020b */ /*06a0*/ LDG.E R5, [R4.64] ; /* 0x0000000604057981 */ /* 0x000ee8000c1e1900 */ /*06b0*/ LDG.E R6, [R6.64] ; /* 0x0000000606067981 */ /* 0x000ee2000c1e1900 */ /*06c0*/ LEA R8, P1, R0, c[0x0][0x168], 0x2 ; /* 0x00005a0000087a11 */ /* 0x000fc800078210ff */ /*06d0*/ LEA.HI.X R9, R0, c[0x0][0x16c], R9, 0x2, P1 ; /* 0x00005b0000097a11 */ /* 0x000fe400008f1409 */ /*06e0*/ ISETP.NE.AND P0, PT, R3, 0x1, PT ; /* 0x000000010300780c */ /* 0x004fe40003f05270 */ /*06f0*/ IADD3 R10, R6, R5, R10 ; /* 0x00000005060a7210 */ /* 0x028fc80007ffe00a */ /*0700*/ ISETP.LT.U32.AND P2, PT, R10, 0x2, PT ; /* 0x000000020a00780c */ /* 0x000fda0003f41070 */ /*0710*/ @!P0 BRA P2, 0x7d0 ; /* 0x000000b000008947 */ /* 0x000fea0001000000 */ /*0720*/ ISETP.GT.AND P1, PT, R10, 0x3, PT ; /* 0x000000030a00780c */ /* 0x000fda0003f24270 */ /*0730*/ @!P0 BRA P1, 0x7b0 ; /* 0x0000007000008947 */ /* 0x000fea0000800000 */ /*0740*/ LOP3.LUT R0, R10, 0x1, RZ, 0xfc, !PT ; /* 0x000000010a007812 */ /* 0x000fc800078efcff */ /*0750*/ ISETP.EQ.AND P0, PT, R0, 0x3, !P0 ; /* 0x000000030000780c */ /* 0x000fda0004702270 */ /*0760*/ @!P0 STG.E [R8.64], R3 ; /* 0x0000000308008986 */ /* 0x0001e2000c101906 */ /*0770*/ @!P0 EXIT ; /* 0x000000000000894d */ /* 0x000fea0003800000 */ /*0780*/ IMAD.MOV.U32 R3, RZ, RZ, 0x1 ; /* 0x00000001ff037424 */ /* 0x001fca00078e00ff */ /*0790*/ STG.E [R8.64], R3 ; /* 0x0000000308007986 */ /* 0x000fe2000c101906 */ /*07a0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*07b0*/ STG.E [R8.64], RZ ; /* 0x000000ff08007986 */ /* 0x000fe2000c101906 */ /*07c0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*07d0*/ STG.E [R8.64], RZ ; /* 0x000000ff08007986 */ /* 0x000fe2000c101906 */ /*07e0*/ EXIT ; /* 0x000000000000794d */ /* 0x000fea0003800000 */ /*07f0*/ BRA 0x7f0; /* 0xfffffff000007947 */ /* 0x000fc0000383ffff */ /*0800*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0810*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0820*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0830*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0840*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0850*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0860*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ /*0870*/ NOP; /* 0x0000000000007918 */ /* 0x000fc00000000000 */ ..........
You are a helpful coding assistant specialized in converting CUDA source code to CUDA host assembly.
#include<stdio.h> #include <cuda.h> #include <sys/time.h> __global__ void compute(int* x,int* y,int n){ int col=threadIdx.x+blockIdx.x*blockDim.x; int row=threadIdx.y+blockIdx.y*blockDim.y; int num=col+row*n; int neighbor=0; //cell in the middle has eight neighbors, //a cell in a corner has only three neighbors, //a cell on an edge has five neighbors. if(col<n && row<n){ //corner 3 //In order to move to corner, //it should move either diagonal or move left/right and move up/down //top left corner if(col==0 && row==0){ neighbor+=x[num+1]; //move right neighbor+=x[num+n]; //move bottom neighbor+=x[num+n+1]; //bottom right } //bottom left corner else if(col==0 && row==n-1){ neighbor+=x[num+1]; //move right neighbor+=x[num-n]; //move up neighbor+=x[num-n+1]; //top right } //bottom right else if(col==n-1 && row==n-1){ neighbor+=x[num-1]; //move left neighbor+=x[num-n]; //move up neighbor+=x[num-n-1]; //top left } //edge 5 //In order to move to edge // it should just move to left/right/up/down (including corner) //top edge else if(row==0 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num+n-1]; //bottom left -- corner neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right -- corner } //bottom edge else if(row==n-1 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left -- corner } //Left edge else if(col==0 && row>0 && row<n-1){ neighbor+=x[num+1]; //right neighbor+=x[num-n]; //Top neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n+1]; //Bottom right-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //cell in the middle has eight neighbors, else{ neighbor+=x[num-1];//left neighbor+=x[num+1]; //right neighbor+=x[num-n-1];//top left neighbor+=x[num-n]; //top neighbor+=x[num-n+1]; //top right neighbor+=x[num+n-1]; //bottom left neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right } //Die : 0 //Live: 1 //A live cell with zero or one live neighbor dies from loneliness. if(x[num]==1 && (neighbor ==0 || neighbor ==1)) y[num]=0; //A live cell with four or more live neighbors dies due to overpopulation. else if(x[num]==1 && neighbor>=4) y[num]=0; //A dead cell with two or three live neighbors becomes alive. else if(x[num]==1 && (neighbor==2 || neighbor==3)) y[num]=1; //Otherwise, a cell's state stays unchanged else y[num] = x[num]; } } int main(void){ int i,j,k; int row= 6; int col= 4; int start[row][col]; int Round[row][col]; dim3 threadsPerBlock(32,32); dim3 numBlocks(row/threadsPerBlock.x,col/threadsPerBlock.x); int* x; int* y; int generation =1;// maximum generation/iteration float milliseconds=0; cudaEvent_t t_start,t_stop; cudaEventCreate(&t_start); cudaEventCreate(&t_stop); //Initilazie the matrix of the x Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) start[i][j]=rand()%2; //Initilazie the matrix of the y Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) Round[i][j]=0; cudaMalloc((void **) &x,sizeof(int)*row*col); cudaMemcpy(x,start,sizeof(int)*row*col,cudaMemcpyHostToDevice); printf("Start\n"); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(start[i][j]){ printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } for(k=0;k<= generation;k++) { cudaEventRecord(t_start); compute<<<numBlocks,threadsPerBlock>>>(y,y,row); cudaEventRecord(t_stop); cudaMalloc((void **) &y,sizeof(int)*row*col); cudaMemcpy(Round,y,sizeof(int)*row*col,cudaMemcpyDeviceToHost); printf("\n Round %d \n",k); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(Round[i][j]) { printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } cudaEventElapsedTime(&milliseconds,t_start,t_stop); printf("Time taken for this computation = %f milliseconds\n\n",milliseconds); } return 0; }
.file "tmpxft_000edb03_00000000-6_game.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__Z7computePiS_iPiS_i .type _Z29__device_stub__Z7computePiS_iPiS_i, @function _Z29__device_stub__Z7computePiS_iPiS_i: .LFB2082: .cfi_startproc endbr64 subq $136, %rsp .cfi_def_cfa_offset 144 movq %rdi, 24(%rsp) movq %rsi, 16(%rsp) movl %edx, 12(%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 12(%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 _Z7computePiS_i(%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 _Z29__device_stub__Z7computePiS_iPiS_i, .-_Z29__device_stub__Z7computePiS_iPiS_i .globl _Z7computePiS_i .type _Z7computePiS_i, @function _Z7computePiS_i: .LFB2083: .cfi_startproc endbr64 subq $8, %rsp .cfi_def_cfa_offset 16 call _Z29__device_stub__Z7computePiS_iPiS_i addq $8, %rsp .cfi_def_cfa_offset 8 ret .cfi_endproc .LFE2083: .size _Z7computePiS_i, .-_Z7computePiS_i .section .rodata.str1.1,"aMS",@progbits,1 .LC1: .string "Start\n" .LC2: .string "-------\n" .LC3: .string " 0" .LC4: .string " 1" .LC5: .string "\n" .LC6: .string "\n Round %d \n" .section .rodata.str1.8,"aMS",@progbits,1 .align 8 .LC7: .string "Time taken for this computation = %f milliseconds\n\n" .text .globl main .type main, @function main: .LFB2057: .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 $296, %rsp .cfi_def_cfa_offset 352 movq %fs:40, %rax movq %rax, 280(%rsp) xorl %eax, %eax movl $32, 60(%rsp) movl $1, 64(%rsp) movl $0, 68(%rsp) movl $0, 72(%rsp) movl $1, 76(%rsp) movl $0x00000000, 20(%rsp) leaq 40(%rsp), %rdi call cudaEventCreate@PLT leaq 48(%rsp), %rdi call cudaEventCreate@PLT leaq 96(%rsp), %rbp leaq 192(%rsp), %r14 movq %rbp, %r12 jmp .L12 .L36: addq $16, %r12 cmpq %r14, %r12 je .L35 .L12: leaq -16(%r12), %rbx .L13: call rand@PLT movl %eax, %edx shrl $31, %edx addl %edx, %eax andl $1, %eax subl %edx, %eax movl %eax, (%rbx) addq $4, %rbx cmpq %r12, %rbx jne .L13 jmp .L36 .L35: leaq 176(%rsp), %rax movq %rax, 8(%rsp) leaq 272(%rsp), %rdx .L14: movl $0, (%rax) movl $0, 4(%rax) movl $0, 8(%rax) movl $0, 12(%rax) addq $16, %rax cmpq %rdx, %rax jne .L14 leaq 24(%rsp), %rdi movl $96, %esi call cudaMalloc@PLT leaq 80(%rsp), %rsi movl $1, %ecx movl $96, %edx movq 24(%rsp), %rdi call cudaMemcpy@PLT 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 .LC4(%rip), %r13 leaq .LC3(%rip), %r12 leaq .LC5(%rip), %r15 jmp .L15 .L16: movq %r13, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT .L17: addq $4, %rbx cmpq %rbp, %rbx je .L37 .L18: cmpl $0, (%rbx) je .L16 movq %r12, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT jmp .L17 .L37: movq %r15, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addq $16, %rbp cmpq %r14, %rbp je .L27 .L15: leaq -16(%rbp), %rbx jmp .L18 .L27: movl $0, 4(%rsp) leaq .LC4(%rip), %r13 leaq .LC3(%rip), %r12 jmp .L19 .L40: movq 32(%rsp), %rdi movl $6, %edx movq %rdi, %rsi call _Z29__device_stub__Z7computePiS_iPiS_i jmp .L20 .L22: movq %r13, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT .L23: addq $4, %rbx cmpq %rbx, %rbp je .L38 .L24: cmpl $0, (%rbx) je .L22 movq %r12, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT jmp .L23 .L38: movq %r14, %rsi movl $2, %edi movl $0, %eax call __printf_chk@PLT addq $16, %rbp cmpq %r15, %rbp je .L25 .L21: leaq -16(%rbp), %rbx jmp .L24 .L25: leaq 20(%rsp), %rdi movq 48(%rsp), %rdx movq 40(%rsp), %rsi call cudaEventElapsedTime@PLT pxor %xmm0, %xmm0 cvtss2sd 20(%rsp), %xmm0 leaq .LC7(%rip), %rsi movl $2, %edi movl $1, %eax call __printf_chk@PLT addl $1, 4(%rsp) movl 4(%rsp), %eax cmpl $2, %eax je .L39 .L19: movl $0, %esi movq 40(%rsp), %rdi call cudaEventRecord@PLT movl $32, 56(%rsp) movl 64(%rsp), %ecx movl $0, %r9d movl $0, %r8d movq 56(%rsp), %rdx movq 68(%rsp), %rdi movl 76(%rsp), %esi call __cudaPushCallConfiguration@PLT testl %eax, %eax je .L40 .L20: movl $0, %esi movq 48(%rsp), %rdi call cudaEventRecord@PLT leaq 32(%rsp), %rdi movl $96, %esi call cudaMalloc@PLT leaq 176(%rsp), %rdi movl $2, %ecx movl $96, %edx movq 32(%rsp), %rsi call cudaMemcpy@PLT movl 4(%rsp), %edx leaq .LC6(%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 movq 8(%rsp), %rax leaq 16(%rax), %rbp leaq 112(%rax), %r15 leaq .LC5(%rip), %r14 jmp .L21 .L39: movq 280(%rsp), %rax subq %fs:40, %rax jne .L41 movl $0, %eax addq $296, %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 .L41: .cfi_restore_state call __stack_chk_fail@PLT .cfi_endproc .LFE2057: .size main, .-main .section .rodata.str1.1 .LC8: .string "_Z7computePiS_i" .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 _Z7computePiS_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 .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> #include <cuda.h> #include <sys/time.h> __global__ void compute(int* x,int* y,int n){ int col=threadIdx.x+blockIdx.x*blockDim.x; int row=threadIdx.y+blockIdx.y*blockDim.y; int num=col+row*n; int neighbor=0; //cell in the middle has eight neighbors, //a cell in a corner has only three neighbors, //a cell on an edge has five neighbors. if(col<n && row<n){ //corner 3 //In order to move to corner, //it should move either diagonal or move left/right and move up/down //top left corner if(col==0 && row==0){ neighbor+=x[num+1]; //move right neighbor+=x[num+n]; //move bottom neighbor+=x[num+n+1]; //bottom right } //bottom left corner else if(col==0 && row==n-1){ neighbor+=x[num+1]; //move right neighbor+=x[num-n]; //move up neighbor+=x[num-n+1]; //top right } //bottom right else if(col==n-1 && row==n-1){ neighbor+=x[num-1]; //move left neighbor+=x[num-n]; //move up neighbor+=x[num-n-1]; //top left } //edge 5 //In order to move to edge // it should just move to left/right/up/down (including corner) //top edge else if(row==0 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num+n-1]; //bottom left -- corner neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right -- corner } //bottom edge else if(row==n-1 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left -- corner } //Left edge else if(col==0 && row>0 && row<n-1){ neighbor+=x[num+1]; //right neighbor+=x[num-n]; //Top neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n+1]; //Bottom right-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //cell in the middle has eight neighbors, else{ neighbor+=x[num-1];//left neighbor+=x[num+1]; //right neighbor+=x[num-n-1];//top left neighbor+=x[num-n]; //top neighbor+=x[num-n+1]; //top right neighbor+=x[num+n-1]; //bottom left neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right } //Die : 0 //Live: 1 //A live cell with zero or one live neighbor dies from loneliness. if(x[num]==1 && (neighbor ==0 || neighbor ==1)) y[num]=0; //A live cell with four or more live neighbors dies due to overpopulation. else if(x[num]==1 && neighbor>=4) y[num]=0; //A dead cell with two or three live neighbors becomes alive. else if(x[num]==1 && (neighbor==2 || neighbor==3)) y[num]=1; //Otherwise, a cell's state stays unchanged else y[num] = x[num]; } } int main(void){ int i,j,k; int row= 6; int col= 4; int start[row][col]; int Round[row][col]; dim3 threadsPerBlock(32,32); dim3 numBlocks(row/threadsPerBlock.x,col/threadsPerBlock.x); int* x; int* y; int generation =1;// maximum generation/iteration float milliseconds=0; cudaEvent_t t_start,t_stop; cudaEventCreate(&t_start); cudaEventCreate(&t_stop); //Initilazie the matrix of the x Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) start[i][j]=rand()%2; //Initilazie the matrix of the y Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) Round[i][j]=0; cudaMalloc((void **) &x,sizeof(int)*row*col); cudaMemcpy(x,start,sizeof(int)*row*col,cudaMemcpyHostToDevice); printf("Start\n"); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(start[i][j]){ printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } for(k=0;k<= generation;k++) { cudaEventRecord(t_start); compute<<<numBlocks,threadsPerBlock>>>(y,y,row); cudaEventRecord(t_stop); cudaMalloc((void **) &y,sizeof(int)*row*col); cudaMemcpy(Round,y,sizeof(int)*row*col,cudaMemcpyDeviceToHost); printf("\n Round %d \n",k); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(Round[i][j]) { printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } cudaEventElapsedTime(&milliseconds,t_start,t_stop); printf("Time taken for this computation = %f milliseconds\n\n",milliseconds); } return 0; }
#include<stdio.h> #include <hip/hip_runtime.h> #include <sys/time.h> __global__ void compute(int* x,int* y,int n){ int col=threadIdx.x+blockIdx.x*blockDim.x; int row=threadIdx.y+blockIdx.y*blockDim.y; int num=col+row*n; int neighbor=0; //cell in the middle has eight neighbors, //a cell in a corner has only three neighbors, //a cell on an edge has five neighbors. if(col<n && row<n){ //corner 3 //In order to move to corner, //it should move either diagonal or move left/right and move up/down //top left corner if(col==0 && row==0){ neighbor+=x[num+1]; //move right neighbor+=x[num+n]; //move bottom neighbor+=x[num+n+1]; //bottom right } //bottom left corner else if(col==0 && row==n-1){ neighbor+=x[num+1]; //move right neighbor+=x[num-n]; //move up neighbor+=x[num-n+1]; //top right } //bottom right else if(col==n-1 && row==n-1){ neighbor+=x[num-1]; //move left neighbor+=x[num-n]; //move up neighbor+=x[num-n-1]; //top left } //edge 5 //In order to move to edge // it should just move to left/right/up/down (including corner) //top edge else if(row==0 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num+n-1]; //bottom left -- corner neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right -- corner } //bottom edge else if(row==n-1 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left -- corner } //Left edge else if(col==0 && row>0 && row<n-1){ neighbor+=x[num+1]; //right neighbor+=x[num-n]; //Top neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n+1]; //Bottom right-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //cell in the middle has eight neighbors, else{ neighbor+=x[num-1];//left neighbor+=x[num+1]; //right neighbor+=x[num-n-1];//top left neighbor+=x[num-n]; //top neighbor+=x[num-n+1]; //top right neighbor+=x[num+n-1]; //bottom left neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right } //Die : 0 //Live: 1 //A live cell with zero or one live neighbor dies from loneliness. if(x[num]==1 && (neighbor ==0 || neighbor ==1)) y[num]=0; //A live cell with four or more live neighbors dies due to overpopulation. else if(x[num]==1 && neighbor>=4) y[num]=0; //A dead cell with two or three live neighbors becomes alive. else if(x[num]==1 && (neighbor==2 || neighbor==3)) y[num]=1; //Otherwise, a cell's state stays unchanged else y[num] = x[num]; } } int main(void){ int i,j,k; int row= 6; int col= 4; int start[row][col]; int Round[row][col]; dim3 threadsPerBlock(32,32); dim3 numBlocks(row/threadsPerBlock.x,col/threadsPerBlock.x); int* x; int* y; int generation =1;// maximum generation/iteration float milliseconds=0; hipEvent_t t_start,t_stop; hipEventCreate(&t_start); hipEventCreate(&t_stop); //Initilazie the matrix of the x Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) start[i][j]=rand()%2; //Initilazie the matrix of the y Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) Round[i][j]=0; hipMalloc((void **) &x,sizeof(int)*row*col); hipMemcpy(x,start,sizeof(int)*row*col,hipMemcpyHostToDevice); printf("Start\n"); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(start[i][j]){ printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } for(k=0;k<= generation;k++) { hipEventRecord(t_start); compute<<<numBlocks,threadsPerBlock>>>(y,y,row); hipEventRecord(t_stop); hipMalloc((void **) &y,sizeof(int)*row*col); hipMemcpy(Round,y,sizeof(int)*row*col,hipMemcpyDeviceToHost); printf("\n Round %d \n",k); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(Round[i][j]) { printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } hipEventElapsedTime(&milliseconds,t_start,t_stop); printf("Time taken for this computation = %f milliseconds\n\n",milliseconds); } return 0; }
You are a helpful coding assistant specialized in converting HIP source code to AMD device assembly.
#include<stdio.h> #include <hip/hip_runtime.h> #include <sys/time.h> __global__ void compute(int* x,int* y,int n){ int col=threadIdx.x+blockIdx.x*blockDim.x; int row=threadIdx.y+blockIdx.y*blockDim.y; int num=col+row*n; int neighbor=0; //cell in the middle has eight neighbors, //a cell in a corner has only three neighbors, //a cell on an edge has five neighbors. if(col<n && row<n){ //corner 3 //In order to move to corner, //it should move either diagonal or move left/right and move up/down //top left corner if(col==0 && row==0){ neighbor+=x[num+1]; //move right neighbor+=x[num+n]; //move bottom neighbor+=x[num+n+1]; //bottom right } //bottom left corner else if(col==0 && row==n-1){ neighbor+=x[num+1]; //move right neighbor+=x[num-n]; //move up neighbor+=x[num-n+1]; //top right } //bottom right else if(col==n-1 && row==n-1){ neighbor+=x[num-1]; //move left neighbor+=x[num-n]; //move up neighbor+=x[num-n-1]; //top left } //edge 5 //In order to move to edge // it should just move to left/right/up/down (including corner) //top edge else if(row==0 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num+n-1]; //bottom left -- corner neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right -- corner } //bottom edge else if(row==n-1 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left -- corner } //Left edge else if(col==0 && row>0 && row<n-1){ neighbor+=x[num+1]; //right neighbor+=x[num-n]; //Top neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n+1]; //Bottom right-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //cell in the middle has eight neighbors, else{ neighbor+=x[num-1];//left neighbor+=x[num+1]; //right neighbor+=x[num-n-1];//top left neighbor+=x[num-n]; //top neighbor+=x[num-n+1]; //top right neighbor+=x[num+n-1]; //bottom left neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right } //Die : 0 //Live: 1 //A live cell with zero or one live neighbor dies from loneliness. if(x[num]==1 && (neighbor ==0 || neighbor ==1)) y[num]=0; //A live cell with four or more live neighbors dies due to overpopulation. else if(x[num]==1 && neighbor>=4) y[num]=0; //A dead cell with two or three live neighbors becomes alive. else if(x[num]==1 && (neighbor==2 || neighbor==3)) y[num]=1; //Otherwise, a cell's state stays unchanged else y[num] = x[num]; } } int main(void){ int i,j,k; int row= 6; int col= 4; int start[row][col]; int Round[row][col]; dim3 threadsPerBlock(32,32); dim3 numBlocks(row/threadsPerBlock.x,col/threadsPerBlock.x); int* x; int* y; int generation =1;// maximum generation/iteration float milliseconds=0; hipEvent_t t_start,t_stop; hipEventCreate(&t_start); hipEventCreate(&t_stop); //Initilazie the matrix of the x Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) start[i][j]=rand()%2; //Initilazie the matrix of the y Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) Round[i][j]=0; hipMalloc((void **) &x,sizeof(int)*row*col); hipMemcpy(x,start,sizeof(int)*row*col,hipMemcpyHostToDevice); printf("Start\n"); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(start[i][j]){ printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } for(k=0;k<= generation;k++) { hipEventRecord(t_start); compute<<<numBlocks,threadsPerBlock>>>(y,y,row); hipEventRecord(t_stop); hipMalloc((void **) &y,sizeof(int)*row*col); hipMemcpy(Round,y,sizeof(int)*row*col,hipMemcpyDeviceToHost); printf("\n Round %d \n",k); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(Round[i][j]) { printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } hipEventElapsedTime(&milliseconds,t_start,t_stop); printf("Time taken for this computation = %f milliseconds\n\n",milliseconds); } return 0; }
.text .amdgcn_target "amdgcn-amd-amdhsa--gfx1100" .protected _Z7computePiS_i .globl _Z7computePiS_i .p2align 8 .type _Z7computePiS_i,@function _Z7computePiS_i: s_clause 0x1 s_load_b32 s2, s[0:1], 0x24 s_load_b32 s7, s[0:1], 0x10 v_and_b32_e32 v1, 0x3ff, v0 v_bfe_u32 v0, v0, 10, 10 s_waitcnt lgkmcnt(0) s_and_b32 s3, s2, 0xffff s_lshr_b32 s2, s2, 16 v_mad_u64_u32 v[6:7], null, s14, s3, v[1:2] v_mad_u64_u32 v[4:5], null, s15, s2, v[0:1] s_mov_b32 s2, exec_lo s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1) v_max_i32_e32 v0, v6, v4 v_cmpx_gt_i32_e64 s7, v0 s_cbranch_execz .LBB0_41 s_load_b64 s[8:9], s[0:1], 0x0 v_or_b32_e32 v2, v4, v6 v_mad_u64_u32 v[0:1], null, v4, s7, v[6:7] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_cmp_ne_u32_e32 vcc_lo, 0, v2 s_and_saveexec_b32 s2, vcc_lo s_xor_b32 s10, exec_lo, s2 s_cbranch_execz .LBB0_27 s_add_i32 s13, s7, -1 v_cmp_eq_u32_e64 s2, 0, v6 v_cmp_eq_u32_e64 s3, s13, v4 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_b32 s4, s2, s3 s_xor_b32 s4, s4, -1 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_saveexec_b32 s5, s4 s_xor_b32 s11, exec_lo, s5 s_cbranch_execz .LBB0_24 v_cmp_eq_u32_e32 vcc_lo, s13, v6 s_and_b32 s4, vcc_lo, s3 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_xor_b32 s4, s4, -1 s_and_saveexec_b32 s5, s4 s_delay_alu instid0(SALU_CYCLE_1) s_xor_b32 s12, exec_lo, s5 s_cbranch_execz .LBB0_21 v_cmp_eq_u32_e64 s6, 0, v4 v_cmp_lt_i32_e64 s4, 0, v6 v_cmp_gt_i32_e64 s5, s13, v6 s_delay_alu instid0(VALU_DEP_2) s_and_b32 s6, s4, s6 s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) s_and_b32 s6, s5, s6 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_xor_b32 s6, s6, -1 s_and_saveexec_b32 s14, s6 s_delay_alu instid0(SALU_CYCLE_1) s_xor_b32 s6, exec_lo, s14 s_cbranch_execz .LBB0_18 s_and_b32 s3, s4, s3 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_b32 s3, s5, s3 s_xor_b32 s3, s3, -1 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_saveexec_b32 s4, s3 s_xor_b32 s5, exec_lo, s4 s_cbranch_execz .LBB0_15 v_cmp_lt_i32_e64 s3, 0, v4 v_cmp_gt_i32_e64 s4, s13, v4 s_delay_alu instid0(VALU_DEP_2) s_and_b32 s2, s2, s3 s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) s_and_b32 s2, s4, s2 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_xor_b32 s2, s2, -1 s_and_saveexec_b32 s13, s2 s_delay_alu instid0(SALU_CYCLE_1) s_xor_b32 s13, exec_lo, s13 s_cbranch_execz .LBB0_12 v_ashrrev_i32_e32 v1, 31, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(VALU_DEP_1) v_lshlrev_b64 v[1:2], 2, v[0:1] s_waitcnt lgkmcnt(0) v_add_co_u32 v4, s2, s8, v1 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_add_co_ci_u32_e64 v5, s2, s9, v2, s2 s_and_b32 s2, vcc_lo, s3 s_and_b32 s2, s4, s2 global_load_b32 v1, v[4:5], off offset:-4 s_xor_b32 s2, s2, -1 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_saveexec_b32 s3, s2 s_xor_b32 s2, exec_lo, s3 s_cbranch_execz .LBB0_9 v_subrev_nc_u32_e32 v2, s7, v0 v_add_nc_u32_e32 v6, s7, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_ashrrev_i32_e32 v7, 31, v6 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[2:3], 2, v[2:3] v_lshlrev_b64 v[6:7], 2, v[6:7] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v2, vcc_lo, s8, v2 v_add_co_ci_u32_e32 v3, vcc_lo, s9, v3, vcc_lo s_clause 0x1 global_load_b32 v8, v[4:5], off offset:4 global_load_b32 v9, v[2:3], off offset:-4 v_add_co_u32 v4, vcc_lo, s8, v6 v_add_co_ci_u32_e32 v5, vcc_lo, s9, v7, vcc_lo s_clause 0x2 global_load_b64 v[2:3], v[2:3], off global_load_b32 v6, v[4:5], off offset:-4 global_load_b32 v7, v[4:5], off s_waitcnt vmcnt(3) v_add3_u32 v1, v8, v1, v9 s_waitcnt vmcnt(2) s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_3) | instid1(VALU_DEP_3) v_add3_u32 v1, v1, v2, v3 v_add_co_u32 v2, vcc_lo, v4, 4 v_add_co_ci_u32_e32 v3, vcc_lo, 0, v5, vcc_lo s_waitcnt vmcnt(0) v_add3_u32 v7, v1, v6, v7 .LBB0_9: s_and_not1_saveexec_b32 s2, s2 s_cbranch_execz .LBB0_11 v_subrev_nc_u32_e32 v2, s7, v0 v_add_nc_u32_e32 v4, s7, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_ashrrev_i32_e32 v5, 31, v4 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[2:3], 2, v[2:3] v_lshlrev_b64 v[4:5], 2, v[4:5] s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v2, vcc_lo, s8, v2 v_add_co_ci_u32_e32 v3, vcc_lo, s9, v3, vcc_lo s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_4) v_add_co_u32 v4, vcc_lo, s8, v4 v_add_co_ci_u32_e32 v5, vcc_lo, s9, v5, vcc_lo s_clause 0x2 global_load_b32 v6, v[2:3], off global_load_b32 v7, v[4:5], off global_load_b32 v2, v[2:3], off offset:-4 s_waitcnt vmcnt(2) v_add_nc_u32_e32 v1, v6, v1 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_1) v_add3_u32 v7, v1, v2, v7 v_add_co_u32 v2, vcc_lo, v4, -4 v_add_co_ci_u32_e32 v3, vcc_lo, -1, v5, vcc_lo .LBB0_11: s_or_b32 exec_lo, exec_lo, s2 .LBB0_12: s_and_not1_saveexec_b32 s2, s13 s_cbranch_execz .LBB0_14 v_subrev_nc_u32_e32 v2, s7, v0 s_waitcnt vmcnt(0) v_ashrrev_i32_e32 v1, 31, v0 v_add_nc_u32_e32 v4, s7, v0 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_ashrrev_i32_e32 v3, 31, v2 v_lshlrev_b64 v[6:7], 2, v[0:1] s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_ashrrev_i32_e32 v5, 31, v4 v_lshlrev_b64 v[1:2], 2, v[2:3] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_4) v_lshlrev_b64 v[3:4], 2, v[4:5] s_waitcnt lgkmcnt(0) v_add_co_u32 v5, vcc_lo, s8, v6 v_add_co_ci_u32_e32 v6, vcc_lo, s9, v7, vcc_lo s_delay_alu instid0(VALU_DEP_4) v_add_co_u32 v1, vcc_lo, s8, v1 v_add_co_ci_u32_e32 v2, vcc_lo, s9, v2, vcc_lo v_add_co_u32 v3, vcc_lo, s8, v3 v_add_co_ci_u32_e32 v4, vcc_lo, s9, v4, vcc_lo s_clause 0x2 global_load_b32 v5, v[5:6], off offset:4 global_load_b64 v[1:2], v[1:2], off global_load_b32 v6, v[3:4], off s_waitcnt vmcnt(1) v_add_nc_u32_e32 v1, v1, v5 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_1) v_add3_u32 v7, v1, v2, v6 v_add_co_u32 v2, vcc_lo, v3, 4 v_add_co_ci_u32_e32 v3, vcc_lo, 0, v4, vcc_lo .LBB0_14: s_or_b32 exec_lo, exec_lo, s2 .LBB0_15: s_and_not1_saveexec_b32 s2, s5 s_cbranch_execz .LBB0_17 v_subrev_nc_u32_e32 v2, s7, v0 s_waitcnt vmcnt(0) v_ashrrev_i32_e32 v1, 31, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_lshlrev_b64 v[4:5], 2, v[0:1] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_lshlrev_b64 v[1:2], 2, v[2:3] s_waitcnt lgkmcnt(0) v_add_co_u32 v3, vcc_lo, s8, v4 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_ci_u32_e32 v4, vcc_lo, s9, v5, vcc_lo v_add_co_u32 v5, vcc_lo, s8, v1 s_delay_alu instid0(VALU_DEP_4) v_add_co_ci_u32_e32 v6, vcc_lo, s9, v2, vcc_lo s_clause 0x2 global_load_b32 v7, v[3:4], off offset:-4 global_load_b32 v3, v[3:4], off offset:4 global_load_b64 v[1:2], v[5:6], off s_waitcnt vmcnt(1) v_add_nc_u32_e32 v3, v3, v7 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_1) v_add3_u32 v7, v3, v2, v1 v_add_co_u32 v2, vcc_lo, v5, -4 v_add_co_ci_u32_e32 v3, vcc_lo, -1, v6, vcc_lo .LBB0_17: s_or_b32 exec_lo, exec_lo, s2 .LBB0_18: s_and_not1_saveexec_b32 s2, s6 s_cbranch_execz .LBB0_20 v_add_nc_u32_e32 v2, s7, v0 s_waitcnt vmcnt(0) v_ashrrev_i32_e32 v1, 31, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_lshlrev_b64 v[4:5], 2, v[0:1] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_lshlrev_b64 v[1:2], 2, v[2:3] s_waitcnt lgkmcnt(0) v_add_co_u32 v3, vcc_lo, s8, v4 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_ci_u32_e32 v4, vcc_lo, s9, v5, vcc_lo v_add_co_u32 v5, vcc_lo, s8, v1 s_delay_alu instid0(VALU_DEP_4) v_add_co_ci_u32_e32 v6, vcc_lo, s9, v2, vcc_lo s_clause 0x3 global_load_b32 v1, v[3:4], off offset:-4 global_load_b32 v2, v[3:4], off offset:4 global_load_b32 v3, v[5:6], off offset:-4 global_load_b32 v4, v[5:6], off s_waitcnt vmcnt(2) v_add_nc_u32_e32 v1, v2, v1 v_add_co_u32 v2, vcc_lo, v5, 4 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_2) v_add3_u32 v7, v1, v3, v4 v_add_co_ci_u32_e32 v3, vcc_lo, 0, v6, vcc_lo .LBB0_20: s_or_b32 exec_lo, exec_lo, s2 .LBB0_21: s_and_not1_saveexec_b32 s2, s12 s_cbranch_execz .LBB0_23 v_subrev_nc_u32_e32 v2, s7, v0 s_waitcnt vmcnt(0) v_ashrrev_i32_e32 v1, 31, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_lshlrev_b64 v[4:5], 2, v[0:1] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_lshlrev_b64 v[1:2], 2, v[2:3] s_waitcnt lgkmcnt(0) v_add_co_u32 v3, vcc_lo, s8, v4 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_ci_u32_e32 v4, vcc_lo, s9, v5, vcc_lo v_add_co_u32 v5, vcc_lo, s8, v1 s_delay_alu instid0(VALU_DEP_4) v_add_co_ci_u32_e32 v6, vcc_lo, s9, v2, vcc_lo s_clause 0x1 global_load_b32 v1, v[3:4], off offset:-4 global_load_b32 v2, v[5:6], off s_waitcnt vmcnt(0) v_add_nc_u32_e32 v7, v2, v1 v_add_co_u32 v2, vcc_lo, v5, -4 v_add_co_ci_u32_e32 v3, vcc_lo, -1, v6, vcc_lo .LBB0_23: s_or_b32 exec_lo, exec_lo, s2 .LBB0_24: s_and_not1_saveexec_b32 s2, s11 s_cbranch_execz .LBB0_26 v_subrev_nc_u32_e32 v2, s7, v0 s_waitcnt vmcnt(0) v_ashrrev_i32_e32 v1, 31, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_ashrrev_i32_e32 v3, 31, v2 v_lshlrev_b64 v[4:5], 2, v[0:1] s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2) v_lshlrev_b64 v[1:2], 2, v[2:3] s_waitcnt lgkmcnt(0) v_add_co_u32 v3, vcc_lo, s8, v4 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_ci_u32_e32 v4, vcc_lo, s9, v5, vcc_lo v_add_co_u32 v5, vcc_lo, s8, v1 s_delay_alu instid0(VALU_DEP_4) v_add_co_ci_u32_e32 v6, vcc_lo, s9, v2, vcc_lo s_clause 0x1 global_load_b32 v1, v[3:4], off offset:4 global_load_b32 v2, v[5:6], off s_waitcnt vmcnt(0) v_add_nc_u32_e32 v7, v2, v1 v_add_co_u32 v2, vcc_lo, v5, 4 v_add_co_ci_u32_e32 v3, vcc_lo, 0, v6, vcc_lo .LBB0_26: s_or_b32 exec_lo, exec_lo, s2 .LBB0_27: s_or_saveexec_b32 s2, s10 s_waitcnt vmcnt(0) s_delay_alu instid0(VALU_DEP_2) v_ashrrev_i32_e32 v1, 31, v0 s_xor_b32 exec_lo, exec_lo, s2 s_cbranch_execz .LBB0_29 v_add_nc_u32_e32 v2, s7, v0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2) v_lshlrev_b64 v[4:5], 2, v[0:1] v_ashrrev_i32_e32 v3, 31, v2 s_waitcnt lgkmcnt(0) s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v4, vcc_lo, s8, v4 v_add_co_ci_u32_e32 v5, vcc_lo, s9, v5, vcc_lo s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_lshlrev_b64 v[2:3], 2, v[2:3] v_add_co_u32 v2, vcc_lo, s8, v2 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v3, vcc_lo, s9, v3, vcc_lo s_clause 0x1 global_load_b32 v4, v[4:5], off offset:4 global_load_b32 v5, v[2:3], 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_add_nc_u32_e32 v7, v5, v4 .LBB0_29: s_or_b32 exec_lo, exec_lo, s2 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_2) | instid1(VALU_DEP_1) v_lshlrev_b64 v[4:5], 2, v[0:1] s_load_b64 s[2:3], s[0:1], 0x8 s_waitcnt lgkmcnt(0) v_add_co_u32 v4, vcc_lo, s8, v4 s_delay_alu instid0(VALU_DEP_2) v_add_co_ci_u32_e32 v5, vcc_lo, s9, v5, vcc_lo global_load_b32 v3, v[2:3], off global_load_b32 v2, v[4:5], off s_waitcnt vmcnt(1) v_add_nc_u32_e32 v3, v7, v3 s_waitcnt vmcnt(0) v_cmp_ne_u32_e64 s0, 1, v2 v_cmp_eq_u32_e32 vcc_lo, 1, v2 s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1) v_cmp_lt_u32_e64 s1, 1, v3 s_or_b32 s0, s0, s1 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_saveexec_b32 s1, s0 s_xor_b32 s1, exec_lo, s1 s_cbranch_execz .LBB0_39 v_cmp_gt_i32_e64 s0, 4, v3 s_xor_b32 s5, vcc_lo, -1 s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1) s_or_b32 s0, s5, s0 s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1) s_and_saveexec_b32 s4, s0 s_xor_b32 s4, exec_lo, s4 s_cbranch_execz .LBB0_36 v_lshlrev_b64 v[0:1], 2, v[0:1] v_and_b32_e32 v3, -2, v3 s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_3) v_cmp_ne_u32_e32 vcc_lo, 2, v3 v_add_co_u32 v0, s0, s2, v0 s_delay_alu instid0(VALU_DEP_1) | instskip(SKIP_1) | instid1(SALU_CYCLE_1) v_add_co_ci_u32_e64 v1, s0, s3, v1, s0 s_or_b32 s0, s5, vcc_lo s_and_saveexec_b32 s5, s0 s_delay_alu instid0(SALU_CYCLE_1) s_xor_b32 s0, exec_lo, s5 s_cbranch_execz .LBB0_33 global_store_b32 v[0:1], v2, off .LBB0_33: s_and_not1_saveexec_b32 s0, s0 s_cbranch_execz .LBB0_35 v_mov_b32_e32 v2, 1 global_store_b32 v[0:1], v2, off .LBB0_35: s_or_b32 exec_lo, exec_lo, s0 .LBB0_36: s_and_not1_saveexec_b32 s0, s4 s_cbranch_execz .LBB0_38 v_lshlrev_b64 v[0:1], 2, v[0:1] v_mov_b32_e32 v2, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v0, vcc_lo, s2, v0 v_add_co_ci_u32_e32 v1, vcc_lo, s3, v1, vcc_lo global_store_b32 v[0:1], v2, off .LBB0_38: s_or_b32 exec_lo, exec_lo, s0 .LBB0_39: s_and_not1_saveexec_b32 s0, s1 s_cbranch_execz .LBB0_41 v_lshlrev_b64 v[0:1], 2, v[0:1] v_mov_b32_e32 v2, 0 s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3) v_add_co_u32 v0, vcc_lo, s2, v0 v_add_co_ci_u32_e32 v1, vcc_lo, s3, v1, vcc_lo global_store_b32 v[0:1], v2, off .LBB0_41: s_nop 0 s_sendmsg sendmsg(MSG_DEALLOC_VGPRS) s_endpgm .section .rodata,"a",@progbits .p2align 6, 0x0 .amdhsa_kernel _Z7computePiS_i .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 10 .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 _Z7computePiS_i, .Lfunc_end0-_Z7computePiS_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 - .offset: 16 .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: _Z7computePiS_i .private_segment_fixed_size: 0 .sgpr_count: 18 .sgpr_spill_count: 0 .symbol: _Z7computePiS_i.kd .uniform_work_group_size: 1 .uses_dynamic_stack: false .vgpr_count: 10 .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 <hip/hip_runtime.h> #include <sys/time.h> __global__ void compute(int* x,int* y,int n){ int col=threadIdx.x+blockIdx.x*blockDim.x; int row=threadIdx.y+blockIdx.y*blockDim.y; int num=col+row*n; int neighbor=0; //cell in the middle has eight neighbors, //a cell in a corner has only three neighbors, //a cell on an edge has five neighbors. if(col<n && row<n){ //corner 3 //In order to move to corner, //it should move either diagonal or move left/right and move up/down //top left corner if(col==0 && row==0){ neighbor+=x[num+1]; //move right neighbor+=x[num+n]; //move bottom neighbor+=x[num+n+1]; //bottom right } //bottom left corner else if(col==0 && row==n-1){ neighbor+=x[num+1]; //move right neighbor+=x[num-n]; //move up neighbor+=x[num-n+1]; //top right } //bottom right else if(col==n-1 && row==n-1){ neighbor+=x[num-1]; //move left neighbor+=x[num-n]; //move up neighbor+=x[num-n-1]; //top left } //edge 5 //In order to move to edge // it should just move to left/right/up/down (including corner) //top edge else if(row==0 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num+n-1]; //bottom left -- corner neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right -- corner } //bottom edge else if(row==n-1 && col>0 && col<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num+1]; //right neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left -- corner } //Left edge else if(col==0 && row>0 && row<n-1){ neighbor+=x[num+1]; //right neighbor+=x[num-n]; //Top neighbor+=x[num-n+1]; //Top right-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n+1]; //Bottom right-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //Right edge else if(col==n-1 && row>0 && row<n-1){ neighbor+=x[num-1]; //left neighbor+=x[num-n]; //Top neighbor+=x[num-n-1]; //Top left-- corner neighbor+=x[num+n]; //Bottom neighbor+=x[num+n-1]; //Bottom left-- corner } //cell in the middle has eight neighbors, else{ neighbor+=x[num-1];//left neighbor+=x[num+1]; //right neighbor+=x[num-n-1];//top left neighbor+=x[num-n]; //top neighbor+=x[num-n+1]; //top right neighbor+=x[num+n-1]; //bottom left neighbor+=x[num+n]; //bottom neighbor+=x[num+n+1]; //bottom right } //Die : 0 //Live: 1 //A live cell with zero or one live neighbor dies from loneliness. if(x[num]==1 && (neighbor ==0 || neighbor ==1)) y[num]=0; //A live cell with four or more live neighbors dies due to overpopulation. else if(x[num]==1 && neighbor>=4) y[num]=0; //A dead cell with two or three live neighbors becomes alive. else if(x[num]==1 && (neighbor==2 || neighbor==3)) y[num]=1; //Otherwise, a cell's state stays unchanged else y[num] = x[num]; } } int main(void){ int i,j,k; int row= 6; int col= 4; int start[row][col]; int Round[row][col]; dim3 threadsPerBlock(32,32); dim3 numBlocks(row/threadsPerBlock.x,col/threadsPerBlock.x); int* x; int* y; int generation =1;// maximum generation/iteration float milliseconds=0; hipEvent_t t_start,t_stop; hipEventCreate(&t_start); hipEventCreate(&t_stop); //Initilazie the matrix of the x Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) start[i][j]=rand()%2; //Initilazie the matrix of the y Generated cell for(i=0;i<row;i++) for(j=0;j<col;j++) Round[i][j]=0; hipMalloc((void **) &x,sizeof(int)*row*col); hipMemcpy(x,start,sizeof(int)*row*col,hipMemcpyHostToDevice); printf("Start\n"); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(start[i][j]){ printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } for(k=0;k<= generation;k++) { hipEventRecord(t_start); compute<<<numBlocks,threadsPerBlock>>>(y,y,row); hipEventRecord(t_stop); hipMalloc((void **) &y,sizeof(int)*row*col); hipMemcpy(Round,y,sizeof(int)*row*col,hipMemcpyDeviceToHost); printf("\n Round %d \n",k); printf("-------\n"); for(i=0;i<row;i++) { for(j=0;j<col;j++) { if(Round[i][j]) { printf(" 0"); } else{ printf(" 1"); } } printf("\n"); } hipEventElapsedTime(&milliseconds,t_start,t_stop); printf("Time taken for this computation = %f milliseconds\n\n",milliseconds); } return 0; }
.text .file "game.hip" .globl _Z22__device_stub__computePiS_i # -- Begin function _Z22__device_stub__computePiS_i .p2align 4, 0x90 .type _Z22__device_stub__computePiS_i,@function _Z22__device_stub__computePiS_i: # @_Z22__device_stub__computePiS_i .cfi_startproc # %bb.0: subq $104, %rsp .cfi_def_cfa_offset 112 movq %rdi, 72(%rsp) movq %rsi, 64(%rsp) movl %edx, 12(%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 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 $_Z7computePiS_i, %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 _Z22__device_stub__computePiS_i, .Lfunc_end0-_Z22__device_stub__computePiS_i .cfi_endproc # -- End function .globl main # -- Begin function 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 $344, %rsp # imm = 0x158 .cfi_def_cfa_offset 400 .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 $0, 8(%rsp) leaq 24(%rsp), %rdi callq hipEventCreate leaq 16(%rsp), %rdi callq hipEventCreate leaq 240(%rsp), %rbx xorl %r14d, %r14d .p2align 4, 0x90 .LBB1_1: # %.preheader74 # =>This Loop Header: Depth=1 # Child Loop BB1_2 Depth 2 xorl %r15d, %r15d .p2align 4, 0x90 .LBB1_2: # Parent Loop BB1_1 Depth=1 # => This Inner Loop Header: Depth=2 callq rand movl %eax, %ecx shrl $31, %ecx addl %eax, %ecx andl $-2, %ecx subl %ecx, %eax movl %eax, (%rbx,%r15,4) incq %r15 cmpq $4, %r15 jne .LBB1_2 # %bb.3: # in Loop: Header=BB1_1 Depth=1 incq %r14 addq $16, %rbx cmpq $6, %r14 jne .LBB1_1 # %bb.4: # %.preheader72.preheader xorps %xmm0, %xmm0 movaps %xmm0, 224(%rsp) movaps %xmm0, 208(%rsp) movaps %xmm0, 192(%rsp) movaps %xmm0, 176(%rsp) movaps %xmm0, 160(%rsp) movaps %xmm0, 144(%rsp) leaq 40(%rsp), %rdi movl $96, %esi callq hipMalloc movq 40(%rsp), %rdi leaq 240(%rsp), %rbx movl $96, %edx movq %rbx, %rsi movl $1, %ecx callq hipMemcpy movl $.Lstr, %edi callq puts@PLT movl $.Lstr.2, %edi callq puts@PLT xorl %r14d, %r14d jmp .LBB1_5 .p2align 4, 0x90 .LBB1_9: # in Loop: Header=BB1_5 Depth=1 movl $10, %edi callq putchar@PLT incq %r14 addq $16, %rbx cmpq $6, %r14 je .LBB1_10 .LBB1_5: # %.preheader71 # =>This Loop Header: Depth=1 # Child Loop BB1_6 Depth 2 xorl %r15d, %r15d jmp .LBB1_6 .p2align 4, 0x90 .LBB1_8: # in Loop: Header=BB1_6 Depth=2 xorl %eax, %eax callq printf incq %r15 cmpq $4, %r15 je .LBB1_9 .LBB1_6: # Parent Loop BB1_5 Depth=1 # => This Inner Loop Header: Depth=2 cmpl $0, (%rbx,%r15,4) movl $.L.str.3, %edi je .LBB1_8 # %bb.7: # in Loop: Header=BB1_6 Depth=2 movl $.L.str.2, %edi jmp .LBB1_8 .LBB1_10: # %.preheader70 leaq 144(%rsp), %rbp leaq 8(%rsp), %r12 xorl %r13d, %r13d jmp .LBB1_11 .p2align 4, 0x90 .LBB1_19: # in Loop: Header=BB1_11 Depth=1 movq 24(%rsp), %rsi movq 16(%rsp), %rdx movq %r12, %rdi callq hipEventElapsedTime movss 8(%rsp), %xmm0 # xmm0 = mem[0],zero,zero,zero cvtss2sd %xmm0, %xmm0 movl $.L.str.6, %edi movb $1, %al callq printf leal 1(%r13), %eax testl %r13d, %r13d movl %eax, %r13d jne .LBB1_20 .LBB1_11: # =>This Loop Header: Depth=1 # Child Loop BB1_14 Depth 2 # Child Loop BB1_15 Depth 3 movq 24(%rsp), %rdi xorl %esi, %esi callq hipEventRecord xorl %edi, %edi movl $1, %esi movabsq $137438953504, %rdx # imm = 0x2000000020 movl $1, %ecx xorl %r8d, %r8d xorl %r9d, %r9d callq __hipPushCallConfiguration testl %eax, %eax jne .LBB1_13 # %bb.12: # in Loop: Header=BB1_11 Depth=1 movq 32(%rsp), %rax movq %rax, 104(%rsp) movq %rax, 96(%rsp) movl $6, 12(%rsp) leaq 104(%rsp), %rax movq %rax, 112(%rsp) leaq 96(%rsp), %rax movq %rax, 120(%rsp) leaq 12(%rsp), %rax movq %rax, 128(%rsp) leaq 80(%rsp), %rdi leaq 64(%rsp), %rsi leaq 56(%rsp), %rdx leaq 48(%rsp), %rcx callq __hipPopCallConfiguration movq 80(%rsp), %rsi movl 88(%rsp), %edx movq 64(%rsp), %rcx movl 72(%rsp), %r8d movl $_Z7computePiS_i, %edi leaq 112(%rsp), %r9 pushq 48(%rsp) .cfi_adjust_cfa_offset 8 pushq 64(%rsp) .cfi_adjust_cfa_offset 8 callq hipLaunchKernel addq $16, %rsp .cfi_adjust_cfa_offset -16 .LBB1_13: # in Loop: Header=BB1_11 Depth=1 movq 16(%rsp), %rdi xorl %esi, %esi callq hipEventRecord movl $96, %esi leaq 32(%rsp), %rdi callq hipMalloc movq 32(%rsp), %rsi movl $96, %edx movq %rbp, %rdi movl $2, %ecx callq hipMemcpy movl $.L.str.5, %edi movl %r13d, %esi xorl %eax, %eax callq printf movl $.Lstr.2, %edi callq puts@PLT movq %rbp, %r15 xorl %ebx, %ebx jmp .LBB1_14 .p2align 4, 0x90 .LBB1_18: # in Loop: Header=BB1_14 Depth=2 movl $10, %edi callq putchar@PLT incq %rbx addq $16, %r15 cmpq $6, %rbx je .LBB1_19 .LBB1_14: # %.preheader # Parent Loop BB1_11 Depth=1 # => This Loop Header: Depth=2 # Child Loop BB1_15 Depth 3 xorl %r14d, %r14d jmp .LBB1_15 .p2align 4, 0x90 .LBB1_17: # in Loop: Header=BB1_15 Depth=3 xorl %eax, %eax callq printf incq %r14 cmpq $4, %r14 je .LBB1_18 .LBB1_15: # Parent Loop BB1_11 Depth=1 # Parent Loop BB1_14 Depth=2 # => This Inner Loop Header: Depth=3 cmpl $0, (%r15,%r14,4) movl $.L.str.3, %edi je .LBB1_17 # %bb.16: # in Loop: Header=BB1_15 Depth=3 movl $.L.str.2, %edi jmp .LBB1_17 .LBB1_20: xorl %eax, %eax addq $344, %rsp # imm = 0x158 .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 $_Z7computePiS_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 _Z7computePiS_i,@object # @_Z7computePiS_i .section .rodata,"a",@progbits .globl _Z7computePiS_i .p2align 3, 0x0 _Z7computePiS_i: .quad _Z22__device_stub__computePiS_i .size _Z7computePiS_i, 8 .type .L.str.2,@object # @.str.2 .section .rodata.str1.1,"aMS",@progbits,1 .L.str.2: .asciz " 0" .size .L.str.2, 3 .type .L.str.3,@object # @.str.3 .L.str.3: .asciz " 1" .size .L.str.3, 3 .type .L.str.5,@object # @.str.5 .L.str.5: .asciz "\n Round %d \n" .size .L.str.5, 13 .type .L.str.6,@object # @.str.6 .L.str.6: .asciz "Time taken for this computation = %f milliseconds\n\n" .size .L.str.6, 52 .type .L__unnamed_1,@object # @0 .L__unnamed_1: .asciz "_Z7computePiS_i" .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 "Start" .size .Lstr, 6 .type .Lstr.2,@object # @str.2 .Lstr.2: .asciz "-------" .size .Lstr.2, 8 .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 _Z22__device_stub__computePiS_i .addrsig_sym __hip_module_ctor .addrsig_sym __hip_module_dtor .addrsig_sym _Z7computePiS_i .addrsig_sym __hip_fatbin .addrsig_sym __hip_fatbin_wrapper .addrsig_sym __hip_cuid_