file stringlengths 6 242 | name stringlengths 1 11 | kernel stringclasses 4 values | input stringlengths 531 14.6k | output stringlengths 176 17.2k |
|---|---|---|---|---|
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp1pe0pk2n | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<835>;
.reg .b32 %r<127>;
.reg .b64 %rd<32>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[448];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[57344];
ld.param.u64 %rd1, [default_function_kernel0_param_0];
ld.param.u64 %rd2, [default_function_kernel0_param_1];
mov.u32 %r13, %tid.x;
mul.lo.s32 %r1, %r13, 3584;
mov.u32 %r123, 0;
mov.f32 %f771, 0f00000000;
cvta.to.global.u64 %rd5, %rd1;
shl.b32 %r17, %r13, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r19, %r18, %r17;
cvta.to.global.u64 %rd8, %rd2;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(772+i*1), %f771;
$L__BB0_1:
mov.u32 %r124, 0;
bar.sync 0;
mul.lo.s32 %r3, %r123, 112;
add.s32 %r16, %r3, %r13;
mul.wide.s32 %rd6, %r16, 4;
add.s64 %rd7, %rd5, %rd6;
for.size.2 i in range(0, 56, 1):
ld.global.nc.f32 %f(321+i*1), [%rd7+(0+i*8)];
st.shared.f32 [%r19+(0+i*8)], %f(321+i*1);
$L__BB0_2:
mul.lo.s32 %r117, %r123, 112;
shl.b32 %r21, %r124, 1;
add.s32 %r22, %r21, %r13;
mul.hi.s32 %r23, %r22, -1840700269;
add.s32 %r24, %r23, %r22;
shr.u32 %r25, %r24, 31;
shr.s32 %r26, %r24, 6;
add.s32 %r27, %r26, %r25;
mul.lo.s32 %r28, %r27, 112;
sub.s32 %r29, %r22, %r28;
add.s32 %r30, %r29, %r117;
mad.lo.s32 %r31, %r27, 784, %r30;
mul.wide.s32 %rd9, %r31, 4;
add.s64 %rd10, %rd8, %rd9;
ld.global.nc.f32 %f377, [%rd10];
shl.b32 %r32, %r22, 2;
mov.u32 %r33, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r34, %r33, %r32;
for.size.14 i in range(0, 7, 1):
st.shared.f32 [%r34+(0+i*8)], %f(377+i*1);
add.s32 %r(35+i*10), %r22, (2+i*2);
mul.hi.s32 %r(36+i*10), %r(35+i*10), -1840700269;
add.s32 %r(37+i*10), %r(36+i*10), %r(35+i*10);
shr.u32 %r(38+i*10), %r(37+i*10), 31;
shr.s32 %r(39+i*10), %r(37+i*10), 6;
add.s32 %r(40+i*10), %r(39+i*10), %r(38+i*10);
mul.lo.s32 %r(41+i*10), %r(40+i*10), 112;
sub.s32 %r(42+i*10), %r(35+i*10), %r(41+i*10);
add.s32 %r(43+i*10), %r(42+i*10), %r117;
mad.lo.s32 %r(44+i*10), %r(40+i*10), 784, %r(43+i*10);
mul.wide.s32 %rd(11+i*2), %r(44+i*10), 4;
add.s64 %rd(12+i*2), %rd8, %rd(11+i*2);
ld.global.nc.f32 %f(378+i*1), [%rd(12+i*2)+0];
st.shared.f32 [%r34+56], %f384;
add.s32 %r124, %r124, 8;
setp.ne.s32 %p1, %r124, 7168;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r125, 0;
$L__BB0_4:
mov.u32 %r126, 0;
$L__BB0_5:
mul.lo.s32 %r121, %r125, 56;
mov.u32 %r119, _ZZ24default_function_kernel0E18placeholder_shared;
mov.u32 %r118, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.3 i in range(0, 2, 1):
add.s32 %r(107+i*4), %r(126+i*-19), %r(121+i*-120);
shl.b32 %r(108+i*4), %r(107+i*4), 2;
add.s32 %r(110+i*4), %r(118+i*1), %r(108+i*4);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(385+i*1), [%r(114+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(834+i*-1), %f386, %f(385+i*2), %f(834+i*-1);
ld.shared.f32 %f(387+i*1), [%r114+(28672+i*-28224)];
for.size.3 i in range(0, 30, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(832+i*-2+j*-1), %f386, %f(388+i*2+j*1), %f(832+i*-2+j*-1);
ld.shared.f32 %f(389+i*2+j*1), [%r114+(29120+i*448+j*-28224)];
fma.rn.f32 %f772, %f386, %f448, %f772;
ld.shared.f32 %f449, [%r114+42560];
fma.rn.f32 %f771, %f386, %f449, %f771;
add.s32 %r126, %r126, 1;
setp.ne.s32 %p2, %r126, 56;
@%p2 bra $L__BB0_5;
add.s32 %r125, %r125, 1;
setp.lt.u32 %p3, %r125, 2;
@%p3 bra $L__BB0_4;
add.s32 %r123, %r123, 1;
setp.lt.u32 %p4, %r123, 7;
@%p4 bra $L__BB0_1;
ld.param.u64 %rd31, [default_function_kernel0_param_2];
ld.param.u64 %rd30, [default_function_kernel0_param_3];
shl.b32 %r116, %r13, 5;
cvta.to.global.u64 %rd25, %rd30;
mul.wide.s32 %rd26, %r116, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f450, [%rd27];
add.f32 %f451, %f834, %f450;
mov.f32 %f452, 0f00000000;
max.f32 %f453, %f451, %f452;
cvta.to.global.u64 %rd28, %rd31;
add.s64 %rd29, %rd28, %rd26;
for.size.5 i in range(0, 31, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(0+i*4+j*256)], %f(453+i*6+j*3);
ld.global.nc.f32 %f(454+i*6+j*3), [%rd27+(256+i*4+j*-252)];
add.f32 %f(455+i*6+j*3), %f(833+i*-2+j*-1), %f(454+i*6+j*3);
max.f32 %f(456+i*6+j*3), %f(455+i*6+j*3), %f452;
st.global.f32 [%rd29+124], %f639;
ld.global.nc.f32 %f640, [%rd27+380];
add.f32 %f641, %f771, %f640;
max.f32 %f642, %f641, %f452;
st.global.f32 [%rd29+380], %f642;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[112];
__shared__ float placeholder_shared[14336];
T_dense[0] = 0.000000e+00f;
T_dense[32] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[33] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[34] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[35] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[36] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[37] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[38] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[39] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[40] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[41] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[42] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[43] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[44] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[45] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[46] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[47] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[48] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[49] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[50] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[51] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[52] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[53] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[54] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[55] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[56] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[57] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[58] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[59] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[60] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[61] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[62] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
T_dense[63] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 7; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 112 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
placeholder_d_shared[index0 + 6] = placeholder[index1 + 6];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 10] = placeholder[index1 + 10];
placeholder_d_shared[index0 + 12] = placeholder[index1 + 12];
placeholder_d_shared[index0 + 14] = placeholder[index1 + 14];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 18] = placeholder[index1 + 18];
placeholder_d_shared[index0 + 20] = placeholder[index1 + 20];
placeholder_d_shared[index0 + 22] = placeholder[index1 + 22];
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
placeholder_d_shared[index0 + 26] = placeholder[index1 + 26];
placeholder_d_shared[index0 + 28] = placeholder[index1 + 28];
placeholder_d_shared[index0 + 30] = placeholder[index1 + 30];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 34] = placeholder[index1 + 34];
placeholder_d_shared[index0 + 36] = placeholder[index1 + 36];
placeholder_d_shared[index0 + 38] = placeholder[index1 + 38];
placeholder_d_shared[index0 + 40] = placeholder[index1 + 40];
placeholder_d_shared[index0 + 42] = placeholder[index1 + 42];
placeholder_d_shared[index0 + 44] = placeholder[index1 + 44];
placeholder_d_shared[index0 + 46] = placeholder[index1 + 46];
placeholder_d_shared[index0 + 48] = placeholder[index1 + 48];
placeholder_d_shared[index0 + 50] = placeholder[index1 + 50];
placeholder_d_shared[index0 + 52] = placeholder[index1 + 52];
placeholder_d_shared[index0 + 54] = placeholder[index1 + 54];
placeholder_d_shared[index0 + 56] = placeholder[index1 + 56];
placeholder_d_shared[index0 + 58] = placeholder[index1 + 58];
placeholder_d_shared[index0 + 60] = placeholder[index1 + 60];
placeholder_d_shared[index0 + 62] = placeholder[index1 + 62];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 66] = placeholder[index1 + 66];
placeholder_d_shared[index0 + 68] = placeholder[index1 + 68];
placeholder_d_shared[index0 + 70] = placeholder[index1 + 70];
placeholder_d_shared[index0 + 72] = placeholder[index1 + 72];
placeholder_d_shared[index0 + 74] = placeholder[index1 + 74];
placeholder_d_shared[index0 + 76] = placeholder[index1 + 76];
placeholder_d_shared[index0 + 78] = placeholder[index1 + 78];
placeholder_d_shared[index0 + 80] = placeholder[index1 + 80];
placeholder_d_shared[index0 + 82] = placeholder[index1 + 82];
placeholder_d_shared[index0 + 84] = placeholder[index1 + 84];
placeholder_d_shared[index0 + 86] = placeholder[index1 + 86];
placeholder_d_shared[index0 + 88] = placeholder[index1 + 88];
placeholder_d_shared[index0 + 90] = placeholder[index1 + 90];
placeholder_d_shared[index0 + 92] = placeholder[index1 + 92];
placeholder_d_shared[index0 + 94] = placeholder[index1 + 94];
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
placeholder_d_shared[index0 + 98] = placeholder[index1 + 98];
placeholder_d_shared[index0 + 100] = placeholder[index1 + 100];
placeholder_d_shared[index0 + 102] = placeholder[index1 + 102];
placeholder_d_shared[index0 + 104] = placeholder[index1 + 104];
placeholder_d_shared[index0 + 106] = placeholder[index1 + 106];
placeholder_d_shared[index0 + 108] = placeholder[index1 + 108];
placeholder_d_shared[index0 + 110] = placeholder[index1 + 110];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 7168; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x) / 112 * 784 + k_outer_outer * 112 + (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x) % 112;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 56; ++k_inner) {
int index4 = k_outer_inner * 56 + k_inner;
int index5 = (int)threadIdx.x * 3584 + k_outer_inner * 56 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[32] = T_dense[32] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7168]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 112]);
T_dense[33] = T_dense[33] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7280]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 224]);
T_dense[34] = T_dense[34] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7392]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 336]);
T_dense[35] = T_dense[35] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7504]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 448]);
T_dense[36] = T_dense[36] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7616]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 560]);
T_dense[37] = T_dense[37] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7728]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 672]);
T_dense[38] = T_dense[38] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7840]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[39] = T_dense[39] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7952]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 896]);
T_dense[40] = T_dense[40] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8064]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1008]);
T_dense[41] = T_dense[41] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8176]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1120]);
T_dense[42] = T_dense[42] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8288]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1232]);
T_dense[43] = T_dense[43] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8400]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1344]);
T_dense[44] = T_dense[44] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8512]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1456]);
T_dense[45] = T_dense[45] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8624]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1568]);
T_dense[46] = T_dense[46] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8736]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1680]);
T_dense[47] = T_dense[47] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8848]);
T_dense[16] = T_dense[16] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1792]);
T_dense[48] = T_dense[48] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8960]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1904]);
T_dense[49] = T_dense[49] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9072]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2016]);
T_dense[50] = T_dense[50] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9184]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2128]);
T_dense[51] = T_dense[51] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9296]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2240]);
T_dense[52] = T_dense[52] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9408]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2352]);
T_dense[53] = T_dense[53] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9520]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2464]);
T_dense[54] = T_dense[54] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9632]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2576]);
T_dense[55] = T_dense[55] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9744]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2688]);
T_dense[56] = T_dense[56] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9856]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2800]);
T_dense[57] = T_dense[57] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9968]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2912]);
T_dense[58] = T_dense[58] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10080]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3024]);
T_dense[59] = T_dense[59] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10192]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3136]);
T_dense[60] = T_dense[60] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10304]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3248]);
T_dense[61] = T_dense[61] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10416]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3360]);
T_dense[62] = T_dense[62] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10528]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3472]);
T_dense[63] = T_dense[63] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10640]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 32; ++ax1_inner) {
int index6 = (int)threadIdx.x * 32 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 64] = max(T_dense[index7 + 32] + placeholder2[index6 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp_pu9j6an | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<6>;
.reg .f32 %f<82>;
.reg .b32 %r<20>;
.reg .b64 %rd<27>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[4];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[64];
ld.param.u64 %rd12, [default_function_kernel0_param_0];
ld.param.u64 %rd13, [default_function_kernel0_param_1];
ld.param.u64 %rd10, [default_function_kernel0_param_2];
ld.param.u64 %rd11, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd14, %rd13;
mov.u32 %r1, %tid.x;
shl.b32 %r3, %r1, 2;
mov.u32 %r11, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r11, %r3;
shl.b32 %r12, %r1, 4;
mov.u32 %r13, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r4, %r13, %r12;
mov.u32 %r5, %ctaid.x;
mul.lo.s32 %r14, %r1, 3136;
mad.lo.s32 %r15, %r5, 12544, %r14;
mul.wide.s32 %rd15, %r15, 4;
add.s64 %rd16, %rd14, %rd15;
add.s64 %rd26, %rd16, 6272;
mul.lo.s32 %r16, %r1, 784;
cvta.to.global.u64 %rd17, %rd12;
mul.wide.s32 %rd18, %r16, 4;
add.s64 %rd19, %rd17, %rd18;
add.s64 %rd25, %rd19, 8;
mov.u32 %r19, 0;
mov.f32 %f78, 0f00000000;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(79+i*1), %f78;
$L__BB0_1:
bar.sync 0;
add.s64 %rd5, %rd25, -8;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f25, [%rd5];
st.shared.f32 [%r2], %f25;
$L__BB0_3:
add.s64 %rd6, %rd26, -6272;
for.size.1 i in range(0, 4, 1):
ld.global.nc.f32 %f(26+i*1), [%rd26+(3136+i*-3136)];
st.shared.v4.f32 [%r4], {%f29, %f28, %f27, %f26};
bar.sync 0;
ld.shared.f32 %f30, [%r4];
ld.shared.f32 %f31, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(5+i*1), %f31, %f(30+i*2), %f(81+i*-1);
ld.shared.f32 %f(32+i*1), [%r4+(4+i*4)];
fma.rn.f32 %f7, %f31, %f33, %f79;
ld.shared.f32 %f34, [%r4+12];
fma.rn.f32 %f8, %f31, %f34, %f78;
bar.sync 0;
@%p1 bra $L__BB0_5;
ld.global.nc.f32 %f35, [%rd5+4];
st.shared.f32 [%r2], %f35;
$L__BB0_5:
for.size.1 i in range(0, 4, 1):
ld.global.nc.f32 %f(36+i*1), [%rd6+(9412+i*-3136)];
st.shared.v4.f32 [%r4], {%f39, %f38, %f37, %f36};
bar.sync 0;
ld.shared.f32 %f40, [%r4];
ld.shared.f32 %f41, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(9+i*1), %f41, %f(40+i*2), %f(5+i*1);
ld.shared.f32 %f(42+i*1), [%r4+(4+i*4)];
fma.rn.f32 %f11, %f41, %f43, %f7;
ld.shared.f32 %f44, [%r4+12];
fma.rn.f32 %f12, %f41, %f44, %f8;
bar.sync 0;
@%p1 bra $L__BB0_7;
ld.global.nc.f32 %f45, [%rd5+8];
st.shared.f32 [%r2], %f45;
$L__BB0_7:
for.size.1 i in range(0, 4, 1):
ld.global.nc.f32 %f(46+i*1), [%rd6+(9416+i*-3136)];
st.shared.v4.f32 [%r4], {%f49, %f48, %f47, %f46};
bar.sync 0;
ld.shared.f32 %f50, [%r4];
ld.shared.f32 %f51, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(13+i*1), %f51, %f(50+i*2), %f(9+i*1);
ld.shared.f32 %f(52+i*1), [%r4+(4+i*4)];
fma.rn.f32 %f15, %f51, %f53, %f11;
ld.shared.f32 %f54, [%r4+12];
fma.rn.f32 %f16, %f51, %f54, %f12;
bar.sync 0;
@%p1 bra $L__BB0_9;
ld.global.nc.f32 %f55, [%rd5+12];
st.shared.f32 [%r2], %f55;
$L__BB0_9:
for.size.1 i in range(0, 4, 1):
ld.global.nc.f32 %f(56+i*1), [%rd6+(9420+i*-3136)];
st.shared.v4.f32 [%r4], {%f59, %f58, %f57, %f56};
bar.sync 0;
ld.shared.f32 %f60, [%r4];
ld.shared.f32 %f61, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(81+i*-1), %f61, %f(60+i*2), %f(13+i*1);
ld.shared.f32 %f(62+i*1), [%r4+(4+i*4)];
fma.rn.f32 %f79, %f61, %f63, %f15;
ld.shared.f32 %f64, [%r4+12];
fma.rn.f32 %f78, %f61, %f64, %f16;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(26+i*-1), %rd(26+i*-1), 16;
add.s32 %r19, %r19, 4;
setp.ne.s32 %p5, %r19, 784;
@%p5 bra $L__BB0_1;
cvta.to.global.u64 %rd20, %rd11;
shl.b32 %r17, %r5, 4;
add.s32 %r18, %r17, %r3;
mul.wide.s32 %rd21, %r18, 4;
add.s64 %rd22, %rd20, %rd21;
ld.global.nc.f32 %f65, [%rd22];
add.f32 %f66, %f81, %f65;
mov.f32 %f67, 0f00000000;
max.f32 %f68, %f66, %f67;
cvta.to.global.u64 %rd23, %rd10;
add.s64 %rd24, %rd23, %rd21;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd24+(0+i*4)], %f(68+i*3);
ld.global.nc.f32 %f(69+i*3), [%rd22+(4+i*4)];
add.f32 %f(70+i*3), %f(80+i*-1), %f(69+i*3);
max.f32 %f(71+i*3), %f(70+i*3), %f67;
st.global.f32 [%rd24+12], %f77;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[16];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_d_shared[index0] = placeholder[index1];
}
int4 _1 = make_int4((int)blockIdx.x * 12544 + (int)threadIdx.x * 3136 + k_outer_outer+784*0, (int)blockIdx.x * 12544 + (int)threadIdx.x * 3136 + k_outer_outer+784*1, (int)blockIdx.x * 12544 + (int)threadIdx.x * 3136 + k_outer_outer+784*2, (int)blockIdx.x * 12544 + (int)threadIdx.x * 3136 + k_outer_outer+784*3);
int index2 = _1.x;
int index3 = _1.y;
int index4 = _1.z;
int index5 = _1.w;
((float4*)(placeholder_shared + (int)threadIdx.x * 4))[0] = make_float4(placeholder1[index2],placeholder1[index3],placeholder1[index4],placeholder1[index5]);
__syncthreads();
int index0 = (int)threadIdx.x;
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 4];
int index6 = (int)threadIdx.x * 4 + 1;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index6];
int index7 = (int)threadIdx.x * 4 + 2;
T_dense[2] = T_dense[2] + placeholder_d_shared[0] * placeholder_shared[index7];
int index8 = (int)threadIdx.x * 4 + 3;
T_dense[3] = T_dense[3] + placeholder_d_shared[0] * placeholder_shared[index8];
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index9 = (int)blockIdx.x * 16 + (int)threadIdx.x * 4 + ax1_inner;
int index10 = ax1_inner;
T_relu[index9] = max(T_dense[index10] + placeholder2[index9], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp56uqx0tl | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<67>;
.reg .b32 %r<91>;
.reg .b64 %rd<51>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[28];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[112];
ld.param.u64 %rd20, [default_function_kernel0_param_0];
ld.param.u64 %rd21, [default_function_kernel0_param_1];
ld.param.u64 %rd22, [default_function_kernel0_param_2];
ld.param.u64 %rd19, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd50, %rd21;
cvta.to.global.u64 %rd2, %rd22;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(9+i*1), %r1, (2+i*2);
mov.u32 %r11, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r11, %r10;
mov.u32 %r12, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r12, %r9;
add.s32 %r13, %r1, 6;
mul.hi.s32 %r14, %r13, -1840700269;
add.s32 %r15, %r14, %r13;
shr.u32 %r16, %r15, 31;
shr.s32 %r17, %r15, 2;
add.s32 %r18, %r17, %r16;
mul.lo.s32 %r19, %r18, 7;
sub.s32 %r20, %r13, %r19;
for.size.6 i in range(0, 3, 1):
add.s32 %r(21+i*6), %r1, (8+i*2);
mul.hi.s32 %r(22+i*6), %r(21+i*6), -1840700269;
add.s32 %r(23+i*6), %r(22+i*6), %r(21+i*6);
shr.u32 %r(24+i*6), %r(23+i*6), 31;
shr.s32 %r(25+i*6), %r(23+i*6), 2;
add.s32 %r(26+i*6), %r(25+i*6), %r(24+i*6);
for.size.6 i in range(0, 6, 1):
add.s32 %r(39+i*6), %r1, (16+i*2);
mul.hi.s32 %r(40+i*6), %r(39+i*6), -1840700269;
add.s32 %r(41+i*6), %r(40+i*6), %r(39+i*6);
shr.u32 %r(42+i*6), %r(41+i*6), 31;
shr.s32 %r(43+i*6), %r(41+i*6), 2;
add.s32 %r(44+i*6), %r(43+i*6), %r(42+i*6);
mad.lo.s32 %r4, %r1, 28, %r12;
mov.u32 %r5, %ctaid.x;
mul.lo.s32 %r75, %r5, 3136;
add.s32 %r76, %r1, %r75;
for.size.3 i in range(0, 3, 1):
mad.lo.s32 %r(77+i*1), %r(74+i*-6), 784, %r76;
mul.wide.s32 %rd(23+i*1), %r(77+i*1), 4;
add.s64 %rd(3+i*1), %rd(23+i*1), (20+i*-8);
add.s32 %r80, %r20, %r75;
for.size.2 i in range(0, 2, 1):
mad.lo.s32 %r(81+i*1), %r(56+i*-6), 784, %r(80+i*-4);
mul.wide.s32 %rd(6+i*20), %r(81+i*1), 4;
for.size.4 i in range(0, 2, 1):
for.size.3 j in range(0, 2, 1):
add.s64 %rd(7+i*2+j*1), %rd(26+i*2+j*1), (16+i*4+j*-8);
mad.lo.s32 %r(83+i*2+j*1), %r(44+i*-12+j*-6), 784, %r76;
mul.wide.s32 %rd(27+i*2+j*1), %r(83+i*2+j*1), 4;
add.s64 %rd11, %rd30, 4;
mad.lo.s32 %r87, %r18, 784, %r80;
mul.wide.s32 %rd12, %r87, 4;
cvta.to.global.u64 %rd31, %rd20;
mul.wide.s32 %rd32, %r9, 4;
add.s64 %rd33, %rd31, %rd32;
add.s64 %rd49, %rd33, 8;
mul.wide.s32 %rd14, %r76, 4;
mov.f32 %f65, 0f00000000;
mov.u32 %r90, 0;
mov.f32 %f66, %f65;
$L__BB0_1:
bar.sync 0;
for.size.2 i in range(0, 3, 1):
ld.global.nc.f32 %f(7+i*1), [%rd49+(-8+i*4)];
st.shared.f32 [%r2+(0+i*4)], %f(7+i*1);
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f10, [%rd49+4];
st.shared.f32 [%r2+12], %f10;
$L__BB0_3:
add.s64 %rd34, %rd50, %rd14;
for.size.2 i in range(0, 3, 1):
ld.global.nc.f32 %f(11+i*1), [%rd34+(0+i*8)];
st.shared.f32 [%r3+(0+i*8)], %f(11+i*1);
for.size.3 i in range(0, 4, 1):
add.s64 %rd(35+i*1), %rd50, %rd(12+i*-1);
ld.global.nc.f32 %f(14+i*1), [%rd(35+i*1)+0];
st.shared.f32 [%r3+(24+i*8)], %f(14+i*1);
ld.global.nc.f32 %f18, [%rd34+6272];
for.size.3 i in range(0, 6, 1):
st.shared.f32 [%r3+(56+i*8)], %f(18+i*1);
add.s64 %rd(39+i*1), %rd50, %rd(8+i*-1);
ld.global.nc.f32 %f(19+i*1), [%rd(39+i*1)+0];
st.shared.f32 [%r3+104], %f24;
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(25+i*5), [%r4+(0+i*4)];
ld.shared.f32 %f(26+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(27+i*5), %f(26+i*5), %f(25+i*5), %f(65+i*-38);
ld.shared.f32 %f(28+i*5), [%r4+(56+i*4)];
fma.rn.f32 %f(29+i*5), %f(26+i*5), %f(28+i*5), %f(66+i*-37);
for.size.5 i in range(0, 4, 1):
ld.shared.f32 %f(35+i*5), [%r4+(8+i*4)];
ld.shared.f32 %f(36+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(37+i*5), %f(36+i*5), %f(35+i*5), %f(32+i*5);
ld.shared.f32 %f(38+i*5), [%r4+(64+i*4)];
fma.rn.f32 %f(39+i*5), %f(36+i*5), %f(38+i*5), %f(34+i*5);
ld.shared.f32 %f55, [%r4+24];
ld.shared.f32 %f56, [_ZZ24default_function_kernel0E20placeholder_d_shared+24];
fma.rn.f32 %f65, %f56, %f55, %f52;
ld.shared.f32 %f57, [%r4+80];
fma.rn.f32 %f66, %f56, %f57, %f54;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(50+i*-1), %rd(50+i*-1), 28;
add.s32 %r90, %r90, 1;
setp.ne.s32 %p2, %r90, 112;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd45, %rd19;
shl.b32 %r88, %r5, 2;
add.s32 %r89, %r88, %r1;
mul.wide.s32 %rd46, %r89, 4;
add.s64 %rd47, %rd45, %rd46;
ld.global.nc.f32 %f58, [%rd47];
add.f32 %f59, %f65, %f58;
mov.f32 %f60, 0f00000000;
max.f32 %f61, %f59, %f60;
add.s64 %rd48, %rd2, %rd46;
st.global.f32 [%rd48], %f61;
ld.global.nc.f32 %f62, [%rd47+8];
add.f32 %f63, %f66, %f62;
max.f32 %f64, %f63, %f60;
st.global.f32 [%rd48+8], %f64;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[7];
__shared__ float placeholder_shared[28];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 112; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 7 + (int)threadIdx.x * 4;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
if ((int)threadIdx.x < 1) {
placeholder_d_shared[index0 + 3] = placeholder[index1 + 3];
}
int index2 = (int)threadIdx.x;
int index3 = (int)blockIdx.x * 3136 + k_outer_outer * 7 + (int)threadIdx.x;
placeholder_shared[index2] = placeholder1[index3];
placeholder_shared[index2 + 2] = placeholder1[index3 + 2];
placeholder_shared[index2 + 4] = placeholder1[index3 + 4];
int index4 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 6) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 6) % 7;
placeholder_shared[index2 + 6] = placeholder1[index4];
int index5 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 8) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 1);
placeholder_shared[index2 + 8] = placeholder1[index5];
int index6 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 10) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 3);
placeholder_shared[index2 + 10] = placeholder1[index6];
int index7 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 12) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 5);
placeholder_shared[index2 + 12] = placeholder1[index7];
placeholder_shared[index2 + 14] = placeholder1[index3 + 1568];
int index8 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 16) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 2);
placeholder_shared[index2 + 16] = placeholder1[index8];
int index9 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 18) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 4);
placeholder_shared[index2 + 18] = placeholder1[index9];
int index10 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 20) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 6) % 7;
placeholder_shared[index2 + 20] = placeholder1[index10];
int index11 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 22) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 1);
placeholder_shared[index2 + 22] = placeholder1[index11];
int index12 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 24) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 3);
placeholder_shared[index2 + 24] = placeholder1[index12];
int index13 = (int)blockIdx.x * 3136 + ((int)threadIdx.x + 26) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 5);
placeholder_shared[index2 + 26] = placeholder1[index13];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index2 * 7];
int index14 = (int)threadIdx.x * 7 + 14;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index14];
int index15 = (int)threadIdx.x * 7 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 7 + 15;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index16];
int index17 = (int)threadIdx.x * 7 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index17];
int index18 = (int)threadIdx.x * 7 + 16;
T_dense[1] = T_dense[1] + placeholder_d_shared[2] * placeholder_shared[index18];
int index19 = (int)threadIdx.x * 7 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index19];
int index20 = (int)threadIdx.x * 7 + 17;
T_dense[1] = T_dense[1] + placeholder_d_shared[3] * placeholder_shared[index20];
int index21 = (int)threadIdx.x * 7 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index21];
int index22 = (int)threadIdx.x * 7 + 18;
T_dense[1] = T_dense[1] + placeholder_d_shared[4] * placeholder_shared[index22];
int index23 = (int)threadIdx.x * 7 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index23];
int index24 = (int)threadIdx.x * 7 + 19;
T_dense[1] = T_dense[1] + placeholder_d_shared[5] * placeholder_shared[index24];
int index25 = (int)threadIdx.x * 7 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index25];
int index26 = (int)threadIdx.x * 7 + 20;
T_dense[1] = T_dense[1] + placeholder_d_shared[6] * placeholder_shared[index26];
}
int index27 = (int)blockIdx.x * 4 + (int)threadIdx.x;
T_relu[index27] = max(T_dense[0] + placeholder2[index27], 0.000000e+00f);
T_relu[index27 + 2] = max(T_dense[1] + placeholder2[index27 + 2], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpasbj8_ay | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<131>;
.reg .b32 %r<100>;
.reg .b64 %rd<22>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[392];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[6272];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
shl.b32 %r27, %r1, 2;
mov.u32 %r28, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r28, %r27;
mov.u32 %r3, %ctaid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(4+i*1), %r(3+i*-2), (12544+i*-12152);
shl.b32 %r6, %r1, 1;
add.s32 %r7, %r6, 24;
shl.b32 %r29, %r1, 3;
mov.u32 %r30, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r31, %r30, %r29;
add.s32 %r8, %r31, 64;
mov.u32 %r94, 0;
mov.f32 %f123, 0f00000000;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(124+i*1), %f123;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r10, %r94, 98;
add.s32 %r32, %r10, %r1;
mul.wide.s32 %rd10, %r32, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.2 i in range(0, 24, 1):
ld.global.nc.f32 %f(21+i*1), [%rd5+(0+i*16)];
st.shared.f32 [%r2+(0+i*16)], %f(21+i*1);
setp.gt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f45, [%rd5+384];
st.shared.f32 [%r2+384], %f45;
$L__BB0_3:
add.s32 %r12, %r10, %r4;
mov.u32 %r98, 64;
mov.u32 %r95, %r8;
mov.u32 %r96, %r6;
mov.u32 %r97, %r7;
$L__BB0_4:
for.size.13 i in range(0, 2, 1):
mul.hi.s32 %r(34+i*13), %r(96+i*-50), 1402438301;
shr.u32 %r(35+i*13), %r(34+i*13), 31;
shr.s32 %r(36+i*13), %r(34+i*13), 5;
add.s32 %r(37+i*13), %r(36+i*13), %r(35+i*13);
mul.lo.s32 %r(38+i*13), %r(37+i*13), 98;
sub.s32 %r(39+i*13), %r(96+i*-50), %r(38+i*13);
add.s32 %r(40+i*13), %r12, %r(39+i*13);
mad.lo.s32 %r(41+i*13), %r(37+i*13), 784, %r(40+i*13);
mul.wide.s32 %rd(11+i*2), %r(41+i*13), 4;
add.s64 %rd(12+i*2), %rd1, %rd(11+i*2);
ld.global.nc.v2.u32 {%r(42+i*13), %r(43+i*13)}, [%rd(12+i*2)+0];
st.shared.v2.u32 [%r95+(-64+i*32)], {%r(42+i*13), %r(43+i*13)};
add.s32 %r(46+i*13), %r97, (-16+i*8);
for.size.12 i in range(0, 2, 1):
mul.hi.s32 %r(60+i*12), %r(59+i*38), 1402438301;
shr.u32 %r(61+i*12), %r(60+i*12), 31;
shr.s32 %r(62+i*12), %r(60+i*12), 5;
add.s32 %r(63+i*12), %r(62+i*12), %r(61+i*12);
mul.lo.s32 %r(64+i*12), %r(63+i*12), 98;
sub.s32 %r(65+i*12), %r(59+i*38), %r(64+i*12);
add.s32 %r(66+i*12), %r12, %r(65+i*12);
mad.lo.s32 %r(67+i*12), %r(63+i*12), 784, %r(66+i*12);
mul.wide.s32 %rd(15+i*2), %r(67+i*12), 4;
add.s64 %rd(16+i*2), %rd1, %rd(15+i*2);
ld.global.nc.v2.u32 {%r(68+i*12), %r(69+i*12)}, [%rd(16+i*2)+0];
st.shared.v2.u32 [%r95+(0+i*32)], {%r(68+i*12), %r(69+i*12)};
for.size.1 i in range(0, 2, 1):
add.s32 %r(97+i*-1), %r(97+i*-1), 32;
for.size.1 i in range(0, 2, 1):
add.s32 %r(95+i*3), %r(95+i*3), 128;
setp.ne.s32 %p2, %r98, 6336;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r99, 0;
bra.uni $L__BB0_6;
$L__BB0_7:
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(60+i*1), [%r(23+i*-1)+8];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(62+i*2), %f61, %f(60+i*3), %f(126+i*-1);
ld.shared.f32 %f(63+i*2), [%r23+(400+i*392)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(66+i*2), %f61, %f(65+i*2), %f(124+i*-1);
ld.shared.f32 %f(67+i*2), [%r23+(1184+i*-1172)];
for.size.9 i in range(0, 4, 1):
ld.shared.f32 %f(70+i*9), [%r22+(12+i*4)];
fma.rn.f32 %f(71+i*9), %f(70+i*9), %f(69+i*9), %f(62+i*9);
ld.shared.f32 %f(72+i*9), [%r23+(404+i*4)];
fma.rn.f32 %f(73+i*9), %f(70+i*9), %f(72+i*9), %f(64+i*9);
ld.shared.f32 %f(74+i*9), [%r23+(796+i*4)];
fma.rn.f32 %f(75+i*9), %f(70+i*9), %f(74+i*9), %f(66+i*9);
ld.shared.f32 %f(76+i*9), [%r23+(1188+i*4)];
fma.rn.f32 %f(77+i*9), %f(70+i*9), %f(76+i*9), %f(68+i*9);
ld.shared.f32 %f(78+i*9), [%r23+(16+i*4)];
ld.shared.f32 %f106, [%r22+28];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(126+i*-1), %f106, %f(105+i*2), %f(98+i*2);
ld.shared.f32 %f(107+i*1), [%r23+(420+i*392)];
fma.rn.f32 %f124, %f106, %f108, %f102;
ld.shared.f32 %f109, [%r23+1204];
fma.rn.f32 %f123, %f106, %f109, %f104;
add.s32 %r99, %r99, 4;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(85+i*1), %r99, (1+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(22+i*66), %r(28+i*57), %r(86+i*-81);
shl.b32 %r89, %r88, 2;
add.s32 %r23, %r30, %r89;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(46+i*1), [%r(23+i*-1)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(48+i*2), %f47, %f(46+i*3), %f(126+i*-1);
ld.shared.f32 %f(49+i*2), [%r23+(392+i*392)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(52+i*2), %f47, %f(51+i*2), %f(124+i*-1);
ld.shared.f32 %f(53+i*2), [%r23+(1176+i*-1172)];
ld.shared.f32 %f56, [%r22+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(126+i*-1), %f56, %f(55+i*2), %f(48+i*2);
ld.shared.f32 %f(57+i*1), [%r23+(396+i*392)];
fma.rn.f32 %f124, %f56, %f58, %f52;
ld.shared.f32 %f59, [%r23+1180];
fma.rn.f32 %f123, %f56, %f59, %f54;
setp.eq.s32 %p3, %r99, 48;
@%p3 bra $L__BB0_8;
bra.uni $L__BB0_7;
$L__BB0_8:
add.s32 %r94, %r94, 1;
setp.lt.u32 %p4, %r94, 8;
@%p4 bra $L__BB0_1;
shl.b32 %r91, %r3, 4;
add.s32 %r93, %r91, %r27;
mul.wide.s32 %rd19, %r93, 4;
add.s64 %rd20, %rd4, %rd19;
ld.global.nc.f32 %f110, [%rd20];
add.f32 %f111, %f126, %f110;
mov.f32 %f112, 0f00000000;
max.f32 %f113, %f111, %f112;
add.s64 %rd21, %rd3, %rd19;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd21+(0+i*4)], %f(113+i*3);
ld.global.nc.f32 %f(114+i*3), [%rd20+(4+i*4)];
add.f32 %f(115+i*3), %f(125+i*-1), %f(114+i*3);
max.f32 %f(116+i*3), %f(115+i*3), %f112;
st.global.f32 [%rd21+12], %f122;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[98];
__shared__ float placeholder_shared[1568];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 8; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 98 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 12] = placeholder[index1 + 12];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 20] = placeholder[index1 + 20];
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
placeholder_d_shared[index0 + 28] = placeholder[index1 + 28];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 36] = placeholder[index1 + 36];
placeholder_d_shared[index0 + 40] = placeholder[index1 + 40];
placeholder_d_shared[index0 + 44] = placeholder[index1 + 44];
placeholder_d_shared[index0 + 48] = placeholder[index1 + 48];
placeholder_d_shared[index0 + 52] = placeholder[index1 + 52];
placeholder_d_shared[index0 + 56] = placeholder[index1 + 56];
placeholder_d_shared[index0 + 60] = placeholder[index1 + 60];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 68] = placeholder[index1 + 68];
placeholder_d_shared[index0 + 72] = placeholder[index1 + 72];
placeholder_d_shared[index0 + 76] = placeholder[index1 + 76];
placeholder_d_shared[index0 + 80] = placeholder[index1 + 80];
placeholder_d_shared[index0 + 84] = placeholder[index1 + 84];
placeholder_d_shared[index0 + 88] = placeholder[index1 + 88];
placeholder_d_shared[index0 + 92] = placeholder[index1 + 92];
if ((int)threadIdx.x < 2) {
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 196; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x * 2) / 98 * 784 + k_outer_outer * 98 + (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x * 2) % 98))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 49; ++k_outer_inner) {
int index2 = k_outer_inner * 2;
int index3 = (int)threadIdx.x * 392 + k_outer_inner * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 98]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 196]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 294]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 99]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 197]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 295]);
}
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index4 = (int)blockIdx.x * 16 + (int)threadIdx.x * 4 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpc6ax8elt | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<71>;
.reg .b32 %r<97>;
.reg .b64 %rd<18>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[1568];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[100352];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd5, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
mov.u32 %r1, %tid.x;
shl.b32 %r21, %r1, 2;
mov.u32 %r22, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r22, %r21;
mov.u32 %r23, %ctaid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(3+i*1), %r(23+i*-22), (50176+i*-49392);
shl.b32 %r24, %r1, 3;
add.s32 %r5, %r24, 256;
shl.b32 %r25, %r1, 5;
mov.u32 %r26, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r27, %r26, %r25;
add.s32 %r6, %r27, 1040;
mov.u32 %r92, 0;
mov.f32 %f67, 0f00000000;
mov.f32 %f68, %f67;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r92, 392;
add.s32 %r28, %r8, %r1;
mul.wide.s32 %rd8, %r28, 4;
add.s64 %rd3, %rd2, %rd8;
for.size.2 i in range(0, 12, 1):
ld.global.nc.f32 %f(9+i*1), [%rd3+(0+i*128)];
st.shared.f32 [%r2+(0+i*128)], %f(9+i*1);
setp.gt.s32 %p1, %r1, 7;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f21, [%rd3+1536];
st.shared.f32 [%r2+1536], %f21;
$L__BB0_3:
add.s32 %r10, %r8, %r3;
mov.u32 %r95, 1024;
mov.u32 %r93, %r6;
mov.u32 %r94, %r5;
$L__BB0_4:
add.s32 %r30, %r94, -256;
for.size.14 i in range(0, 2, 1):
mul.hi.s32 %r(31+i*24), %r(30+i*64), 1402438301;
shr.u32 %r(32+i*24), %r(31+i*24), 31;
shr.s32 %r(33+i*24), %r(31+i*24), 7;
add.s32 %r(34+i*24), %r(33+i*24), %r(32+i*24);
mul.lo.s32 %r(35+i*24), %r(34+i*24), 392;
sub.s32 %r(36+i*24), %r(30+i*64), %r(35+i*24);
add.s32 %r(37+i*24), %r10, %r(36+i*24);
mad.lo.s32 %r(38+i*24), %r(34+i*24), 784, %r(37+i*24);
mul.wide.s32 %rd(9+i*2), %r(38+i*24), 4;
add.s64 %rd(10+i*2), %rd1, %rd(9+i*2);
ld.global.nc.v4.u32 {%r(39+i*24), %r(40+i*24), %r(41+i*24), %r(42+i*24)}, [%rd(10+i*2)+0];
ld.global.nc.v4.u32 {%r(47+i*24), %r(48+i*24), %r(49+i*24), %r(50+i*24)}, [%rd(10+i*2)+16];
st.shared.v4.u32 [%r93+(-1040+i*1024)], {%r(39+i*24), %r(40+i*24), %r(41+i*24), %r(42+i*24)};
st.shared.v4.u32 [%r93+(-1024+i*1024)], {%r(47+i*24), %r(48+i*24), %r(49+i*24), %r(50+i*24)};
for.size.1 i in range(0, 2, 1):
add.s32 %r(94+i*-1), %r(94+i*-1), (512+i*1536);
add.s32 %r95, %r95, 2048;
setp.ne.s32 %p2, %r95, 101376;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r96, 0;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(80+i*1), %r96, (1+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(83+i*1), %r(22+i*58), %r(81+i*-77);
shl.b32 %r85, %r84, 2;
add.s32 %r87, %r26, %r85;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(22+i*1), [%r(87+i*-4)+0];
for.size.10 i in range(0, 2, 1):
fma.rn.f32 %f(24+i*10), %f(23+i*10), %f(22+i*10), %f(68+i*-41);
ld.shared.f32 %f(25+i*10), [%r87+(4+i*8)];
ld.shared.f32 %f(26+i*10), [%r83+(4+i*8)];
fma.rn.f32 %f(27+i*10), %f(26+i*10), %f(25+i*10), %f(24+i*10);
ld.shared.f32 %f(28+i*10), [%r87+(1568+i*8)];
fma.rn.f32 %f(29+i*10), %f(23+i*10), %f(28+i*10), %f(67+i*-36);
ld.shared.f32 %f(30+i*10), [%r87+(1572+i*8)];
fma.rn.f32 %f(31+i*10), %f(26+i*10), %f(30+i*10), %f(29+i*10);
ld.shared.f32 %f(32+i*10), [%r87+(8+i*8)];
ld.shared.f32 %f(33+i*10), [%r83+(8+i*8)];
fma.rn.f32 %f44, %f43, %f42, %f37;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(45+i*1), [%r(87+i*-4)+20];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(47+i*2), %f(46+i*-3), %f(45+i*3), %f(44+i*-3);
ld.shared.f32 %f(48+i*2), [%r87+(1584+i*4)];
for.size.3 i in range(0, 2, 1):
fma.rn.f32 %f(51+i*3), %f(46+i*7), %f(50+i*2), %f(49+i*-2);
ld.shared.f32 %f(52+i*3), [%r87+(24+i*4)];
ld.shared.f32 %f(53+i*3), [%r83+(24+i*4)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(68+i*-10), %f(56+i*-3), %f(55+i*2), %f(54+i*-3);
ld.shared.f32 %f(57+i*2), [%r87+(1592+i*4)];
fma.rn.f32 %f67, %f56, %f59, %f58;
add.s32 %r96, %r96, 4;
setp.ne.s32 %p3, %r96, 196;
@%p3 bra $L__BB0_6;
add.s32 %r92, %r92, 1;
setp.lt.u32 %p4, %r92, 2;
@%p4 bra $L__BB0_1;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(89+i*1), %r(23+i*-22), (6+i*-5);
add.s32 %r91, %r89, %r90;
cvta.to.global.u64 %rd13, %rd5;
mul.wide.s32 %rd14, %r91, 4;
add.s64 %rd15, %rd13, %rd14;
ld.global.nc.f32 %f60, [%rd15];
add.f32 %f61, %f68, %f60;
mov.f32 %f62, 0f00000000;
max.f32 %f63, %f61, %f62;
cvta.to.global.u64 %rd16, %rd4;
add.s64 %rd17, %rd16, %rd14;
st.global.f32 [%rd17], %f63;
ld.global.nc.f32 %f64, [%rd15+4];
add.f32 %f65, %f67, %f64;
max.f32 %f66, %f65, %f62;
st.global.f32 [%rd17+4], %f66;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[392];
__shared__ float placeholder_shared[25088];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 2; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 392 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
placeholder_d_shared[index0 + 128] = placeholder[index1 + 128];
placeholder_d_shared[index0 + 160] = placeholder[index1 + 160];
placeholder_d_shared[index0 + 192] = placeholder[index1 + 192];
placeholder_d_shared[index0 + 224] = placeholder[index1 + 224];
placeholder_d_shared[index0 + 256] = placeholder[index1 + 256];
placeholder_d_shared[index0 + 288] = placeholder[index1 + 288];
placeholder_d_shared[index0 + 320] = placeholder[index1 + 320];
placeholder_d_shared[index0 + 352] = placeholder[index1 + 352];
if ((int)threadIdx.x < 8) {
placeholder_d_shared[index0 + 384] = placeholder[index1 + 384];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
ulonglong4*(placeholder_shared + ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 8)[0] = ulonglong4*(placeholder1 + (int)blockIdx.x * 50176 + (ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 8) / 392 * 784 + k_outer_outer * 392 + (ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 8) % 392)[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
int index2 = k_outer_inner * 2;
int index3 = (int)threadIdx.x * 784 + k_outer_inner * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 392]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 393]);
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index4 = (int)blockIdx.x * 64 + (int)threadIdx.x * 2 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpjaj0lm38 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<651>;
.reg .b32 %r<128>;
.reg .b64 %rd<38>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[1568];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd8, [default_function_kernel0_param_0];
ld.param.u64 %rd9, [default_function_kernel0_param_1];
cvta.to.global.u64 %rd12, %rd8;
cvta.to.global.u64 %rd2, %rd9;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 25088;
shl.b32 %r17, %r1, 3;
mov.u32 %r18, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r4, %r18, %r17;
add.s64 %rd4, %rd12, 16;
mov.u32 %r123, 0;
mov.f32 %f523, 0f00000000;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(524+i*1), %f523;
$L__BB0_1:
mov.u32 %r124, 0;
mov.u32 %r121, %tid.x;
shl.b32 %r120, %r121, 1;
bar.sync 0;
mul.lo.s32 %r6, %r123, 392;
add.s32 %r21, %r120, %r6;
mul.wide.s32 %rd13, %r21, 4;
add.s64 %rd37, %rd4, %rd13;
mov.u32 %r125, 16;
$L__BB0_2:
ld.global.nc.v2.u32 {%r22, %r23}, [%rd37+-16];
add.s32 %r26, %r4, %r124;
st.shared.v2.u32 [%r26], {%r22, %r23};
ld.global.nc.v2.u32 {%r27, %r28}, [%rd37];
st.shared.v2.u32 [%r26+16], {%r27, %r28};
add.s64 %rd37, %rd37, 32;
add.s32 %r9, %r125, 32;
setp.ne.s32 %p1, %r9, 1584;
add.s32 %r124, %r125, 16;
mov.u32 %r125, %r9;
@%p1 bra $L__BB0_2;
mov.u32 %r126, 0;
$L__BB0_4:
mul.lo.s32 %r117, %r123, 392;
shl.b32 %r32, %r126, 1;
add.s32 %r33, %r32, %r1;
mul.hi.s32 %r34, %r33, 1402438301;
shr.u32 %r35, %r34, 31;
shr.s32 %r36, %r34, 7;
add.s32 %r37, %r36, %r35;
mul.lo.s32 %r38, %r37, 392;
sub.s32 %r39, %r33, %r38;
add.s32 %r40, %r39, %r117;
mad.lo.s32 %r41, %r37, 784, %r40;
mul.wide.s32 %rd14, %r41, 4;
add.s64 %rd15, %rd2, %rd14;
ld.global.nc.f32 %f257, [%rd15];
shl.b32 %r42, %r33, 2;
mov.u32 %r43, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r44, %r43, %r42;
for.size.13 i in range(0, 7, 1):
st.shared.f32 [%r44+(0+i*8)], %f(257+i*1);
add.s32 %r(45+i*9), %r33, (2+i*2);
mul.hi.s32 %r(46+i*9), %r(45+i*9), 1402438301;
shr.u32 %r(47+i*9), %r(46+i*9), 31;
shr.s32 %r(48+i*9), %r(46+i*9), 7;
add.s32 %r(49+i*9), %r(48+i*9), %r(47+i*9);
mul.lo.s32 %r(50+i*9), %r(49+i*9), 392;
sub.s32 %r(51+i*9), %r(45+i*9), %r(50+i*9);
add.s32 %r(52+i*9), %r(51+i*9), %r117;
mad.lo.s32 %r(53+i*9), %r(49+i*9), 784, %r(52+i*9);
mul.wide.s32 %rd(16+i*2), %r(53+i*9), 4;
add.s64 %rd(17+i*2), %rd2, %rd(16+i*2);
ld.global.nc.f32 %f(258+i*1), [%rd(17+i*2)+0];
st.shared.f32 [%r44+56], %f264;
add.s32 %r126, %r126, 8;
setp.ne.s32 %p2, %r126, 25088;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r127, 0;
$L__BB0_6:
mov.u32 %r119, _ZZ24default_function_kernel0E20placeholder_d_shared;
mov.u32 %r118, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r109, %r127, %r2;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(110+i*3), %r(109+i*18), 2;
add.s32 %r(112+i*3), %r(118+i*1), %r(110+i*3);
ld.shared.f32 %f(265+i*1), [%r(112+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(586+i*-1), %f266, %f(265+i*2), %f(586+i*-1);
ld.shared.f32 %f(267+i*1), [%r112+(1568+i*1568)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(584+i*-1), %f266, %f(268+i*1), %f(584+i*-1);
ld.shared.f32 %f(269+i*1), [%r112+(4704+i*1568)];
fma.rn.f32 %f523, %f266, %f329, %f523;
add.s32 %r127, %r127, 1;
setp.ne.s32 %p3, %r127, 392;
@%p3 bra $L__BB0_6;
add.s32 %r123, %r123, 1;
setp.lt.u32 %p4, %r123, 2;
@%p4 bra $L__BB0_1;
ld.param.u64 %rd36, [default_function_kernel0_param_2];
cvta.to.global.u64 %rd35, %rd36;
ld.param.u64 %rd34, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd33, %rd34;
shl.b32 %r116, %r1, 6;
mul.wide.s32 %rd30, %r116, 4;
add.s64 %rd31, %rd33, %rd30;
ld.global.nc.f32 %f330, [%rd31];
add.f32 %f331, %f586, %f330;
mov.f32 %f332, 0f00000000;
max.f32 %f333, %f331, %f332;
add.s64 %rd32, %rd35, %rd30;
for.size.4 i in range(0, 63, 1):
st.global.f32 [%rd32+(0+i*4)], %f(333+i*3);
ld.global.nc.f32 %f(334+i*3), [%rd31+(4+i*4)];
add.f32 %f(335+i*3), %f(585+i*-1), %f(334+i*3);
max.f32 %f(336+i*3), %f(335+i*3), %f332;
st.global.f32 [%rd32+252], %f522;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[392];
__shared__ float placeholder_shared[50176];
for (int j_inner_init = 0; j_inner_init < 64; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 2; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_d_shared + ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder + k_outer_outer * 392 + ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x * 2))[0];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 25088; ++ax0_ax1_fused_outer_outer1) {
int index1 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
int index2 = (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) / 392 * 784 + k_outer_outer * 392 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) % 392;
placeholder_shared[index1] = placeholder1[index2];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 392; ++k_outer_inner) {
for (int j_inner = 0; j_inner < 64; ++j_inner) {
int index3 = j_inner;
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 25088 + j_inner * 392 + k_outer_inner;
T_dense[index3] = T_dense[index3] + placeholder_d_shared[index4] * placeholder_shared[index5];
}
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index6 = (int)threadIdx.x * 64 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpzybl8g4k | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<87>;
.reg .f32 %f<82>;
.reg .b32 %r<94>;
.reg .b64 %rd<20>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[401408];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd11, %rd10;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 28;
for.size.1 i in range(0, 2, 1):
setp.gt.s32 %p(1+i*1), %r(1+i*1), (27+i*756);
mul.wide.s32 %rd12, %r2, 4;
add.s64 %rd1, %rd11, %rd12;
shl.b32 %r14, %r2, 2;
mov.u32 %r15, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r15, %r14;
or.pred %p3, %p1, %p2;
@%p3 bra $L__BB0_2;
ld.global.nc.f32 %f5, [%rd1];
st.shared.f32 [%r3], %f5;
$L__BB0_2:
setp.gt.s32 %p5, %r2, 782;
or.pred %p6, %p1, %p5;
@%p6 bra $L__BB0_4;
ld.global.nc.f32 %f6, [%rd1+4];
st.shared.f32 [%r3+4], %f6;
$L__BB0_4:
setp.gt.s32 %p8, %r2, 781;
or.pred %p9, %p1, %p8;
@%p9 bra $L__BB0_6;
ld.global.nc.f32 %f7, [%rd1+8];
st.shared.f32 [%r3+8], %f7;
$L__BB0_6:
setp.gt.s32 %p11, %r2, 780;
or.pred %p12, %p1, %p11;
@%p12 bra $L__BB0_8;
ld.global.nc.f32 %f8, [%rd1+12];
st.shared.f32 [%r3+12], %f8;
$L__BB0_8:
setp.gt.s32 %p14, %r2, 779;
or.pred %p15, %p1, %p14;
@%p15 bra $L__BB0_10;
ld.global.nc.f32 %f9, [%rd1+16];
st.shared.f32 [%r3+16], %f9;
$L__BB0_10:
setp.gt.s32 %p17, %r2, 778;
or.pred %p18, %p1, %p17;
@%p18 bra $L__BB0_12;
ld.global.nc.f32 %f10, [%rd1+20];
st.shared.f32 [%r3+20], %f10;
$L__BB0_12:
setp.gt.s32 %p20, %r2, 777;
or.pred %p21, %p1, %p20;
@%p21 bra $L__BB0_14;
ld.global.nc.f32 %f11, [%rd1+24];
st.shared.f32 [%r3+24], %f11;
$L__BB0_14:
setp.gt.s32 %p23, %r2, 776;
or.pred %p24, %p1, %p23;
@%p24 bra $L__BB0_16;
ld.global.nc.f32 %f12, [%rd1+28];
st.shared.f32 [%r3+28], %f12;
$L__BB0_16:
setp.gt.s32 %p26, %r2, 775;
or.pred %p27, %p1, %p26;
@%p27 bra $L__BB0_18;
ld.global.nc.f32 %f13, [%rd1+32];
st.shared.f32 [%r3+32], %f13;
$L__BB0_18:
setp.gt.s32 %p29, %r2, 774;
or.pred %p30, %p1, %p29;
@%p30 bra $L__BB0_20;
ld.global.nc.f32 %f14, [%rd1+36];
st.shared.f32 [%r3+36], %f14;
$L__BB0_20:
setp.gt.s32 %p32, %r2, 773;
or.pred %p33, %p1, %p32;
@%p33 bra $L__BB0_22;
ld.global.nc.f32 %f15, [%rd1+40];
st.shared.f32 [%r3+40], %f15;
$L__BB0_22:
setp.gt.s32 %p35, %r2, 772;
or.pred %p36, %p1, %p35;
@%p36 bra $L__BB0_24;
ld.global.nc.f32 %f16, [%rd1+44];
st.shared.f32 [%r3+44], %f16;
$L__BB0_24:
setp.gt.s32 %p38, %r2, 771;
or.pred %p39, %p1, %p38;
@%p39 bra $L__BB0_26;
ld.global.nc.f32 %f17, [%rd1+48];
st.shared.f32 [%r3+48], %f17;
$L__BB0_26:
setp.gt.s32 %p41, %r2, 770;
or.pred %p42, %p1, %p41;
@%p42 bra $L__BB0_28;
ld.global.nc.f32 %f18, [%rd1+52];
st.shared.f32 [%r3+52], %f18;
$L__BB0_28:
setp.gt.s32 %p44, %r2, 769;
or.pred %p45, %p1, %p44;
@%p45 bra $L__BB0_30;
ld.global.nc.f32 %f19, [%rd1+56];
st.shared.f32 [%r3+56], %f19;
$L__BB0_30:
setp.gt.s32 %p47, %r2, 768;
or.pred %p48, %p1, %p47;
@%p48 bra $L__BB0_32;
ld.global.nc.f32 %f20, [%rd1+60];
st.shared.f32 [%r3+60], %f20;
$L__BB0_32:
setp.gt.s32 %p50, %r2, 767;
or.pred %p51, %p1, %p50;
@%p51 bra $L__BB0_34;
ld.global.nc.f32 %f21, [%rd1+64];
st.shared.f32 [%r3+64], %f21;
$L__BB0_34:
setp.gt.s32 %p53, %r2, 766;
or.pred %p54, %p1, %p53;
@%p54 bra $L__BB0_36;
ld.global.nc.f32 %f22, [%rd1+68];
st.shared.f32 [%r3+68], %f22;
$L__BB0_36:
setp.gt.s32 %p56, %r2, 765;
or.pred %p57, %p1, %p56;
@%p57 bra $L__BB0_38;
ld.global.nc.f32 %f23, [%rd1+72];
st.shared.f32 [%r3+72], %f23;
$L__BB0_38:
setp.gt.s32 %p59, %r2, 764;
or.pred %p60, %p1, %p59;
@%p60 bra $L__BB0_40;
ld.global.nc.f32 %f24, [%rd1+76];
st.shared.f32 [%r3+76], %f24;
$L__BB0_40:
setp.gt.s32 %p62, %r2, 763;
or.pred %p63, %p1, %p62;
@%p63 bra $L__BB0_42;
ld.global.nc.f32 %f25, [%rd1+80];
st.shared.f32 [%r3+80], %f25;
$L__BB0_42:
setp.gt.s32 %p65, %r2, 762;
or.pred %p66, %p1, %p65;
@%p66 bra $L__BB0_44;
ld.global.nc.f32 %f26, [%rd1+84];
st.shared.f32 [%r3+84], %f26;
$L__BB0_44:
setp.gt.s32 %p68, %r2, 761;
or.pred %p69, %p1, %p68;
@%p69 bra $L__BB0_46;
ld.global.nc.f32 %f27, [%rd1+88];
st.shared.f32 [%r3+88], %f27;
$L__BB0_46:
setp.gt.s32 %p71, %r2, 760;
or.pred %p72, %p1, %p71;
@%p72 bra $L__BB0_48;
ld.global.nc.f32 %f28, [%rd1+92];
st.shared.f32 [%r3+92], %f28;
$L__BB0_48:
setp.gt.s32 %p74, %r2, 759;
or.pred %p75, %p1, %p74;
@%p75 bra $L__BB0_50;
ld.global.nc.f32 %f29, [%rd1+96];
st.shared.f32 [%r3+96], %f29;
$L__BB0_50:
setp.gt.s32 %p77, %r2, 758;
or.pred %p78, %p1, %p77;
@%p78 bra $L__BB0_52;
ld.global.nc.f32 %f30, [%rd1+100];
st.shared.f32 [%r3+100], %f30;
$L__BB0_52:
setp.gt.s32 %p80, %r2, 757;
or.pred %p81, %p1, %p80;
@%p81 bra $L__BB0_54;
ld.global.nc.f32 %f31, [%rd1+104];
st.shared.f32 [%r3+104], %f31;
$L__BB0_54:
setp.gt.s32 %p83, %r2, 756;
or.pred %p84, %p1, %p83;
@%p84 bra $L__BB0_56;
ld.global.nc.f32 %f32, [%rd1+108];
st.shared.f32 [%r3+108], %f32;
$L__BB0_56:
cvta.to.global.u64 %rd13, %rd7;
cvta.to.global.u64 %rd2, %rd9;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(17+i*1), %r1, (1+i*2);
mov.u32 %r19, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r20, %r19, %r18;
add.s32 %r90, %r20, 4096;
mul.wide.s32 %rd14, %r17, 4;
add.s64 %rd15, %rd13, %rd14;
add.s64 %rd19, %rd15, 4096;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r91, 0;
$L__BB0_57:
for.size.2 i in range(0, 16, 1):
ld.global.nc.v2.u32 {%r(21+i*4), %r(22+i*4)}, [%rd19+(-4096+i*512)];
st.shared.v2.u32 [%r90+(-4096+i*512)], {%r(21+i*4), %r(22+i*4)};
add.s32 %r90, %r90, 8192;
add.s64 %rd19, %rd19, 8192;
add.s32 %r91, %r91, 16;
setp.ne.s32 %p85, %r91, 784;
@%p85 bra $L__BB0_57;
bar.sync 0;
mad.lo.s32 %r87, %r1, 3136, %r19;
add.s32 %r92, %r87, 200732;
mov.f32 %f80, 0f00000000;
mov.u32 %r93, 200704;
mov.f32 %f81, %f80;
$L__BB0_59:
add.s32 %r89, %r15, %r93;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(35+i*5), [%r92+(-200732+i*4)];
ld.shared.f32 %f(36+i*5), [%r89+(-200704+i*4)];
fma.rn.f32 %f(37+i*5), %f(36+i*5), %f(35+i*5), %f(80+i*-43);
ld.shared.f32 %f(38+i*5), [%r92+(-28+i*4)];
fma.rn.f32 %f(39+i*5), %f(36+i*5), %f(38+i*5), %f(81+i*-42);
for.size.5 i in range(0, 5, 1):
ld.shared.f32 %f(45+i*5), [%r92+(-200724+i*4)];
ld.shared.f32 %f(46+i*5), [%r89+(-200696+i*4)];
fma.rn.f32 %f(47+i*5), %f(46+i*5), %f(45+i*5), %f(42+i*5);
ld.shared.f32 %f(48+i*5), [%r92+(-20+i*4)];
fma.rn.f32 %f(49+i*5), %f(46+i*5), %f(48+i*5), %f(44+i*5);
ld.shared.f32 %f70, [%r92+-200704];
ld.shared.f32 %f71, [%r89+-200676];
fma.rn.f32 %f80, %f71, %f70, %f67;
ld.shared.f32 %f72, [%r92];
fma.rn.f32 %f81, %f71, %f72, %f69;
for.size.1 i in range(0, 2, 1):
add.s32 %r(92+i*1), %r(92+i*1), 32;
setp.ne.s32 %p86, %r93, 203840;
@%p86 bra $L__BB0_59;
mul.wide.s32 %rd16, %r1, 4;
add.s64 %rd17, %rd2, %rd16;
ld.global.nc.f32 %f73, [%rd17];
add.f32 %f74, %f80, %f73;
mov.f32 %f75, 0f00000000;
max.f32 %f76, %f74, %f75;
add.s64 %rd18, %rd4, %rd16;
st.global.f32 [%rd18], %f76;
ld.global.nc.f32 %f77, [%rd17+256];
add.f32 %f78, %f81, %f77;
max.f32 %f79, %f78, %f75;
st.global.f32 [%rd18+256], %f79;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[100352];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int ax0_ax1_fused_inner_s = 0; ax0_ax1_fused_inner_s < 28; ++ax0_ax1_fused_inner_s) {
if ((int)threadIdx.x * 28 + ax0_ax1_fused_inner_s < 784) {
if ((int)threadIdx.x < 28) {
int index0 = (int)threadIdx.x * 28 + ax0_ax1_fused_inner_s;
placeholder_d_shared[index0] = placeholder[index0];
}
}
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 784; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer * 128 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder1 + ax0_ax1_fused_outer_outer * 128 + (int)threadIdx.x * 2))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 98; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 8; ++k_inner) {
int index1 = k_outer_inner * 8 + k_inner;
int index2 = (int)threadIdx.x * 784 + k_outer_inner * 8 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index1] * placeholder_shared[index2];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index1] * placeholder_shared[index2 + 50176]);
}
}
int index3 = (int)threadIdx.x;
T_relu[index3] = max(T_dense[0] + placeholder2[index3], 0.000000e+00f);
T_relu[index3 + 64] = max(T_dense[1] + placeholder2[index3 + 64], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmppi9d78sx | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<643>;
.reg .b32 %r<1060>;
.reg .b64 %rd<18>;
// demoted variable
.shared .align 16 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 16 .b8 _ZZ24default_function_kernel0E18placeholder_shared[2048];
ld.param.u64 %rd1, [default_function_kernel0_param_0];
ld.param.u64 %rd2, [default_function_kernel0_param_1];
mov.u32 %r1058, 0;
mov.f32 %f579, 0f00000000;
cvta.to.global.u64 %rd5, %rd1;
mov.u32 %r24, %ctaid.x;
cvta.to.global.u64 %rd8, %rd2;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(580+i*1), %f579;
$L__BB0_1:
mov.u32 %r1059, 0;
bar.sync 0;
shl.b32 %r7, %r1058, 3;
mul.wide.s32 %rd6, %r7, 4;
add.s64 %rd7, %rd5, %rd6;
for.size.2 i in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(8+i*8), %r(9+i*8), %r(10+i*8), %r(11+i*8)}, [%rd7+(0+i*16)];
st.shared.v4.u32 [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*16)], {%r(8+i*8), %r(9+i*8), %r(10+i*8), %r(11+i*8)};
mad.lo.s32 %r25, %r24, 50176, %r7;
mul.wide.s32 %rd9, %r25, 4;
add.s64 %rd10, %rd8, %rd9;
for.size.4 i in range(0, 64, 1):
for.size.1 j in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(26+i*16+j*8), %r(27+i*16+j*8), %r(28+i*16+j*8), %r(29+i*16+j*8)}, [%rd10+(0+i*3136+j*16)];
for.size.1 j in range(0, 2, 1):
st.shared.v4.u32 [_ZZ24default_function_kernel0E18placeholder_shared+(0+i*32+j*16)], {%r(26+i*16+j*8), %r(27+i*16+j*8), %r(28+i*16+j*8), %r(29+i*16+j*8)};
bar.sync 0;
$L__BB0_2:
shl.b32 %r1050, %r1059, 2;
mov.u32 %r1051, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r1052, %r1051, %r1050;
ld.shared.f32 %f257, [%r1052];
mov.u32 %r1053, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r1054, %r1053, %r1050;
ld.shared.f32 %f258, [%r1054];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(642+i*-32), %f258, %f(257+i*2), %f(642+i*-32);
ld.shared.f32 %f(259+i*1), [%r1052+(1024+i*-992)];
for.size.3 i in range(0, 30, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(641+i*-1+j*-32), %f258, %f(260+i*2+j*1), %f(641+i*-1+j*-32);
ld.shared.f32 %f(261+i*2+j*1), [%r1052+(1056+i*32+j*-992)];
fma.rn.f32 %f611, %f258, %f320, %f611;
ld.shared.f32 %f321, [%r1052+2016];
fma.rn.f32 %f579, %f258, %f321, %f579;
add.s32 %r1059, %r1059, 1;
setp.ne.s32 %p1, %r1059, 8;
@%p1 bra $L__BB0_2;
add.s32 %r1058, %r1058, 1;
setp.lt.u32 %p2, %r1058, 98;
@%p2 bra $L__BB0_1;
ld.param.u64 %rd17, [default_function_kernel0_param_2];
ld.param.u64 %rd16, [default_function_kernel0_param_3];
shl.b32 %r1056, %r24, 6;
cvta.to.global.u64 %rd11, %rd16;
mul.wide.s32 %rd12, %r1056, 4;
add.s64 %rd13, %rd11, %rd12;
ld.global.nc.f32 %f322, [%rd13];
add.f32 %f323, %f642, %f322;
mov.f32 %f324, 0f00000000;
max.f32 %f325, %f323, %f324;
cvta.to.global.u64 %rd14, %rd17;
add.s64 %rd15, %rd14, %rd12;
for.size.5 i in range(0, 31, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd15+(0+i*4+j*128)], %f(325+i*6+j*3);
ld.global.nc.f32 %f(326+i*6+j*3), [%rd13+(128+i*4+j*-124)];
add.f32 %f(327+i*6+j*3), %f(610+i*-1+j*31), %f(326+i*6+j*3);
max.f32 %f(328+i*6+j*3), %f(327+i*6+j*3), %f324;
st.global.f32 [%rd15+124], %f511;
ld.global.nc.f32 %f512, [%rd13+252];
add.f32 %f513, %f579, %f512;
max.f32 %f514, %f513, %f324;
st.global.f32 [%rd15+252], %f514;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[512];
for (int j_inner_init = 0; j_inner_init < 32; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
T_dense[index0 + 32] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
((float4*)(placeholder_d_shared + 0))[0] = ((float4*)(placeholder + k_outer_outer * 8))[0];
((float4*)(placeholder_d_shared + 4))[0] = ((float4*)(placeholder + k_outer_outer * 8 + 4))[0];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 64; ++ax0_ax1_fused_outer_outer) {
ulonglong4*(placeholder_shared + ax0_ax1_fused_outer_outer * 8)[0] = ulonglong4*(placeholder1 + (int)blockIdx.x * 50176 + ax0_ax1_fused_outer_outer * 784 + k_outer_outer * 8)[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 8; ++k_outer_inner) {
for (int j_inner = 0; j_inner < 32; ++j_inner) {
int index1 = j_inner;
int index2 = k_outer_inner;
int index3 = j_inner * 8 + k_outer_inner;
T_dense[index1] = T_dense[index1] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[index1 + 32] = T_dense[index1 + 32] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 256]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 32; ++ax1_inner) {
int index4 = (int)blockIdx.x * 64 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 32] = max(T_dense[index5 + 32] + placeholder2[index4 + 32], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpizw4ngdm | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<292>;
.reg .b32 %r<18>;
.reg .b64 %rd<23>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[4];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[256];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd1, %rd11;
mov.u32 %r1, %tid.x;
shl.b32 %r4, %r1, 2;
mov.u32 %r10, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r10, %r4;
mov.u32 %r11, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r11, %r4;
shl.b32 %r12, %r1, 4;
add.s32 %r5, %r11, %r12;
mul.lo.s32 %r13, %r1, 784;
cvta.to.global.u64 %rd13, %rd9;
mul.wide.s32 %rd14, %r13, 4;
add.s64 %rd22, %rd13, %rd14;
mov.u32 %r6, %ctaid.x;
mad.lo.s32 %r14, %r6, 50176, %r13;
mul.wide.s32 %rd15, %r14, 4;
add.s64 %rd16, %rd12, %rd15;
add.s64 %rd21, %rd16, 100352;
mov.u32 %r17, 0;
mov.f32 %f260, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(261+i*1), %f260;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f97, [%rd22];
st.shared.f32 [%r2], %f97;
$L__BB0_3:
for.size.2 i in range(0, 32, 1):
ld.global.nc.f32 %f(98+i*1), [%rd21+(-100352+i*6272)];
st.shared.f32 [%r3+(0+i*8)], %f(98+i*1);
bar.sync 0;
ld.shared.f32 %f130, [%r5];
ld.shared.f32 %f131, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(291+i*-1), %f131, %f(130+i*2), %f(291+i*-1);
ld.shared.f32 %f(132+i*1), [%r5+(32+i*32)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(289+i*-1), %f131, %f(133+i*1), %f(289+i*-1);
ld.shared.f32 %f(134+i*1), [%r5+(96+i*32)];
for.size.3 i in range(0, 3, 1):
for.size.2 j in range(0, 8, 1):
fma.rn.f32 %f(284+i*-8+j*-1), %f131, %f(138+i*8+j*1), %f(284+i*-8+j*-1);
ld.shared.f32 %f(139+i*8+j*1), [%r5+(4+i*4+j*32)];
fma.rn.f32 %f260, %f131, %f162, %f260;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(22+i*-1), %rd(22+i*-1), 4;
add.s32 %r17, %r17, 1;
setp.ne.s32 %p2, %r17, 784;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd17, %rd8;
shl.b32 %r15, %r6, 6;
add.s32 %r16, %r15, %r4;
mul.wide.s32 %rd18, %r16, 4;
add.s64 %rd19, %rd17, %rd18;
ld.global.nc.f32 %f163, [%rd19];
add.f32 %f164, %f291, %f163;
mov.f32 %f165, 0f00000000;
max.f32 %f166, %f164, %f165;
add.s64 %rd20, %rd1, %rd18;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd20+(0+i*32)], %f(166+i*3);
ld.global.nc.f32 %f(167+i*3), [%rd19+(32+i*32)];
add.f32 %f(168+i*3), %f(290+i*-1), %f(167+i*3);
max.f32 %f(169+i*3), %f(168+i*3), %f165;
for.size.10 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd20+(224+i*4+j*-220)], %f(187+i*24+j*3);
ld.global.nc.f32 %f(188+i*24+j*3), [%rd19+(4+i*4+j*32)];
add.f32 %f(189+i*24+j*3), %f(283+i*-8+j*-1), %f(188+i*24+j*3);
max.f32 %f(190+i*24+j*3), %f(189+i*24+j*3), %f165;
for.size.4 j in range(0, 6, 1):
st.global.f32 [%rd20+(36+i*4+j*32)], %f(193+i*24+j*3);
ld.global.nc.f32 %f(194+i*24+j*3), [%rd19+(68+i*4+j*32)];
add.f32 %f(195+i*24+j*3), %f(281+i*-8+j*-1), %f(194+i*24+j*3);
max.f32 %f(196+i*24+j*3), %f(195+i*24+j*3), %f165;
st.global.f32 [%rd20+236], %f259;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[64];
T_dense[0] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 784 + k_outer_outer;
placeholder_shared[index0] = placeholder1[index2];
placeholder_shared[index0 + 2] = placeholder1[index2 + 1568];
placeholder_shared[index0 + 4] = placeholder1[index2 + 3136];
placeholder_shared[index0 + 6] = placeholder1[index2 + 4704];
placeholder_shared[index0 + 8] = placeholder1[index2 + 6272];
placeholder_shared[index0 + 10] = placeholder1[index2 + 7840];
placeholder_shared[index0 + 12] = placeholder1[index2 + 9408];
placeholder_shared[index0 + 14] = placeholder1[index2 + 10976];
placeholder_shared[index0 + 16] = placeholder1[index2 + 12544];
placeholder_shared[index0 + 18] = placeholder1[index2 + 14112];
placeholder_shared[index0 + 20] = placeholder1[index2 + 15680];
placeholder_shared[index0 + 22] = placeholder1[index2 + 17248];
placeholder_shared[index0 + 24] = placeholder1[index2 + 18816];
placeholder_shared[index0 + 26] = placeholder1[index2 + 20384];
placeholder_shared[index0 + 28] = placeholder1[index2 + 21952];
placeholder_shared[index0 + 30] = placeholder1[index2 + 23520];
placeholder_shared[index0 + 32] = placeholder1[index2 + 25088];
placeholder_shared[index0 + 34] = placeholder1[index2 + 26656];
placeholder_shared[index0 + 36] = placeholder1[index2 + 28224];
placeholder_shared[index0 + 38] = placeholder1[index2 + 29792];
placeholder_shared[index0 + 40] = placeholder1[index2 + 31360];
placeholder_shared[index0 + 42] = placeholder1[index2 + 32928];
placeholder_shared[index0 + 44] = placeholder1[index2 + 34496];
placeholder_shared[index0 + 46] = placeholder1[index2 + 36064];
placeholder_shared[index0 + 48] = placeholder1[index2 + 37632];
placeholder_shared[index0 + 50] = placeholder1[index2 + 39200];
placeholder_shared[index0 + 52] = placeholder1[index2 + 40768];
placeholder_shared[index0 + 54] = placeholder1[index2 + 42336];
placeholder_shared[index0 + 56] = placeholder1[index2 + 43904];
placeholder_shared[index0 + 58] = placeholder1[index2 + 45472];
placeholder_shared[index0 + 60] = placeholder1[index2 + 47040];
placeholder_shared[index0 + 62] = placeholder1[index2 + 48608];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 4];
int index3 = (int)threadIdx.x * 4 + 8;
T_dense[4] = T_dense[4] + placeholder_d_shared[0] * placeholder_shared[index3];
int index4 = (int)threadIdx.x * 4 + 16;
T_dense[8] = T_dense[8] + placeholder_d_shared[0] * placeholder_shared[index4];
int index5 = (int)threadIdx.x * 4 + 24;
T_dense[12] = T_dense[12] + placeholder_d_shared[0] * placeholder_shared[index5];
int index6 = (int)threadIdx.x * 4 + 32;
T_dense[16] = T_dense[16] + placeholder_d_shared[0] * placeholder_shared[index6];
int index7 = (int)threadIdx.x * 4 + 40;
T_dense[20] = T_dense[20] + placeholder_d_shared[0] * placeholder_shared[index7];
int index8 = (int)threadIdx.x * 4 + 48;
T_dense[24] = T_dense[24] + placeholder_d_shared[0] * placeholder_shared[index8];
int index9 = (int)threadIdx.x * 4 + 56;
T_dense[28] = T_dense[28] + placeholder_d_shared[0] * placeholder_shared[index9];
int index10 = (int)threadIdx.x * 4 + 1;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index10];
int index11 = (int)threadIdx.x * 4 + 9;
T_dense[5] = T_dense[5] + placeholder_d_shared[0] * placeholder_shared[index11];
int index12 = (int)threadIdx.x * 4 + 17;
T_dense[9] = T_dense[9] + placeholder_d_shared[0] * placeholder_shared[index12];
int index13 = (int)threadIdx.x * 4 + 25;
T_dense[13] = T_dense[13] + placeholder_d_shared[0] * placeholder_shared[index13];
int index14 = (int)threadIdx.x * 4 + 33;
T_dense[17] = T_dense[17] + placeholder_d_shared[0] * placeholder_shared[index14];
int index15 = (int)threadIdx.x * 4 + 41;
T_dense[21] = T_dense[21] + placeholder_d_shared[0] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 4 + 49;
T_dense[25] = T_dense[25] + placeholder_d_shared[0] * placeholder_shared[index16];
int index17 = (int)threadIdx.x * 4 + 57;
T_dense[29] = T_dense[29] + placeholder_d_shared[0] * placeholder_shared[index17];
int index18 = (int)threadIdx.x * 4 + 2;
T_dense[2] = T_dense[2] + placeholder_d_shared[0] * placeholder_shared[index18];
int index19 = (int)threadIdx.x * 4 + 10;
T_dense[6] = T_dense[6] + placeholder_d_shared[0] * placeholder_shared[index19];
int index20 = (int)threadIdx.x * 4 + 18;
T_dense[10] = T_dense[10] + placeholder_d_shared[0] * placeholder_shared[index20];
int index21 = (int)threadIdx.x * 4 + 26;
T_dense[14] = T_dense[14] + placeholder_d_shared[0] * placeholder_shared[index21];
int index22 = (int)threadIdx.x * 4 + 34;
T_dense[18] = T_dense[18] + placeholder_d_shared[0] * placeholder_shared[index22];
int index23 = (int)threadIdx.x * 4 + 42;
T_dense[22] = T_dense[22] + placeholder_d_shared[0] * placeholder_shared[index23];
int index24 = (int)threadIdx.x * 4 + 50;
T_dense[26] = T_dense[26] + placeholder_d_shared[0] * placeholder_shared[index24];
int index25 = (int)threadIdx.x * 4 + 58;
T_dense[30] = T_dense[30] + placeholder_d_shared[0] * placeholder_shared[index25];
int index26 = (int)threadIdx.x * 4 + 3;
T_dense[3] = T_dense[3] + placeholder_d_shared[0] * placeholder_shared[index26];
int index27 = (int)threadIdx.x * 4 + 11;
T_dense[7] = T_dense[7] + placeholder_d_shared[0] * placeholder_shared[index27];
int index28 = (int)threadIdx.x * 4 + 19;
T_dense[11] = T_dense[11] + placeholder_d_shared[0] * placeholder_shared[index28];
int index29 = (int)threadIdx.x * 4 + 27;
T_dense[15] = T_dense[15] + placeholder_d_shared[0] * placeholder_shared[index29];
int index30 = (int)threadIdx.x * 4 + 35;
T_dense[19] = T_dense[19] + placeholder_d_shared[0] * placeholder_shared[index30];
int index31 = (int)threadIdx.x * 4 + 43;
T_dense[23] = T_dense[23] + placeholder_d_shared[0] * placeholder_shared[index31];
int index32 = (int)threadIdx.x * 4 + 51;
T_dense[27] = T_dense[27] + placeholder_d_shared[0] * placeholder_shared[index32];
int index33 = (int)threadIdx.x * 4 + 59;
T_dense[31] = T_dense[31] + placeholder_d_shared[0] * placeholder_shared[index33];
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index34 = (int)blockIdx.x * 64 + (int)threadIdx.x * 4 + ax1_inner;
int index35 = ax1_inner;
T_relu[index34] = max(T_dense[index35] + placeholder2[index34], 0.000000e+00f);
T_relu[index34 + 8] = max(T_dense[index35 + 4] + placeholder2[index34 + 8], 0.000000e+00f);
T_relu[index34 + 16] = max(T_dense[index35 + 8] + placeholder2[index34 + 16], 0.000000e+00f);
T_relu[index34 + 24] = max(T_dense[index35 + 12] + placeholder2[index34 + 24], 0.000000e+00f);
T_relu[index34 + 32] = max(T_dense[index35 + 16] + placeholder2[index34 + 32], 0.000000e+00f);
T_relu[index34 + 40] = max(T_dense[index35 + 20] + placeholder2[index34 + 40], 0.000000e+00f);
T_relu[index34 + 48] = max(T_dense[index35 + 24] + placeholder2[index34 + 48], 0.000000e+00f);
T_relu[index34 + 56] = max(T_dense[index35 + 28] + placeholder2[index34 + 56], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp7wub3ptp | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<229>;
.reg .b32 %r<108>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[100352];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r19, %r18;
mul.lo.s32 %r3, %r1, 3136;
add.s32 %r4, %r1, 56;
mov.u32 %r20, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r20, %r18;
mov.u32 %r103, 0;
mov.f32 %f197, 0f00000000;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(198+i*1), %f197;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r7, %r103, 196;
add.s32 %r21, %r7, %r1;
mul.wide.s32 %rd10, %r21, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.2 i in range(0, 24, 1):
ld.global.nc.f32 %f(65+i*1), [%rd5+(0+i*32)];
st.shared.f32 [%r2+(0+i*32)], %f(65+i*1);
setp.gt.s32 %p1, %r1, 3;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f89, [%rd5+768];
st.shared.f32 [%r2+768], %f89;
$L__BB0_3:
mov.u32 %r106, 128;
mov.u32 %r104, %r1;
mov.u32 %r105, %r4;
$L__BB0_4:
mul.hi.s32 %r23, %r104, 1402438301;
shr.u32 %r24, %r23, 31;
shr.s32 %r25, %r23, 6;
add.s32 %r26, %r25, %r24;
mul.lo.s32 %r27, %r26, 196;
sub.s32 %r28, %r104, %r27;
add.s32 %r29, %r28, %r7;
mad.lo.s32 %r30, %r26, 784, %r29;
mul.wide.s32 %rd11, %r30, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f90, [%rd12];
add.s32 %r31, %r5, %r106;
for.size.13 i in range(0, 4, 1):
st.shared.f32 [%r31+(-128+i*32)], %f(90+i*1);
add.s32 %r(32+i*9), %r104, (8+i*8);
mul.hi.s32 %r(33+i*9), %r(32+i*9), 1402438301;
shr.u32 %r(34+i*9), %r(33+i*9), 31;
shr.s32 %r(35+i*9), %r(33+i*9), 6;
add.s32 %r(36+i*9), %r(35+i*9), %r(34+i*9);
mul.lo.s32 %r(37+i*9), %r(36+i*9), 196;
sub.s32 %r(38+i*9), %r(32+i*9), %r(37+i*9);
add.s32 %r(39+i*9), %r(38+i*9), %r7;
mad.lo.s32 %r(40+i*9), %r(36+i*9), 784, %r(39+i*9);
mul.wide.s32 %rd(13+i*2), %r(40+i*9), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(91+i*1), [%rd(14+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.f32 [%r31+(0+i*32)], %f(94+i*1);
add.s32 %r(68+i*9), %r105, (-16+i*8);
mul.hi.s32 %r(69+i*9), %r(68+i*9), 1402438301;
shr.u32 %r(70+i*9), %r(69+i*9), 31;
shr.s32 %r(71+i*9), %r(69+i*9), 6;
add.s32 %r(72+i*9), %r(71+i*9), %r(70+i*9);
mul.lo.s32 %r(73+i*9), %r(72+i*9), 196;
sub.s32 %r(74+i*9), %r(68+i*9), %r(73+i*9);
add.s32 %r(75+i*9), %r(74+i*9), %r7;
mad.lo.s32 %r(76+i*9), %r(72+i*9), 784, %r(75+i*9);
mul.wide.s32 %rd(21+i*2), %r(76+i*9), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(95+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r31+64], %f96;
mul.hi.s32 %r86, %r105, 1402438301;
shr.u32 %r87, %r86, 31;
shr.s32 %r88, %r86, 6;
add.s32 %r89, %r88, %r87;
mul.lo.s32 %r90, %r89, 196;
sub.s32 %r91, %r105, %r90;
add.s32 %r92, %r91, %r7;
mad.lo.s32 %r93, %r89, 784, %r92;
mul.wide.s32 %rd25, %r93, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f97, [%rd26];
st.shared.f32 [%r31+96], %f97;
add.s32 %r106, %r106, 256;
setp.ne.s32 %p2, %r106, 100480;
for.size.1 i in range(0, 2, 1):
add.s32 %r(12+i*92), %r105, (64+i*-56);
mov.u32 %r105, %r12;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r107, 0;
$L__BB0_6:
add.s32 %r95, %r107, %r3;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(96+i*3), %r(95+i*12), 2;
add.s32 %r(98+i*3), %r(20+i*-1), %r(96+i*3);
ld.shared.f32 %f(98+i*1), [%r(98+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(100+i*2), %f99, %f(98+i*3), %f(212+i*-1);
ld.shared.f32 %f(101+i*2), [%r98+(784+i*784)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(104+i*2), %f99, %f(103+i*2), %f(210+i*-1);
ld.shared.f32 %f(105+i*2), [%r98+(2352+i*784)];
fma.rn.f32 %f130, %f99, %f129, %f197;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(131+i*1), [%r(98+i*3)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(212+i*-1), %f132, %f(131+i*2), %f(100+i*2);
ld.shared.f32 %f(133+i*1), [%r98+(788+i*784)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(210+i*-1), %f132, %f(134+i*1), %f(104+i*2);
ld.shared.f32 %f(135+i*1), [%r98+(2356+i*784)];
fma.rn.f32 %f197, %f132, %f147, %f130;
add.s32 %r107, %r107, 2;
setp.ne.s32 %p3, %r107, 196;
@%p3 bra $L__BB0_6;
add.s32 %r103, %r103, 1;
setp.lt.u32 %p4, %r103, 4;
@%p4 bra $L__BB0_1;
shl.b32 %r102, %r1, 4;
mul.wide.s32 %rd27, %r102, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f148, [%rd28];
add.f32 %f149, %f212, %f148;
mov.f32 %f150, 0f00000000;
max.f32 %f151, %f149, %f150;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 15, 1):
st.global.f32 [%rd29+(0+i*4)], %f(151+i*3);
ld.global.nc.f32 %f(152+i*3), [%rd28+(4+i*4)];
add.f32 %f(153+i*3), %f(211+i*-1), %f(152+i*3);
max.f32 %f(154+i*3), %f(153+i*3), %f150;
st.global.f32 [%rd29+60], %f196;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[25088];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 196 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 40] = placeholder[index1 + 40];
placeholder_d_shared[index0 + 48] = placeholder[index1 + 48];
placeholder_d_shared[index0 + 56] = placeholder[index1 + 56];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 72] = placeholder[index1 + 72];
placeholder_d_shared[index0 + 80] = placeholder[index1 + 80];
placeholder_d_shared[index0 + 88] = placeholder[index1 + 88];
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
placeholder_d_shared[index0 + 104] = placeholder[index1 + 104];
placeholder_d_shared[index0 + 112] = placeholder[index1 + 112];
placeholder_d_shared[index0 + 120] = placeholder[index1 + 120];
placeholder_d_shared[index0 + 128] = placeholder[index1 + 128];
placeholder_d_shared[index0 + 136] = placeholder[index1 + 136];
placeholder_d_shared[index0 + 144] = placeholder[index1 + 144];
placeholder_d_shared[index0 + 152] = placeholder[index1 + 152];
placeholder_d_shared[index0 + 160] = placeholder[index1 + 160];
placeholder_d_shared[index0 + 168] = placeholder[index1 + 168];
placeholder_d_shared[index0 + 176] = placeholder[index1 + 176];
placeholder_d_shared[index0 + 184] = placeholder[index1 + 184];
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 192] = placeholder[index1 + 192];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 3136; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x) % 196;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 3136 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 196]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 392]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 588]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 980]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1176]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1372]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1568]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1764]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1960]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2156]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2352]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2548]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2744]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2940]);
}
}
for (int ax1_inner = 0; ax1_inner < 16; ++ax1_inner) {
int index6 = (int)threadIdx.x * 16 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmptrn7w2g4 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<68>;
.reg .b32 %r<20>;
.reg .b64 %rd<34>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E7T_dense[8];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E8red_buf0[8];
ld.param.u64 %rd16, [default_function_kernel0_param_0];
ld.param.u64 %rd17, [default_function_kernel0_param_1];
ld.param.u64 %rd18, [default_function_kernel0_param_2];
ld.param.u64 %rd19, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd17;
cvta.to.global.u64 %rd2, %rd18;
cvta.to.global.u64 %rd3, %rd19;
mov.u32 %r1, %tid.x;
shl.b32 %r9, %r1, 2;
mov.u32 %r10, _ZZ24default_function_kernel0E8red_buf0;
add.s32 %r2, %r10, %r9;
cvta.to.global.u64 %rd20, %rd16;
mul.wide.s32 %rd21, %r1, 4;
add.s64 %rd22, %rd20, %rd21;
add.s64 %rd33, %rd22, 32;
mov.u32 %r3, %ctaid.x;
mad.lo.s32 %r11, %r3, 1568, %r1;
cvt.s64.s32 %rd5, %r11;
mul.wide.s32 %rd23, %r11, 4;
add.s64 %rd24, %rd1, %rd23;
add.s64 %rd30, %rd24, 32;
mov.f32 %f66, 0f00000000;
mov.u32 %r18, 0;
mov.u64 %rd31, %rd33;
$L__BB0_1:
for.size.1 i in range(0, 2, 1):
ld.global.nc.f32 %f(6+i*1), [%rd(30+i*1)+-32];
for.size.3 i in range(0, 2, 1):
fma.rn.f32 %f(8+i*3), %f(7+i*3), %f(6+i*3), %f(66+i*-58);
ld.global.nc.f32 %f(9+i*3), [%rd30+(-24+i*8)];
ld.global.nc.f32 %f(10+i*3), [%rd31+(-24+i*8)];
for.size.3 i in range(0, 5, 1):
fma.rn.f32 %f(14+i*3), %f(13+i*3), %f(12+i*3), %f(11+i*3);
ld.global.nc.f32 %f(15+i*3), [%rd30+(-8+i*8)];
ld.global.nc.f32 %f(16+i*3), [%rd31+(-8+i*8)];
fma.rn.f32 %f66, %f28, %f27, %f26;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(31+i*-1), %rd(31+i*-1), 64;
add.s32 %r18, %r18, 8;
setp.ne.s32 %p1, %r18, 392;
@%p1 bra $L__BB0_1;
setp.gt.s32 %p2, %r1, 0;
bar.sync 0;
st.volatile.shared.f32 [%r2], %f66;
bar.sync 0;
@%p2 bra $L__BB0_4;
for.size.1 i in range(0, 2, 1):
ld.volatile.shared.f32 %f(29+i*1), [%r2+(0+i*4)];
add.f32 %f31, %f29, %f30;
st.volatile.shared.f32 [%r2], %f31;
$L__BB0_4:
bar.sync 0;
mov.u32 %r19, 0;
ld.volatile.shared.f32 %f33, [_ZZ24default_function_kernel0E8red_buf0];
st.shared.f32 [_ZZ24default_function_kernel0E7T_dense], %f33;
shl.b64 %rd25, %rd5, 2;
add.s64 %rd26, %rd1, %rd25;
add.s64 %rd32, %rd26, 3192;
mov.f32 %f67, 0f00000000;
$L__BB0_5:
for.size.3 i in range(0, 2, 1):
ld.global.nc.f32 %f(34+i*3), [%rd32+(-56+i*8)];
ld.global.nc.f32 %f(35+i*3), [%rd33+(-32+i*8)];
fma.rn.f32 %f(36+i*3), %f(35+i*3), %f(34+i*3), %f(67+i*-31);
for.size.3 i in range(0, 5, 1):
ld.global.nc.f32 %f(40+i*3), [%rd32+(-40+i*8)];
ld.global.nc.f32 %f(41+i*3), [%rd33+(-16+i*8)];
fma.rn.f32 %f(42+i*3), %f(41+i*3), %f(40+i*3), %f(39+i*3);
ld.global.nc.f32 %f55, [%rd32];
ld.global.nc.f32 %f56, [%rd33+24];
fma.rn.f32 %f67, %f56, %f55, %f54;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(33+i*-1), %rd(33+i*-1), 64;
add.s32 %r19, %r19, 8;
setp.ne.s32 %p3, %r19, 392;
@%p3 bra $L__BB0_5;
bar.sync 0;
st.volatile.shared.f32 [%r2], %f67;
bar.sync 0;
@%p2 bra $L__BB0_8;
for.size.1 i in range(0, 2, 1):
ld.volatile.shared.f32 %f(57+i*1), [%r2+(0+i*4)];
add.f32 %f59, %f57, %f58;
st.volatile.shared.f32 [%r2], %f59;
$L__BB0_8:
bar.sync 0;
ld.volatile.shared.f32 %f60, [_ZZ24default_function_kernel0E8red_buf0];
mov.u32 %r13, _ZZ24default_function_kernel0E7T_dense;
st.shared.f32 [_ZZ24default_function_kernel0E7T_dense+4], %f60;
bar.sync 0;
shl.b32 %r14, %r3, 1;
add.s32 %r15, %r14, %r1;
mul.wide.s32 %rd27, %r15, 4;
add.s64 %rd28, %rd3, %rd27;
ld.global.nc.f32 %f61, [%rd28];
add.s32 %r17, %r13, %r9;
ld.shared.f32 %f62, [%r17];
add.f32 %f63, %f62, %f61;
mov.f32 %f64, 0f00000000;
max.f32 %f65, %f63, %f64;
add.s64 %rd29, %rd2, %rd27;
st.global.f32 [%rd29], %f65;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
__shared__ float T_dense[2];
float normal_reduce_temp0[1];
__shared__ float red_buf0[2];
for (int j = 0; j < 2; ++j) {
normal_reduce_temp0[0] = 0.000000e+00f;
for (int k_outer = 0; k_outer < 392; ++k_outer) {
int index0 = k_outer * 2 + (int)threadIdx.x;
int index1 = (int)blockIdx.x * 1568 + j * 784 + k_outer * 2 + (int)threadIdx.x;
normal_reduce_temp0[0] = normal_reduce_temp0[0] + placeholder[index0] * placeholder1[index1];
}
__syncthreads();
int index2 = (int)threadIdx.x;
((volatile float*)red_buf0)[index2] = normal_reduce_temp0[0];
__syncthreads();
if ((int)threadIdx.x < 1) {
((volatile float*)red_buf0)[index2] = ((volatile float*)red_buf0)[index2] + ((volatile float*)red_buf0)[index2 + 1];
}
__syncthreads();
int index3 = j;
T_dense[index3] = ((volatile float*)red_buf0)[0];
}
__syncthreads();
int index2 = (int)threadIdx.x;
int index4 = (int)blockIdx.x * 2 + (int)threadIdx.x;
T_relu[index4] = max(T_dense[index2] + placeholder2[index4], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpt0gfnz24 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<72>;
.reg .b32 %r<103>;
.reg .b64 %rd<18>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[1568];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[100352];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd5, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
mov.u32 %r20, %tid.x;
shl.b32 %r1, %r20, 1;
mov.u32 %r21, %ctaid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(2+i*1), %r(21+i*-1), (50176+i*-49392);
shl.b32 %r22, %r20, 3;
add.s32 %r4, %r22, 256;
shl.b32 %r23, %r20, 5;
mov.u32 %r24, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r25, %r24, %r23;
add.s32 %r5, %r25, 1040;
mov.u32 %r98, 0;
mov.f32 %f68, 0f00000000;
mov.f32 %f69, %f68;
$L__BB0_1:
setp.gt.s32 %p1, %r20, 3;
bar.sync 0;
mul.lo.s32 %r7, %r98, 392;
add.s32 %r28, %r7, %r1;
mul.wide.s32 %rd8, %r28, 4;
add.s64 %rd3, %rd2, %rd8;
ld.global.nc.f32 %f9, [%rd3];
mov.u32 %r30, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r31, %r30, %r22;
for.size.3 i in range(0, 5, 1):
for.size.2 j in range(0, 2, 1):
st.shared.f32 [%r31+(0+i*256+j*4)], %f(9+i*2+j*1);
ld.global.nc.f32 %f(10+i*2+j*1), [%rd3+(4+i*256+j*252)];
st.shared.f32 [%r31+1280], %f19;
ld.global.nc.f32 %f20, [%rd3+1284];
st.shared.f32 [%r31+1284], %f20;
@%p1 bra $L__BB0_3;
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(21+i*1), [%rd3+(1536+i*4)];
st.shared.f32 [%r31+(1536+i*4)], %f(21+i*1);
$L__BB0_3:
add.s32 %r9, %r7, %r2;
mov.u32 %r101, 1024;
mov.u32 %r99, %r5;
mov.u32 %r100, %r4;
$L__BB0_4:
add.s32 %r37, %r100, -256;
for.size.14 i in range(0, 2, 1):
mul.hi.s32 %r(38+i*24), %r(37+i*63), 1402438301;
shr.u32 %r(39+i*24), %r(38+i*24), 31;
shr.s32 %r(40+i*24), %r(38+i*24), 7;
add.s32 %r(41+i*24), %r(40+i*24), %r(39+i*24);
mul.lo.s32 %r(42+i*24), %r(41+i*24), 392;
sub.s32 %r(43+i*24), %r(37+i*63), %r(42+i*24);
add.s32 %r(44+i*24), %r9, %r(43+i*24);
mad.lo.s32 %r(45+i*24), %r(41+i*24), 784, %r(44+i*24);
mul.wide.s32 %rd(9+i*2), %r(45+i*24), 4;
add.s64 %rd(10+i*2), %rd1, %rd(9+i*2);
ld.global.nc.v4.u32 {%r(46+i*24), %r(47+i*24), %r(48+i*24), %r(49+i*24)}, [%rd(10+i*2)+0];
ld.global.nc.v4.u32 {%r(54+i*24), %r(55+i*24), %r(56+i*24), %r(57+i*24)}, [%rd(10+i*2)+16];
st.shared.v4.u32 [%r99+(-1040+i*1024)], {%r(46+i*24), %r(47+i*24), %r(48+i*24), %r(49+i*24)};
st.shared.v4.u32 [%r99+(-1024+i*1024)], {%r(54+i*24), %r(55+i*24), %r(56+i*24), %r(57+i*24)};
for.size.1 i in range(0, 2, 1):
add.s32 %r(100+i*-1), %r(100+i*-1), (512+i*1536);
add.s32 %r101, %r101, 2048;
setp.ne.s32 %p2, %r101, 101376;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r102, 0;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(87+i*1), %r102, (1+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(90+i*1), %r(30+i*57), %r(88+i*-85);
shl.b32 %r92, %r91, 2;
add.s32 %r94, %r24, %r92;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(23+i*1), [%r(94+i*-4)+0];
for.size.10 i in range(0, 2, 1):
fma.rn.f32 %f(25+i*10), %f(24+i*10), %f(23+i*10), %f(69+i*-41);
ld.shared.f32 %f(26+i*10), [%r94+(4+i*8)];
ld.shared.f32 %f(27+i*10), [%r90+(4+i*8)];
fma.rn.f32 %f(28+i*10), %f(27+i*10), %f(26+i*10), %f(25+i*10);
ld.shared.f32 %f(29+i*10), [%r94+(1568+i*8)];
fma.rn.f32 %f(30+i*10), %f(24+i*10), %f(29+i*10), %f(68+i*-36);
ld.shared.f32 %f(31+i*10), [%r94+(1572+i*8)];
fma.rn.f32 %f(32+i*10), %f(27+i*10), %f(31+i*10), %f(30+i*10);
ld.shared.f32 %f(33+i*10), [%r94+(8+i*8)];
ld.shared.f32 %f(34+i*10), [%r90+(8+i*8)];
fma.rn.f32 %f45, %f44, %f43, %f38;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(46+i*1), [%r(94+i*-4)+20];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(48+i*2), %f(47+i*-3), %f(46+i*3), %f(45+i*-3);
ld.shared.f32 %f(49+i*2), [%r94+(1584+i*4)];
for.size.3 i in range(0, 2, 1):
fma.rn.f32 %f(52+i*3), %f(47+i*7), %f(51+i*2), %f(50+i*-2);
ld.shared.f32 %f(53+i*3), [%r94+(24+i*4)];
ld.shared.f32 %f(54+i*3), [%r90+(24+i*4)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(69+i*-10), %f(57+i*-3), %f(56+i*2), %f(55+i*-3);
ld.shared.f32 %f(58+i*2), [%r94+(1592+i*4)];
fma.rn.f32 %f68, %f57, %f60, %f59;
add.s32 %r102, %r102, 4;
setp.ne.s32 %p3, %r102, 196;
@%p3 bra $L__BB0_6;
add.s32 %r98, %r98, 1;
setp.lt.u32 %p4, %r98, 2;
@%p4 bra $L__BB0_1;
shl.b32 %r96, %r21, 6;
add.s32 %r97, %r96, %r1;
cvta.to.global.u64 %rd13, %rd5;
mul.wide.s32 %rd14, %r97, 4;
add.s64 %rd15, %rd13, %rd14;
ld.global.nc.f32 %f61, [%rd15];
add.f32 %f62, %f69, %f61;
mov.f32 %f63, 0f00000000;
max.f32 %f64, %f62, %f63;
cvta.to.global.u64 %rd16, %rd4;
add.s64 %rd17, %rd16, %rd14;
st.global.f32 [%rd17], %f64;
ld.global.nc.f32 %f65, [%rd15+4];
add.f32 %f66, %f68, %f65;
max.f32 %f67, %f66, %f63;
st.global.f32 [%rd17+4], %f67;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[392];
__shared__ float placeholder_shared[25088];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 2; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x * 2;
int index1 = k_outer_outer * 392 + (int)threadIdx.x * 2;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 65] = placeholder[index1 + 65];
placeholder_d_shared[index0 + 128] = placeholder[index1 + 128];
placeholder_d_shared[index0 + 129] = placeholder[index1 + 129];
placeholder_d_shared[index0 + 192] = placeholder[index1 + 192];
placeholder_d_shared[index0 + 193] = placeholder[index1 + 193];
placeholder_d_shared[index0 + 256] = placeholder[index1 + 256];
placeholder_d_shared[index0 + 257] = placeholder[index1 + 257];
placeholder_d_shared[index0 + 320] = placeholder[index1 + 320];
placeholder_d_shared[index0 + 321] = placeholder[index1 + 321];
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 384] = placeholder[index1 + 384];
}
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 385] = placeholder[index1 + 385];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
ulonglong4*(placeholder_shared + ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 8)[0] = ulonglong4*(placeholder1 + (int)blockIdx.x * 50176 + (ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 8) / 392 * 784 + k_outer_outer * 392 + (ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 8) % 392)[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
int index2 = k_outer_inner * 2;
int index3 = (int)threadIdx.x * 784 + k_outer_inner * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 392]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 393]);
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index4 = (int)blockIdx.x * 64 + (int)threadIdx.x * 2 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpf9aka2kn | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<197>;
.reg .b32 %r<18>;
.reg .b64 %rd<23>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[8];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[1024];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd12, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd10;
cvta.to.global.u64 %rd1, %rd11;
cvta.to.global.u64 %rd2, %rd12;
mov.u32 %r1, %tid.x;
shl.b32 %r8, %r1, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r9, %r8;
shr.s32 %r10, %r1, 1;
and.b32 %r11, %r1, 1;
mov.u32 %r12, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r12, %r8;
shl.b32 %r13, %r1, 7;
add.s32 %r4, %r12, %r13;
cvta.to.global.u64 %rd14, %rd9;
mul.wide.s32 %rd15, %r1, 4;
add.s64 %rd22, %rd14, %rd15;
mul.lo.s32 %r14, %r10, 784;
or.b32 %r15, %r14, %r11;
mul.wide.s32 %rd16, %r15, 4;
add.s64 %rd17, %rd13, %rd16;
add.s64 %rd21, %rd17, 200704;
mov.u32 %r17, 0;
mov.f32 %f181, 0f00000000;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(182+i*1), %f181;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f49, [%rd22];
st.shared.f32 [%r2], %f49;
$L__BB0_3:
for.size.2 i in range(0, 32, 1):
ld.global.nc.f32 %f(50+i*1), [%rd21+(-200704+i*12544)];
st.shared.f32 [%r3+(0+i*32)], %f(50+i*1);
bar.sync 0;
for.size.33 i in range(0, 2, 1):
ld.shared.f32 %f(82+i*33), [%r4+(0+i*4)];
ld.shared.f32 %f(83+i*33), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(84+i*112), %f(83+i*33), %f(82+i*33), %f(196+i*-112);
ld.shared.f32 %f(85+i*32), [%r4+(8+i*4)];
fma.rn.f32 %f(86+i*109), %f(83+i*33), %f(85+i*32), %f(195+i*-109);
ld.shared.f32 %f(87+i*31), [%r4+(16+i*4)];
fma.rn.f32 %f(88+i*106), %f(83+i*33), %f(87+i*31), %f(194+i*-106);
ld.shared.f32 %f(89+i*30), [%r4+(24+i*4)];
fma.rn.f32 %f(90+i*103), %f(83+i*33), %f(89+i*30), %f(193+i*-103);
ld.shared.f32 %f(91+i*29), [%r4+(32+i*4)];
fma.rn.f32 %f(92+i*100), %f(83+i*33), %f(91+i*29), %f(192+i*-100);
ld.shared.f32 %f(93+i*28), [%r4+(40+i*4)];
fma.rn.f32 %f(94+i*97), %f(83+i*33), %f(93+i*28), %f(191+i*-97);
ld.shared.f32 %f(95+i*27), [%r4+(48+i*4)];
fma.rn.f32 %f(96+i*94), %f(83+i*33), %f(95+i*27), %f(190+i*-94);
ld.shared.f32 %f(97+i*26), [%r4+(56+i*4)];
fma.rn.f32 %f(98+i*91), %f(83+i*33), %f(97+i*26), %f(189+i*-91);
ld.shared.f32 %f(99+i*25), [%r4+(64+i*4)];
fma.rn.f32 %f(100+i*88), %f(83+i*33), %f(99+i*25), %f(188+i*-88);
ld.shared.f32 %f(101+i*24), [%r4+(72+i*4)];
fma.rn.f32 %f(102+i*85), %f(83+i*33), %f(101+i*24), %f(187+i*-85);
ld.shared.f32 %f(103+i*23), [%r4+(80+i*4)];
fma.rn.f32 %f(104+i*82), %f(83+i*33), %f(103+i*23), %f(186+i*-82);
ld.shared.f32 %f(105+i*22), [%r4+(88+i*4)];
fma.rn.f32 %f(106+i*79), %f(83+i*33), %f(105+i*22), %f(185+i*-79);
ld.shared.f32 %f(107+i*21), [%r4+(96+i*4)];
fma.rn.f32 %f(108+i*76), %f(83+i*33), %f(107+i*21), %f(184+i*-76);
ld.shared.f32 %f(109+i*20), [%r4+(104+i*4)];
fma.rn.f32 %f(110+i*73), %f(83+i*33), %f(109+i*20), %f(183+i*-73);
ld.shared.f32 %f(111+i*19), [%r4+(112+i*4)];
fma.rn.f32 %f(112+i*70), %f(83+i*33), %f(111+i*19), %f(182+i*-70);
ld.shared.f32 %f(113+i*18), [%r4+(120+i*4)];
fma.rn.f32 %f(114+i*67), %f(83+i*33), %f(113+i*18), %f(181+i*-67);
for.size.1 i in range(0, 2, 1):
add.s64 %rd(22+i*-1), %rd(22+i*-1), 8;
add.s32 %r17, %r17, 1;
setp.ne.s32 %p2, %r17, 392;
@%p2 bra $L__BB0_1;
shl.b32 %r16, %r1, 4;
mul.wide.s32 %rd18, %r16, 4;
add.s64 %rd19, %rd2, %rd18;
ld.global.nc.f32 %f132, [%rd19];
add.f32 %f133, %f196, %f132;
mov.f32 %f134, 0f00000000;
max.f32 %f135, %f133, %f134;
add.s64 %rd20, %rd1, %rd18;
for.size.4 i in range(0, 15, 1):
st.global.f32 [%rd20+(0+i*4)], %f(135+i*3);
ld.global.nc.f32 %f(136+i*3), [%rd19+(4+i*4)];
add.f32 %f(137+i*3), %f(195+i*-1), %f(136+i*3);
max.f32 %f(138+i*3), %f(137+i*3), %f134;
st.global.f32 [%rd20+60], %f180;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[2];
__shared__ float placeholder_shared[256];
for (int j_inner_init = 0; j_inner_init < 16; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 392; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 2) {
int index1 = (int)threadIdx.x;
int index2 = k_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 32; ++ax0_ax1_fused_outer_outer) {
int index3 = ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
int index4 = ax0_ax1_fused_outer_outer * 3136 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 2 + ((int)threadIdx.x & 1);
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int j_inner = 0; j_inner < 16; ++j_inner) {
int index5 = j_inner;
int index6 = k_outer_inner;
int index7 = (int)threadIdx.x * 32 + j_inner * 2 + k_outer_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
for (int ax1_inner = 0; ax1_inner < 16; ++ax1_inner) {
int index8 = (int)threadIdx.x * 16 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpjaqc7av6 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<243>;
.reg .b32 %r<127>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[112];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[7168];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd7, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd5;
cvta.to.global.u64 %rd3, %rd7;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r1, %tid.x;
shl.b32 %r26, %r1, 2;
mov.u32 %r27, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r27, %r26;
mov.u32 %r3, %ctaid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(4+i*1), %r(3+i*-2), (50176+i*-50148);
add.s32 %r6, %r1, 28;
mov.u32 %r28, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r7, %r28, %r26;
mov.u32 %r121, 0;
mov.f32 %f227, 0f00000000;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(228+i*1), %f227;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r30, %r121, 28;
add.s32 %r31, %r30, %r1;
mul.wide.s32 %rd9, %r31, 4;
add.s64 %rd10, %rd2, %rd9;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(97+i*1), [%rd10+(0+i*16)];
st.shared.f32 [%r2+(0+i*16)], %f(97+i*1);
ld.global.nc.f32 %f101, [%rd10+64];
mov.u32 %r124, 64;
for.size.2 i in range(0, 2, 1):
st.shared.f32 [%r2+(64+i*16)], %f(101+i*1);
ld.global.nc.f32 %f(102+i*1), [%rd10+(80+i*16)];
st.shared.f32 [%r2+96], %f103;
add.s32 %r9, %r30, %r4;
mov.u32 %r122, %r1;
mov.u32 %r123, %r6;
$L__BB0_2:
mul.hi.s32 %r32, %r122, -1840700269;
add.s32 %r33, %r32, %r122;
shr.u32 %r34, %r33, 31;
shr.s32 %r35, %r33, 4;
add.s32 %r36, %r35, %r34;
mul.lo.s32 %r37, %r36, 28;
sub.s32 %r38, %r122, %r37;
add.s32 %r39, %r9, %r38;
mad.lo.s32 %r40, %r36, 784, %r39;
mul.wide.s32 %rd11, %r40, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f104, [%rd12];
add.s32 %r41, %r7, %r124;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r41+(-64+i*16)], %f(104+i*1);
add.s32 %r(42+i*10), %r122, (4+i*4);
mul.hi.s32 %r(43+i*10), %r(42+i*10), -1840700269;
add.s32 %r(44+i*10), %r(43+i*10), %r(42+i*10);
shr.u32 %r(45+i*10), %r(44+i*10), 31;
shr.s32 %r(46+i*10), %r(44+i*10), 4;
add.s32 %r(47+i*10), %r(46+i*10), %r(45+i*10);
mul.lo.s32 %r(48+i*10), %r(47+i*10), 28;
sub.s32 %r(49+i*10), %r(42+i*10), %r(48+i*10);
add.s32 %r(50+i*10), %r9, %r(49+i*10);
mad.lo.s32 %r(51+i*10), %r(47+i*10), 784, %r(50+i*10);
mul.wide.s32 %rd(13+i*2), %r(51+i*10), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(105+i*1), [%rd(14+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r41+(0+i*16)], %f(108+i*1);
add.s32 %r(82+i*10), %r123, (-8+i*4);
mul.hi.s32 %r(83+i*10), %r(82+i*10), -1840700269;
add.s32 %r(84+i*10), %r(83+i*10), %r(82+i*10);
shr.u32 %r(85+i*10), %r(84+i*10), 31;
shr.s32 %r(86+i*10), %r(84+i*10), 4;
add.s32 %r(87+i*10), %r(86+i*10), %r(85+i*10);
mul.lo.s32 %r(88+i*10), %r(87+i*10), 28;
sub.s32 %r(89+i*10), %r(82+i*10), %r(88+i*10);
add.s32 %r(90+i*10), %r9, %r(89+i*10);
mad.lo.s32 %r(91+i*10), %r(87+i*10), 784, %r(90+i*10);
mul.wide.s32 %rd(21+i*2), %r(91+i*10), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(109+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r41+32], %f110;
mul.hi.s32 %r102, %r123, -1840700269;
add.s32 %r103, %r102, %r123;
shr.u32 %r104, %r103, 31;
shr.s32 %r105, %r103, 4;
add.s32 %r106, %r105, %r104;
mul.lo.s32 %r107, %r106, 28;
sub.s32 %r108, %r123, %r107;
add.s32 %r109, %r9, %r108;
mad.lo.s32 %r110, %r106, 784, %r109;
mul.wide.s32 %rd25, %r110, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f111, [%rd26];
st.shared.f32 [%r41+48], %f111;
add.s32 %r124, %r124, 128;
setp.ne.s32 %p1, %r124, 7232;
for.size.1 i in range(0, 2, 1):
add.s32 %r(14+i*108), %r123, (32+i*-28);
mov.u32 %r123, %r14;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r111, 0;
mov.u32 %r125, %r111;
$L__BB0_4:
mul.lo.s32 %r17, %r125, 7;
add.s32 %r18, %r17, %r5;
mov.u32 %r126, %r111;
bra.uni $L__BB0_5;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(129+i*1), [%r(21+i*-1)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(242+i*-1), %f130, %f(129+i*2), %f(242+i*-1);
ld.shared.f32 %f(131+i*1), [%r21+(452+i*448)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(240+i*-1), %f130, %f(132+i*1), %f(240+i*-1);
ld.shared.f32 %f(133+i*1), [%r21+(1348+i*448)];
fma.rn.f32 %f227, %f130, %f145, %f227;
add.s32 %r126, %r126, 2;
$L__BB0_5:
for.size.3 i in range(0, 2, 1):
add.s32 %r(113+i*3), %r(126+i*-108), %r(17+i*109);
shl.b32 %r(114+i*3), %r(113+i*3), 2;
add.s32 %r(20+i*1), %r(27+i*1), %r(114+i*3);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(112+i*1), [%r(21+i*-1)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(242+i*-1), %f113, %f(112+i*2), %f(242+i*-1);
ld.shared.f32 %f(114+i*1), [%r21+(448+i*448)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(240+i*-1), %f113, %f(115+i*1), %f(240+i*-1);
ld.shared.f32 %f(116+i*1), [%r21+(1344+i*448)];
fma.rn.f32 %f227, %f113, %f128, %f227;
setp.eq.s32 %p2, %r126, 6;
@%p2 bra $L__BB0_7;
bra.uni $L__BB0_6;
$L__BB0_7:
add.s32 %r125, %r125, 1;
setp.lt.u32 %p3, %r125, 4;
@%p3 bra $L__BB0_4;
add.s32 %r121, %r121, 1;
setp.lt.u32 %p4, %r121, 28;
@%p4 bra $L__BB0_1;
shl.b32 %r119, %r3, 6;
add.s32 %r120, %r119, %r1;
mul.wide.s32 %rd27, %r120, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f146, [%rd28];
add.f32 %f147, %f242, %f146;
mov.f32 %f148, 0f00000000;
max.f32 %f149, %f147, %f148;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 15, 1):
st.global.f32 [%rd29+(0+i*16)], %f(149+i*3);
ld.global.nc.f32 %f(150+i*3), [%rd28+(16+i*16)];
add.f32 %f(151+i*3), %f(241+i*-1), %f(150+i*3);
max.f32 %f(152+i*3), %f(151+i*3), %f148;
st.global.f32 [%rd29+240], %f194;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[28];
__shared__ float placeholder_shared[1792];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 28; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 28 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 12] = placeholder[index1 + 12];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 20] = placeholder[index1 + 20];
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 448; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 50176 + (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) / 28 * 784 + k_outer_outer * 28 + (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) % 28;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 4; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 7; ++k_inner) {
int index4 = k_outer_inner * 7 + k_inner;
int index5 = (int)threadIdx.x * 28 + k_outer_inner * 7 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 112]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 224]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 336]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 448]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 560]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 672]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 896]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1008]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1120]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1232]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1344]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1456]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1568]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1680]);
}
}
}
for (int vthread_s = 0; vthread_s < 16; ++vthread_s) {
int index6 = (int)blockIdx.x * 64 + vthread_s * 4 + (int)threadIdx.x;
int index7 = vthread_s;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp1xwgmvpj | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<659>;
.reg .b32 %r<48>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[8];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[1024];
ld.param.u64 %rd8, [default_function_kernel0_param_0];
ld.param.u64 %rd9, [default_function_kernel0_param_1];
cvta.to.global.u64 %rd12, %rd9;
cvta.to.global.u64 %rd2, %rd8;
mov.u32 %r1, %tid.x;
shl.b32 %r16, %r1, 3;
mov.u32 %r17, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r17, %r16;
mul.lo.s32 %r3, %r1, 784;
shl.b32 %r18, %r1, 4;
mov.u32 %r19, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r20, %r19, %r18;
add.s32 %r5, %r20, 64;
add.s64 %rd4, %rd12, 25088;
mul.lo.s32 %r6, %r1, 1568;
mov.u32 %r44, 0;
mov.f32 %f531, 0f00000000;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(532+i*1), %f531;
$L__BB0_1:
bar.sync 0;
setp.lt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_2;
bra.uni $L__BB0_3;
$L__BB0_2:
shl.b32 %r21, %r44, 1;
add.s32 %r22, %r21, %r3;
mul.wide.s32 %rd13, %r22, 4;
add.s64 %rd14, %rd2, %rd13;
ld.global.nc.v2.u32 {%r23, %r24}, [%rd14];
st.shared.v2.u32 [%r2], {%r23, %r24};
$L__BB0_3:
shl.b32 %r28, %r44, 1;
add.s32 %r29, %r6, %r28;
mul.wide.s32 %rd15, %r29, 4;
add.s64 %rd23, %rd4, %rd15;
mov.u32 %r46, 64;
mov.u32 %r45, %r5;
$L__BB0_4:
for.size.2 i in range(0, 2, 1):
for.size.1 j in range(0, 2, 1):
ld.global.nc.f32 %f(257+i*2+j*1), [%rd23+(-21948+i*-3136+j*-4)];
for.size.5 i in range(0, 3, 1):
st.shared.v4.f32 [%r45+(-64+i*32)], {%f(260+i*4), %f(259+i*4), %f(258+i*4), %f(257+i*4)};
ld.global.nc.f32 %f(261+i*4), [%rd23+(-9404+i*12544)];
ld.global.nc.f32 %f(262+i*4), [%rd23+(-9408+i*12544)];
ld.global.nc.f32 %f(263+i*4), [%rd23+(-12540+i*12544)];
ld.global.nc.f32 %f(264+i*4), [%rd23+(-12544+i*12544)];
st.shared.v4.f32 [%r45+32], {%f272, %f271, %f270, %f269};
add.s32 %r45, %r45, 128;
add.s64 %rd23, %rd23, 50176;
add.s32 %r46, %r46, 128;
setp.ne.s32 %p2, %r46, 1088;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r47, 0;
$L__BB0_6:
mov.u32 %r43, _ZZ24default_function_kernel0E20placeholder_d_shared;
mov.u32 %r42, _ZZ24default_function_kernel0E18placeholder_shared;
for.size.2 i in range(0, 2, 1):
shl.b32 %r(41+i*-8), %r(1+i*31), (7+i*-5);
add.s32 %r(32+i*3), %r(47+i*-5), %r(41+i*-8);
ld.shared.f32 %f273, [%r35];
shl.b32 %r36, %r47, 2;
add.s32 %r38, %r43, %r36;
ld.shared.f32 %f274, [%r38];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(594+i*-1), %f274, %f(273+i*2), %f(594+i*-1);
ld.shared.f32 %f(275+i*1), [%r35+(8+i*8)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(592+i*-1), %f274, %f(276+i*1), %f(592+i*-1);
ld.shared.f32 %f(277+i*1), [%r35+(24+i*8)];
fma.rn.f32 %f531, %f274, %f337, %f531;
add.s32 %r39, %r47, 1;
mov.u32 %r47, 1;
setp.lt.u32 %p3, %r39, 2;
@%p3 bra $L__BB0_6;
add.s32 %r44, %r44, 1;
setp.lt.u32 %p4, %r44, 392;
@%p4 bra $L__BB0_1;
ld.param.u64 %rd22, [default_function_kernel0_param_2];
cvta.to.global.u64 %rd21, %rd22;
ld.param.u64 %rd20, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd19, %rd20;
shl.b32 %r40, %r1, 6;
mul.wide.s32 %rd16, %r40, 4;
add.s64 %rd17, %rd19, %rd16;
ld.global.nc.f32 %f338, [%rd17];
add.f32 %f339, %f594, %f338;
mov.f32 %f340, 0f00000000;
max.f32 %f341, %f339, %f340;
add.s64 %rd18, %rd21, %rd16;
for.size.4 i in range(0, 63, 1):
st.global.f32 [%rd18+(0+i*4)], %f(341+i*3);
ld.global.nc.f32 %f(342+i*3), [%rd17+(4+i*4)];
add.f32 %f(343+i*3), %f(593+i*-1), %f(342+i*3);
max.f32 %f(344+i*3), %f(343+i*3), %f340;
st.global.f32 [%rd18+252], %f530;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[2];
__shared__ float placeholder_shared[256];
for (int j_inner_init = 0; j_inner_init < 64; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 392; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
((float2*)(placeholder_d_shared + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder + (int)threadIdx.x * 784 + k_outer_outer * 2))[0];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 32; ++ax0_ax1_fused_outer_outer) {
int4 _1;
int4 _2;
int4 _3;
int4 _4 = make_int4(ax0_ax1_fused_outer_outer * 3136 + (int)threadIdx.x * 1568, ax0_ax1_fused_outer_outer * 3136 + (int)threadIdx.x * 1568, ax0_ax1_fused_outer_outer * 3136 + (int)threadIdx.x * 1568, ax0_ax1_fused_outer_outer * 3136 + (int)threadIdx.x * 1568);
int4 _5;
int4 _6;
int4 _7 = make_int4(0+1*0, 0+1*1, 0+1*2, 0+1*3);
int4 _8 = make_int4(2, 2, 2, 2);
_6.x = _7.x%_8.x;
_6.y = _7.y%_8.y;
_6.z = _7.z%_8.z;
_6.w = _7.w%_8.w;
int4 _9;
int4 _10 = make_int4(0+1*0, 0+1*1, 0+1*2, 0+1*3);
int4 _11 = make_int4(2, 2, 2, 2);
_9.x = _10.x/_11.x;
_9.y = _10.y/_11.y;
_9.z = _10.z/_11.z;
_9.w = _10.w/_11.w;
int4 _12;
ushort4 _13;
ushort4 _14;
ushort4 _15;
int4 _16 = make_int4(2, 2, 2, 2);
int4 _17 = make_int4(0, 0, 0, 0);
_15.x = _16.x>=_17.x;
_15.y = _16.y>=_17.y;
_15.z = _16.z>=_17.z;
_15.w = _16.w>=_17.w;
ushort4 _18;
int4 _19 = make_int4(0, 0, 0, 0);
_18.x = _6.x>=_19.x;
_18.y = _6.y>=_19.y;
_18.z = _6.z>=_19.z;
_18.w = _6.w>=_19.w;
_14.x = _15.x&&_18.x;
_14.y = _15.y&&_18.y;
_14.z = _15.z&&_18.z;
_14.w = _15.w&&_18.w;
ushort4 _20;
ushort4 _21;
int4 _22 = make_int4(2, 2, 2, 2);
int4 _23 = make_int4(0, 0, 0, 0);
_21.x = _22.x<_23.x;
_21.y = _22.y<_23.y;
_21.z = _22.z<_23.z;
_21.w = _22.w<_23.w;
ushort4 _24;
int4 _25 = make_int4(0, 0, 0, 0);
_24.x = _6.x<=_25.x;
_24.y = _6.y<=_25.y;
_24.z = _6.z<=_25.z;
_24.w = _6.w<=_25.w;
_20.x = _21.x&&_24.x;
_20.y = _21.y&&_24.y;
_20.z = _21.z&&_24.z;
_20.w = _21.w&&_24.w;
_13.x = _14.x||_20.x;
_13.y = _14.y||_20.y;
_13.z = _14.z||_20.z;
_13.w = _14.w||_20.w;
int4 _26;
int4 _27 = make_int4(1, 1, 1, 1);
_26.x = _9.x-_27.x;
_26.y = _9.y-_27.y;
_26.z = _9.z-_27.z;
_26.w = _9.w-_27.w;
_12.x = bool(_13.x)?_9.x:_26.x;
_12.y = bool(_13.y)?_9.y:_26.y;
_12.z = bool(_13.z)?_9.z:_26.z;
_12.w = bool(_13.w)?_9.w:_26.w;
int4 _28 = make_int4(784, 784, 784, 784);
_5.x = _12.x*_28.x;
_5.y = _12.y*_28.y;
_5.z = _12.z*_28.z;
_5.w = _12.w*_28.w;
_3.x = _4.x+_5.x;
_3.y = _4.y+_5.y;
_3.z = _4.z+_5.z;
_3.w = _4.w+_5.w;
int4 _29 = make_int4(k_outer_outer * 2, k_outer_outer * 2, k_outer_outer * 2, k_outer_outer * 2);
_2.x = _3.x+_29.x;
_2.y = _3.y+_29.y;
_2.z = _3.z+_29.z;
_2.w = _3.w+_29.w;
int4 _30;
int4 _31 = make_int4(0+1*0, 0+1*1, 0+1*2, 0+1*3);
int4 _32 = make_int4(2, 2, 2, 2);
_30.x = _31.x%_32.x;
_30.y = _31.y%_32.y;
_30.z = _31.z%_32.z;
_30.w = _31.w%_32.w;
int4 _33;
ushort4 _34;
ushort4 _35;
ushort4 _36;
int4 _37 = make_int4(2, 2, 2, 2);
int4 _38 = make_int4(0, 0, 0, 0);
_36.x = _37.x>=_38.x;
_36.y = _37.y>=_38.y;
_36.z = _37.z>=_38.z;
_36.w = _37.w>=_38.w;
ushort4 _39;
int4 _40 = make_int4(0, 0, 0, 0);
_39.x = _30.x>=_40.x;
_39.y = _30.y>=_40.y;
_39.z = _30.z>=_40.z;
_39.w = _30.w>=_40.w;
_35.x = _36.x&&_39.x;
_35.y = _36.y&&_39.y;
_35.z = _36.z&&_39.z;
_35.w = _36.w&&_39.w;
ushort4 _41;
ushort4 _42;
int4 _43 = make_int4(2, 2, 2, 2);
int4 _44 = make_int4(0, 0, 0, 0);
_42.x = _43.x<_44.x;
_42.y = _43.y<_44.y;
_42.z = _43.z<_44.z;
_42.w = _43.w<_44.w;
ushort4 _45;
int4 _46 = make_int4(0, 0, 0, 0);
_45.x = _30.x<=_46.x;
_45.y = _30.y<=_46.y;
_45.z = _30.z<=_46.z;
_45.w = _30.w<=_46.w;
_41.x = _42.x&&_45.x;
_41.y = _42.y&&_45.y;
_41.z = _42.z&&_45.z;
_41.w = _42.w&&_45.w;
_34.x = _35.x||_41.x;
_34.y = _35.y||_41.y;
_34.z = _35.z||_41.z;
_34.w = _35.w||_41.w;
int4 _47;
int4 _48 = make_int4(2, 2, 2, 2);
_47.x = _30.x+_48.x;
_47.y = _30.y+_48.y;
_47.z = _30.z+_48.z;
_47.w = _30.w+_48.w;
_33.x = bool(_34.x)?_30.x:_47.x;
_33.y = bool(_34.y)?_30.y:_47.y;
_33.z = bool(_34.z)?_30.z:_47.z;
_33.w = bool(_34.w)?_30.w:_47.w;
_1.x = _2.x+_33.x;
_1.y = _2.y+_33.y;
_1.z = _2.z+_33.z;
_1.w = _2.w+_33.w;
int index1 = _1.x;
int index2 = _1.y;
int index3 = _1.z;
int index4 = _1.w;
((float4*)(placeholder_shared + ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x * 4))[0] = make_float4(placeholder1[index1],placeholder1[index2],placeholder1[index3],placeholder1[index4]);
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int j_inner = 0; j_inner < 64; ++j_inner) {
int index5 = j_inner;
int index6 = k_outer_inner;
int index7 = (int)threadIdx.x * 128 + j_inner * 2 + k_outer_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index8 = (int)threadIdx.x * 64 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpddyc20a0 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<70>;
.reg .b32 %r<39>;
.reg .b64 %rd<26>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[16];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[1024];
ld.param.u64 %rd11, [default_function_kernel0_param_0];
ld.param.u64 %rd12, [default_function_kernel0_param_1];
ld.param.u64 %rd13, [default_function_kernel0_param_2];
ld.param.u64 %rd10, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd14, %rd12;
cvta.to.global.u64 %rd1, %rd13;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(10+i*1), %r1, (2+i*2);
mov.u32 %r12, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r12, %r11;
shr.s32 %r13, %r1, 2;
and.b32 %r14, %r1, 3;
mov.u32 %r15, _ZZ24default_function_kernel0E18placeholder_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(3+i*1), %r15, %r(10+i*1);
mul.lo.s32 %r16, %r1, 784;
cvta.to.global.u64 %rd15, %rd11;
mul.wide.s32 %rd16, %r16, 4;
add.s64 %rd17, %rd15, %rd16;
add.s64 %rd25, %rd17, 16;
mov.u32 %r5, %ctaid.x;
mul.lo.s32 %r17, %r13, 784;
mad.lo.s32 %r18, %r5, 50176, %r17;
or.b32 %r19, %r18, %r14;
mul.wide.s32 %rd18, %r19, 4;
add.s64 %rd19, %rd14, %rd18;
add.s64 %rd24, %rd19, 100352;
mov.f32 %f68, 0f00000000;
mov.u32 %r38, 0;
mov.f32 %f69, %f68;
$L__BB0_1:
bar.sync 0;
add.s64 %rd6, %rd25, -16;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.v4.u32 {%r20, %r21, %r22, %r23}, [%rd6];
st.shared.v4.u32 [%r2], {%r20, %r21, %r22, %r23};
$L__BB0_3:
add.s64 %rd7, %rd24, -100352;
for.size.2 i in range(0, 8, 1):
ld.global.nc.f32 %f(9+i*1), [%rd24+(-100352+i*25088)];
st.shared.f32 [%r3+(0+i*128)], %f(9+i*1);
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(17+i*5), [%r4+(0+i*4)];
ld.shared.f32 %f(18+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(19+i*5), %f(18+i*5), %f(17+i*5), %f(68+i*-49);
ld.shared.f32 %f(20+i*5), [%r4+(512+i*4)];
fma.rn.f32 %f(21+i*5), %f(18+i*5), %f(20+i*5), %f(69+i*-48);
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(27+i*5), [%r4+(8+i*4)];
ld.shared.f32 %f(28+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(29+i*-26), %f(28+i*5), %f(27+i*5), %f(24+i*5);
ld.shared.f32 %f(30+i*4), [%r4+(520+i*4)];
fma.rn.f32 %f(31+i*-27), %f(28+i*5), %f(30+i*4), %f(26+i*5);
bar.sync 0;
@%p1 bra $L__BB0_5;
ld.global.nc.v4.u32 {%r28, %r29, %r30, %r31}, [%rd6+16];
st.shared.v4.u32 [%r2], {%r28, %r29, %r30, %r31};
$L__BB0_5:
for.size.2 i in range(0, 8, 1):
ld.global.nc.f32 %f(35+i*1), [%rd7+(16+i*25088)];
st.shared.f32 [%r3+(0+i*128)], %f(35+i*1);
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(43+i*5), [%r4+(0+i*4)];
ld.shared.f32 %f(44+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(45+i*5), %f(44+i*5), %f(43+i*5), %f(3+i*42);
ld.shared.f32 %f(46+i*5), [%r4+(512+i*4)];
fma.rn.f32 %f(47+i*5), %f(44+i*5), %f(46+i*5), %f(4+i*43);
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(53+i*5), [%r4+(8+i*4)];
ld.shared.f32 %f(54+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(55+i*13), %f(54+i*5), %f(53+i*5), %f(50+i*5);
ld.shared.f32 %f(56+i*4), [%r4+(520+i*4)];
fma.rn.f32 %f(57+i*12), %f(54+i*5), %f(56+i*4), %f(52+i*5);
for.size.1 i in range(0, 2, 1):
add.s64 %rd(25+i*-1), %rd(25+i*-1), 32;
add.s32 %r38, %r38, 2;
setp.ne.s32 %p3, %r38, 196;
@%p3 bra $L__BB0_1;
cvta.to.global.u64 %rd20, %rd10;
shl.b32 %r36, %r5, 6;
add.s32 %r37, %r36, %r1;
mul.wide.s32 %rd21, %r37, 4;
add.s64 %rd22, %rd20, %rd21;
ld.global.nc.f32 %f61, [%rd22];
add.f32 %f62, %f68, %f61;
mov.f32 %f63, 0f00000000;
max.f32 %f64, %f62, %f63;
add.s64 %rd23, %rd1, %rd21;
st.global.f32 [%rd23], %f64;
ld.global.nc.f32 %f65, [%rd22+128];
add.f32 %f66, %f69, %f65;
max.f32 %f67, %f66, %f63;
st.global.f32 [%rd23+128], %f67;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[4];
__shared__ float placeholder_shared[256];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 196; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
((float4*)(placeholder_d_shared + (int)threadIdx.x * 4))[0] = ((float4*)(placeholder + (int)threadIdx.x * 784 + k_outer_outer * 4))[0];
}
int index0 = (int)threadIdx.x;
int index1 = (int)blockIdx.x * 50176 + ((int)threadIdx.x >> 2) * 784 + k_outer_outer * 4 + ((int)threadIdx.x & 3);
placeholder_shared[index0] = placeholder1[index1];
placeholder_shared[index0 + 32] = placeholder1[index1 + 6272];
placeholder_shared[index0 + 64] = placeholder1[index1 + 12544];
placeholder_shared[index0 + 96] = placeholder1[index1 + 18816];
placeholder_shared[index0 + 128] = placeholder1[index1 + 25088];
placeholder_shared[index0 + 160] = placeholder1[index1 + 31360];
placeholder_shared[index0 + 192] = placeholder1[index1 + 37632];
placeholder_shared[index0 + 224] = placeholder1[index1 + 43904];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 4];
int index2 = (int)threadIdx.x * 4 + 128;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index2];
int index3 = (int)threadIdx.x * 4 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index3];
int index4 = (int)threadIdx.x * 4 + 129;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index4];
int index5 = (int)threadIdx.x * 4 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index5];
int index6 = (int)threadIdx.x * 4 + 130;
T_dense[1] = T_dense[1] + placeholder_d_shared[2] * placeholder_shared[index6];
int index7 = (int)threadIdx.x * 4 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index7];
int index8 = (int)threadIdx.x * 4 + 131;
T_dense[1] = T_dense[1] + placeholder_d_shared[3] * placeholder_shared[index8];
}
int index9 = (int)blockIdx.x * 64 + (int)threadIdx.x;
T_relu[index9] = max(T_dense[0] + placeholder2[index9], 0.000000e+00f);
T_relu[index9 + 32] = max(T_dense[1] + placeholder2[index9 + 32], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpt255kjz8 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<6>;
.reg .f32 %f<779>;
.reg .b32 %r<137>;
.reg .b64 %rd<32>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[1568];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd1, [default_function_kernel0_param_0];
ld.param.u64 %rd2, [default_function_kernel0_param_1];
mov.u32 %r14, %tid.x;
mul.lo.s32 %r1, %r14, 6272;
mov.u32 %r132, 0;
mov.f32 %f715, 0f00000000;
cvta.to.global.u64 %rd5, %rd1;
cvta.to.global.u64 %rd8, %rd2;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(716+i*1), %f715;
$L__BB0_1:
mov.u32 %r133, 0;
bar.sync 0;
$L__BB0_2:
mul.lo.s32 %r129, %r132, 392;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(17+i*1), %r(14+i*119), (1+i*1);
add.s32 %r19, %r18, %r17;
shl.b32 %r20, %r19, 2;
mov.u32 %r21, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(22+i*1), %r(21+i*108), %r(20+i*-3);
add.s32 %r24, %r23, %r18;
mul.wide.s32 %rd6, %r24, 4;
add.s64 %rd7, %rd5, %rd6;
for.size.2 i in range(0, 2, 1):
ld.global.nc.v2.u32 {%r(25+i*4), %r(26+i*4)}, [%rd7+(0+i*16)];
st.shared.v2.u32 [%r22+(0+i*16)], {%r(25+i*4), %r(26+i*4)};
add.s32 %r133, %r133, 2;
setp.ne.s32 %p1, %r133, 98;
@%p1 bra $L__BB0_2;
mov.u32 %r134, 0;
$L__BB0_4:
mul.lo.s32 %r125, %r132, 392;
shl.b32 %r35, %r134, 1;
add.s32 %r36, %r35, %r14;
mul.hi.s32 %r37, %r36, 1402438301;
shr.u32 %r38, %r37, 31;
shr.s32 %r39, %r37, 7;
add.s32 %r40, %r39, %r38;
mul.lo.s32 %r41, %r40, 392;
sub.s32 %r42, %r36, %r41;
add.s32 %r43, %r42, %r125;
mad.lo.s32 %r44, %r40, 784, %r43;
mul.wide.s32 %rd9, %r44, 4;
add.s64 %rd10, %rd8, %rd9;
ld.global.nc.f32 %f321, [%rd10];
shl.b32 %r45, %r36, 2;
mov.u32 %r46, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r47, %r46, %r45;
for.size.13 i in range(0, 7, 1):
st.shared.f32 [%r47+(0+i*8)], %f(321+i*1);
add.s32 %r(48+i*9), %r36, (2+i*2);
mul.hi.s32 %r(49+i*9), %r(48+i*9), 1402438301;
shr.u32 %r(50+i*9), %r(49+i*9), 31;
shr.s32 %r(51+i*9), %r(49+i*9), 7;
add.s32 %r(52+i*9), %r(51+i*9), %r(50+i*9);
mul.lo.s32 %r(53+i*9), %r(52+i*9), 392;
sub.s32 %r(54+i*9), %r(48+i*9), %r(53+i*9);
add.s32 %r(55+i*9), %r(54+i*9), %r125;
mad.lo.s32 %r(56+i*9), %r(52+i*9), 784, %r(55+i*9);
mul.wide.s32 %rd(11+i*2), %r(56+i*9), 4;
add.s64 %rd(12+i*2), %rd8, %rd(11+i*2);
ld.global.nc.f32 %f(322+i*1), [%rd(12+i*2)+0];
st.shared.f32 [%r47+56], %f328;
add.s32 %r134, %r134, 8;
setp.ne.s32 %p2, %r134, 25088;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r135, 0;
$L__BB0_6:
mov.u32 %r136, 0;
$L__BB0_7:
shl.b32 %r131, %r135, 1;
mov.u32 %r127, _ZZ24default_function_kernel0E18placeholder_shared;
mov.u32 %r126, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.3 i in range(0, 2, 1):
add.s32 %r(114+i*4), %r(136+i*-22), %r(131+i*-130);
shl.b32 %r(115+i*4), %r(114+i*4), 2;
add.s32 %r(117+i*4), %r(126+i*1), %r(115+i*4);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(329+i*1), [%r(121+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(778+i*-16), %f330, %f(329+i*2), %f(778+i*-16);
ld.shared.f32 %f(331+i*1), [%r121+(50176+i*50176)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(746+i*-16), %f330, %f(332+i*1), %f(746+i*-16);
ld.shared.f32 %f(333+i*1), [%r121+(150528+i*-148960)];
for.size.2 i in range(0, 3, 1):
fma.rn.f32 %f(777+i*-16), %f330, %f(334+i*1), %f(777+i*-16);
ld.shared.f32 %f(335+i*1), [%r121+(51744+i*50176)];
for.size.6 i in range(0, 14, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(729+i*-1+j*47), %f330, %f(337+i*4+j*1), %f(729+i*-1+j*47);
ld.shared.f32 %f(338+i*4+j*1), [%r121+(3136+i*1568+j*50176)];
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(760+i*-1+j*-16), %f330, %f(339+i*4+j*1), %f(760+i*-1+j*-16);
ld.shared.f32 %f(340+i*4+j*1), [%r121+(103488+i*1568+j*50176)];
fma.rn.f32 %f715, %f330, %f393, %f715;
add.s32 %r122, %r136, 1;
mov.u32 %r136, 1;
setp.lt.u32 %p3, %r122, 2;
@%p3 bra $L__BB0_7;
add.s32 %r135, %r135, 1;
setp.lt.u32 %p4, %r135, 196;
@%p4 bra $L__BB0_6;
add.s32 %r132, %r132, 1;
setp.lt.u32 %p5, %r132, 2;
@%p5 bra $L__BB0_1;
ld.param.u64 %rd31, [default_function_kernel0_param_2];
ld.param.u64 %rd30, [default_function_kernel0_param_3];
shl.b32 %r124, %r14, 4;
cvta.to.global.u64 %rd25, %rd30;
mul.wide.s32 %rd26, %r124, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f394, [%rd27];
add.f32 %f395, %f778, %f394;
mov.f32 %f396, 0f00000000;
max.f32 %f397, %f395, %f396;
cvta.to.global.u64 %rd28, %rd31;
add.s64 %rd29, %rd28, %rd26;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd29+(0+i*128)], %f(397+i*3);
ld.global.nc.f32 %f(398+i*3), [%rd27+(128+i*128)];
add.f32 %f(399+i*3), %f(762+i*-16), %f(398+i*3);
max.f32 %f(400+i*3), %f(399+i*3), %f396;
for.size.10 i in range(0, 15, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(384+i*4+j*-380)], %f(406+i*12+j*3);
ld.global.nc.f32 %f(407+i*12+j*3), [%rd27+(4+i*4+j*128)];
add.f32 %f(408+i*12+j*3), %f(777+i*-1+j*-16), %f(407+i*12+j*3);
max.f32 %f(409+i*12+j*3), %f(408+i*12+j*3), %f396;
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(132+i*4+j*128)], %f(412+i*12+j*3);
ld.global.nc.f32 %f(413+i*12+j*3), [%rd27+(260+i*4+j*128)];
add.f32 %f(414+i*12+j*3), %f(745+i*-1+j*-16), %f(413+i*12+j*3);
max.f32 %f(415+i*12+j*3), %f(414+i*12+j*3), %f396;
st.global.f32 [%rd29+444], %f586;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[392];
__shared__ float placeholder_shared[50176];
for (int j_inner_init = 0; j_inner_init < 16; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
T_dense[index0 + 16] = 0.000000e+00f;
T_dense[index0 + 32] = 0.000000e+00f;
T_dense[index0 + 48] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 2; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_d_shared + ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder + k_outer_outer * 392 + ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x * 2))[0];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 25088; ++ax0_ax1_fused_outer_outer1) {
int index1 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
int index2 = (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) / 392 * 784 + k_outer_outer * 392 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) % 392;
placeholder_shared[index1] = placeholder1[index2];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 2; ++k_inner) {
for (int j_inner = 0; j_inner < 16; ++j_inner) {
int index3 = j_inner;
int index4 = k_outer_inner * 2 + k_inner;
int index5 = (int)threadIdx.x * 6272 + j_inner * 392 + k_outer_inner * 2 + k_inner;
T_dense[index3] = T_dense[index3] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[index3 + 16] = T_dense[index3 + 16] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 12544]);
T_dense[index3 + 32] = T_dense[index3 + 32] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 25088]);
T_dense[index3 + 48] = T_dense[index3 + 48] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 37632]);
}
}
}
}
for (int ax1_inner = 0; ax1_inner < 16; ++ax1_inner) {
int index6 = (int)threadIdx.x * 16 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 32] = max(T_dense[index7 + 16] + placeholder2[index6 + 32], 0.000000e+00f);
T_relu[index6 + 64] = max(T_dense[index7 + 32] + placeholder2[index6 + 64], 0.000000e+00f);
T_relu[index6 + 96] = max(T_dense[index7 + 48] + placeholder2[index6 + 96], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmphyw2eb5o | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<807>;
.reg .b32 %r<127>;
.reg .b64 %rd<32>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[224];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[28672];
ld.param.u64 %rd1, [default_function_kernel0_param_0];
ld.param.u64 %rd2, [default_function_kernel0_param_1];
mov.u32 %r13, %tid.x;
mul.lo.s32 %r1, %r13, 3584;
mov.u32 %r123, 0;
mov.f32 %f743, 0f00000000;
cvta.to.global.u64 %rd5, %rd1;
shl.b32 %r17, %r13, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r19, %r18, %r17;
cvta.to.global.u64 %rd8, %rd2;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(744+i*1), %f743;
$L__BB0_1:
mov.u32 %r124, 0;
bar.sync 0;
mul.lo.s32 %r3, %r123, 56;
add.s32 %r16, %r3, %r13;
mul.wide.s32 %rd6, %r16, 4;
add.s64 %rd7, %rd5, %rd6;
for.size.2 i in range(0, 28, 1):
ld.global.nc.f32 %f(321+i*1), [%rd7+(0+i*8)];
st.shared.f32 [%r19+(0+i*8)], %f(321+i*1);
$L__BB0_2:
mul.lo.s32 %r117, %r123, 56;
shl.b32 %r21, %r124, 1;
add.s32 %r22, %r21, %r13;
mul.hi.s32 %r23, %r22, -1840700269;
add.s32 %r24, %r23, %r22;
shr.u32 %r25, %r24, 31;
shr.s32 %r26, %r24, 5;
add.s32 %r27, %r26, %r25;
mul.lo.s32 %r28, %r27, 56;
sub.s32 %r29, %r22, %r28;
add.s32 %r30, %r29, %r117;
mad.lo.s32 %r31, %r27, 784, %r30;
mul.wide.s32 %rd9, %r31, 4;
add.s64 %rd10, %rd8, %rd9;
ld.global.nc.f32 %f349, [%rd10];
shl.b32 %r32, %r22, 2;
mov.u32 %r33, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r34, %r33, %r32;
for.size.14 i in range(0, 7, 1):
st.shared.f32 [%r34+(0+i*8)], %f(349+i*1);
add.s32 %r(35+i*10), %r22, (2+i*2);
mul.hi.s32 %r(36+i*10), %r(35+i*10), -1840700269;
add.s32 %r(37+i*10), %r(36+i*10), %r(35+i*10);
shr.u32 %r(38+i*10), %r(37+i*10), 31;
shr.s32 %r(39+i*10), %r(37+i*10), 5;
add.s32 %r(40+i*10), %r(39+i*10), %r(38+i*10);
mul.lo.s32 %r(41+i*10), %r(40+i*10), 56;
sub.s32 %r(42+i*10), %r(35+i*10), %r(41+i*10);
add.s32 %r(43+i*10), %r(42+i*10), %r117;
mad.lo.s32 %r(44+i*10), %r(40+i*10), 784, %r(43+i*10);
mul.wide.s32 %rd(11+i*2), %r(44+i*10), 4;
add.s64 %rd(12+i*2), %rd8, %rd(11+i*2);
ld.global.nc.f32 %f(350+i*1), [%rd(12+i*2)+0];
st.shared.f32 [%r34+56], %f356;
add.s32 %r124, %r124, 8;
setp.ne.s32 %p1, %r124, 3584;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r125, 0;
$L__BB0_4:
mov.u32 %r126, 0;
$L__BB0_5:
mul.lo.s32 %r121, %r125, 14;
mov.u32 %r119, _ZZ24default_function_kernel0E18placeholder_shared;
mov.u32 %r118, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.3 i in range(0, 2, 1):
add.s32 %r(107+i*4), %r(126+i*-19), %r(121+i*-120);
shl.b32 %r(108+i*4), %r(107+i*4), 2;
add.s32 %r(110+i*4), %r(118+i*1), %r(108+i*4);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(357+i*1), [%r(114+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(806+i*-1), %f358, %f(357+i*2), %f(806+i*-1);
ld.shared.f32 %f(359+i*1), [%r114+(224+i*224)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(804+i*-1), %f358, %f(360+i*1), %f(804+i*-1);
ld.shared.f32 %f(361+i*1), [%r114+(672+i*224)];
fma.rn.f32 %f743, %f358, %f421, %f743;
add.s32 %r126, %r126, 1;
setp.ne.s32 %p2, %r126, 14;
@%p2 bra $L__BB0_5;
add.s32 %r125, %r125, 1;
setp.lt.u32 %p3, %r125, 4;
@%p3 bra $L__BB0_4;
add.s32 %r123, %r123, 1;
setp.lt.u32 %p4, %r123, 14;
@%p4 bra $L__BB0_1;
ld.param.u64 %rd31, [default_function_kernel0_param_2];
ld.param.u64 %rd30, [default_function_kernel0_param_3];
shl.b32 %r116, %r13, 6;
cvta.to.global.u64 %rd25, %rd30;
mul.wide.s32 %rd26, %r116, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f422, [%rd27];
add.f32 %f423, %f806, %f422;
mov.f32 %f424, 0f00000000;
max.f32 %f425, %f423, %f424;
cvta.to.global.u64 %rd28, %rd31;
add.s64 %rd29, %rd28, %rd26;
for.size.4 i in range(0, 63, 1):
st.global.f32 [%rd29+(0+i*4)], %f(425+i*3);
ld.global.nc.f32 %f(426+i*3), [%rd27+(4+i*4)];
add.f32 %f(427+i*3), %f(805+i*-1), %f(426+i*3);
max.f32 %f(428+i*3), %f(427+i*3), %f424;
st.global.f32 [%rd29+252], %f614;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[56];
__shared__ float placeholder_shared[7168];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
T_dense[32] = 0.000000e+00f;
T_dense[33] = 0.000000e+00f;
T_dense[34] = 0.000000e+00f;
T_dense[35] = 0.000000e+00f;
T_dense[36] = 0.000000e+00f;
T_dense[37] = 0.000000e+00f;
T_dense[38] = 0.000000e+00f;
T_dense[39] = 0.000000e+00f;
T_dense[40] = 0.000000e+00f;
T_dense[41] = 0.000000e+00f;
T_dense[42] = 0.000000e+00f;
T_dense[43] = 0.000000e+00f;
T_dense[44] = 0.000000e+00f;
T_dense[45] = 0.000000e+00f;
T_dense[46] = 0.000000e+00f;
T_dense[47] = 0.000000e+00f;
T_dense[48] = 0.000000e+00f;
T_dense[49] = 0.000000e+00f;
T_dense[50] = 0.000000e+00f;
T_dense[51] = 0.000000e+00f;
T_dense[52] = 0.000000e+00f;
T_dense[53] = 0.000000e+00f;
T_dense[54] = 0.000000e+00f;
T_dense[55] = 0.000000e+00f;
T_dense[56] = 0.000000e+00f;
T_dense[57] = 0.000000e+00f;
T_dense[58] = 0.000000e+00f;
T_dense[59] = 0.000000e+00f;
T_dense[60] = 0.000000e+00f;
T_dense[61] = 0.000000e+00f;
T_dense[62] = 0.000000e+00f;
T_dense[63] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 14; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 56 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
placeholder_d_shared[index0 + 6] = placeholder[index1 + 6];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 10] = placeholder[index1 + 10];
placeholder_d_shared[index0 + 12] = placeholder[index1 + 12];
placeholder_d_shared[index0 + 14] = placeholder[index1 + 14];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 18] = placeholder[index1 + 18];
placeholder_d_shared[index0 + 20] = placeholder[index1 + 20];
placeholder_d_shared[index0 + 22] = placeholder[index1 + 22];
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
placeholder_d_shared[index0 + 26] = placeholder[index1 + 26];
placeholder_d_shared[index0 + 28] = placeholder[index1 + 28];
placeholder_d_shared[index0 + 30] = placeholder[index1 + 30];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 34] = placeholder[index1 + 34];
placeholder_d_shared[index0 + 36] = placeholder[index1 + 36];
placeholder_d_shared[index0 + 38] = placeholder[index1 + 38];
placeholder_d_shared[index0 + 40] = placeholder[index1 + 40];
placeholder_d_shared[index0 + 42] = placeholder[index1 + 42];
placeholder_d_shared[index0 + 44] = placeholder[index1 + 44];
placeholder_d_shared[index0 + 46] = placeholder[index1 + 46];
placeholder_d_shared[index0 + 48] = placeholder[index1 + 48];
placeholder_d_shared[index0 + 50] = placeholder[index1 + 50];
placeholder_d_shared[index0 + 52] = placeholder[index1 + 52];
placeholder_d_shared[index0 + 54] = placeholder[index1 + 54];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 3584; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x) / 56 * 784 + k_outer_outer * 56 + (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x) % 56;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 4; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 14; ++k_inner) {
int index4 = k_outer_inner * 14 + k_inner;
int index5 = (int)threadIdx.x * 3584 + k_outer_inner * 14 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 56]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 112]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 168]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 224]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 280]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 336]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 392]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 448]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 504]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 560]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 616]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 672]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 728]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 840]);
T_dense[16] = T_dense[16] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 896]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 952]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1008]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1064]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1120]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1176]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1232]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1288]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1344]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1400]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1456]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1512]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1568]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1624]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1680]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1736]);
T_dense[32] = T_dense[32] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1792]);
T_dense[33] = T_dense[33] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1848]);
T_dense[34] = T_dense[34] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1904]);
T_dense[35] = T_dense[35] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1960]);
T_dense[36] = T_dense[36] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2016]);
T_dense[37] = T_dense[37] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2072]);
T_dense[38] = T_dense[38] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2128]);
T_dense[39] = T_dense[39] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2184]);
T_dense[40] = T_dense[40] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2240]);
T_dense[41] = T_dense[41] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2296]);
T_dense[42] = T_dense[42] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2352]);
T_dense[43] = T_dense[43] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2408]);
T_dense[44] = T_dense[44] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2464]);
T_dense[45] = T_dense[45] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2520]);
T_dense[46] = T_dense[46] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2576]);
T_dense[47] = T_dense[47] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2632]);
T_dense[48] = T_dense[48] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2688]);
T_dense[49] = T_dense[49] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2744]);
T_dense[50] = T_dense[50] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2800]);
T_dense[51] = T_dense[51] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2856]);
T_dense[52] = T_dense[52] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2912]);
T_dense[53] = T_dense[53] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2968]);
T_dense[54] = T_dense[54] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3024]);
T_dense[55] = T_dense[55] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3080]);
T_dense[56] = T_dense[56] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3136]);
T_dense[57] = T_dense[57] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3192]);
T_dense[58] = T_dense[58] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3248]);
T_dense[59] = T_dense[59] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3304]);
T_dense[60] = T_dense[60] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3360]);
T_dense[61] = T_dense[61] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3416]);
T_dense[62] = T_dense[62] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3472]);
T_dense[63] = T_dense[63] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3528]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index6 = (int)threadIdx.x * 64 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpsbahwun6 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<667>;
.reg .b32 %r<47>;
.reg .b64 %rd<27>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[401408];
ld.param.u64 %rd11, [default_function_kernel0_param_0];
ld.param.u64 %rd8, [default_function_kernel0_param_1];
cvta.to.global.u64 %rd12, %rd11;
mov.u32 %r17, %tid.x;
shl.b32 %r18, %r17, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r20, %r19, %r18;
add.s32 %r41, %r20, 32;
cvt.s64.s32 %rd1, %r17;
mul.wide.s32 %rd13, %r17, 4;
add.s64 %rd14, %rd12, %rd13;
add.s64 %rd25, %rd14, 32;
mov.u32 %r42, 0;
$L__BB0_1:
for.size.2 i in range(0, 8, 1):
ld.global.nc.f32 %f(193+i*1), [%rd25+(-32+i*8)];
st.shared.f32 [%r41+(-32+i*8)], %f(193+i*1);
add.s32 %r41, %r41, 64;
add.s64 %rd25, %rd25, 64;
add.s32 %r42, %r42, 8;
setp.ne.s32 %p1, %r42, 392;
@%p1 bra $L__BB0_1;
cvt.u32.u64 %r22, %rd1;
shl.b32 %r23, %r22, 2;
mov.u32 %r24, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r25, %r24, %r23;
add.s32 %r43, %r25, 64;
cvta.to.global.u64 %rd15, %rd8;
shl.b64 %rd16, %rd1, 2;
add.s64 %rd17, %rd15, %rd16;
add.s64 %rd26, %rd17, 64;
mov.u32 %r44, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(201+i*1), [%rd26+(-64+i*8)];
st.shared.f32 [%r43+(-64+i*8)], %f(201+i*1);
add.s32 %r43, %r43, 128;
add.s64 %rd26, %rd26, 128;
add.s32 %r44, %r44, 16;
setp.ne.s32 %p2, %r44, 50176;
@%p2 bra $L__BB0_3;
bar.sync 0;
mov.u32 %r26, 0;
mov.f32 %f603, 0f00000000;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(604+i*1), %f603;
mov.u32 %r45, %r26;
$L__BB0_5:
mov.u32 %r46, %r26;
$L__BB0_6:
mul.lo.s32 %r40, %r45, 56;
mov.u32 %r39, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r28, %r46, %r40;
shl.b32 %r29, %r28, 2;
add.s32 %r31, %r39, %r29;
mad.lo.s32 %r33, %r17, 12544, %r28;
shl.b32 %r34, %r33, 2;
add.s32 %r36, %r24, %r34;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(281+i*1), [%r(36+i*-5)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(666+i*-1), %f282, %f(281+i*2), %f(666+i*-1);
ld.shared.f32 %f(283+i*1), [%r36+(100352+i*100352)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(664+i*-1), %f282, %f(284+i*1), %f(664+i*-1);
ld.shared.f32 %f(285+i*1), [%r36+(301056+i*-297920)];
for.size.2 i in range(0, 3, 1):
fma.rn.f32 %f(662+i*-1), %f282, %f(286+i*1), %f(662+i*-1);
ld.shared.f32 %f(287+i*1), [%r36+(103488+i*100352)];
for.size.3 i in range(0, 14, 1):
for.size.2 j in range(0, 4, 1):
fma.rn.f32 %f(659+i*-4+j*-1), %f282, %f(289+i*4+j*1), %f(659+i*-4+j*-1);
ld.shared.f32 %f(290+i*4+j*1), [%r36+(6272+i*3136+j*100352)];
fma.rn.f32 %f603, %f282, %f345, %f603;
add.s32 %r46, %r46, 1;
setp.ne.s32 %p3, %r46, 56;
@%p3 bra $L__BB0_6;
add.s32 %r45, %r45, 1;
setp.lt.u32 %p4, %r45, 14;
@%p4 bra $L__BB0_5;
ld.param.u64 %rd24, [default_function_kernel0_param_2];
ld.param.u64 %rd23, [default_function_kernel0_param_3];
shl.b32 %r38, %r17, 4;
cvta.to.global.u64 %rd18, %rd23;
mul.wide.s32 %rd19, %r38, 4;
add.s64 %rd20, %rd18, %rd19;
ld.global.nc.f32 %f346, [%rd20];
add.f32 %f347, %f666, %f346;
mov.f32 %f348, 0f00000000;
max.f32 %f349, %f347, %f348;
cvta.to.global.u64 %rd21, %rd24;
add.s64 %rd22, %rd21, %rd19;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd22+(0+i*128)], %f(349+i*3);
ld.global.nc.f32 %f(350+i*3), [%rd20+(128+i*128)];
add.f32 %f(351+i*3), %f(665+i*-1), %f(350+i*3);
max.f32 %f(352+i*3), %f(351+i*3), %f348;
for.size.10 i in range(0, 15, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd22+(384+i*4+j*-380)], %f(358+i*12+j*3);
ld.global.nc.f32 %f(359+i*12+j*3), [%rd20+(4+i*4+j*128)];
add.f32 %f(360+i*12+j*3), %f(662+i*-4+j*-1), %f(359+i*12+j*3);
max.f32 %f(361+i*12+j*3), %f(360+i*12+j*3), %f348;
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd22+(132+i*4+j*128)], %f(364+i*12+j*3);
ld.global.nc.f32 %f(365+i*12+j*3), [%rd20+(260+i*4+j*128)];
add.f32 %f(366+i*12+j*3), %f(660+i*-4+j*-1), %f(365+i*12+j*3);
max.f32 %f(367+i*12+j*3), %f(366+i*12+j*3), %f348;
st.global.f32 [%rd22+444], %f538;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[100352];
T_dense[0] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[32] = 0.000000e+00f;
T_dense[48] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[33] = 0.000000e+00f;
T_dense[49] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[34] = 0.000000e+00f;
T_dense[50] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[35] = 0.000000e+00f;
T_dense[51] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[36] = 0.000000e+00f;
T_dense[52] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[37] = 0.000000e+00f;
T_dense[53] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[38] = 0.000000e+00f;
T_dense[54] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[39] = 0.000000e+00f;
T_dense[55] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[40] = 0.000000e+00f;
T_dense[56] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[41] = 0.000000e+00f;
T_dense[57] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[42] = 0.000000e+00f;
T_dense[58] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[43] = 0.000000e+00f;
T_dense[59] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[44] = 0.000000e+00f;
T_dense[60] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[45] = 0.000000e+00f;
T_dense[61] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[46] = 0.000000e+00f;
T_dense[62] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
T_dense[47] = 0.000000e+00f;
T_dense[63] = 0.000000e+00f;
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 392; ++ax0_ax1_fused_outer_outer) {
int index0 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index0];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 50176; ++ax0_ax1_fused_outer_outer1) {
int index1 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index1];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 14; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 56; ++k_inner) {
int index2 = k_outer_inner * 56 + k_inner;
int index3 = (int)threadIdx.x * 12544 + k_outer_inner * 56 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[16] = T_dense[16] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 25088]);
T_dense[32] = T_dense[32] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 50176]);
T_dense[48] = T_dense[48] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 75264]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 784]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 25872]);
T_dense[33] = T_dense[33] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 50960]);
T_dense[49] = T_dense[49] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 76048]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 1568]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 26656]);
T_dense[34] = T_dense[34] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 51744]);
T_dense[50] = T_dense[50] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 76832]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 2352]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 27440]);
T_dense[35] = T_dense[35] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 52528]);
T_dense[51] = T_dense[51] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 77616]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 3136]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 28224]);
T_dense[36] = T_dense[36] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 53312]);
T_dense[52] = T_dense[52] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 78400]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 3920]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 29008]);
T_dense[37] = T_dense[37] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 54096]);
T_dense[53] = T_dense[53] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 79184]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 4704]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 29792]);
T_dense[38] = T_dense[38] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 54880]);
T_dense[54] = T_dense[54] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 79968]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 5488]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 30576]);
T_dense[39] = T_dense[39] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 55664]);
T_dense[55] = T_dense[55] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 80752]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 6272]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 31360]);
T_dense[40] = T_dense[40] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 56448]);
T_dense[56] = T_dense[56] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 81536]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 7056]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 32144]);
T_dense[41] = T_dense[41] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 57232]);
T_dense[57] = T_dense[57] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 82320]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 7840]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 32928]);
T_dense[42] = T_dense[42] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 58016]);
T_dense[58] = T_dense[58] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 83104]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 8624]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 33712]);
T_dense[43] = T_dense[43] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 58800]);
T_dense[59] = T_dense[59] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 83888]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 9408]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 34496]);
T_dense[44] = T_dense[44] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 59584]);
T_dense[60] = T_dense[60] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 84672]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 10192]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 35280]);
T_dense[45] = T_dense[45] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 60368]);
T_dense[61] = T_dense[61] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 85456]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 10976]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 36064]);
T_dense[46] = T_dense[46] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 61152]);
T_dense[62] = T_dense[62] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 86240]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 11760]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 36848]);
T_dense[47] = T_dense[47] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 61936]);
T_dense[63] = T_dense[63] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 87024]);
}
}
for (int ax1_inner = 0; ax1_inner < 16; ++ax1_inner) {
int index4 = (int)threadIdx.x * 16 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 32] = max(T_dense[index5 + 16] + placeholder2[index4 + 32], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[index5 + 32] + placeholder2[index4 + 64], 0.000000e+00f);
T_relu[index4 + 96] = max(T_dense[index5 + 48] + placeholder2[index4 + 96], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp8x_t1kud | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<48>;
.reg .f32 %f<76>;
.reg .b32 %r<117>;
.reg .b64 %rd<22>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[1568];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd7;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
for.size.2 i in range(0, 2, 1):
mul.lo.s32 %r(2+i*1), %r1, (14+i*378);
setp.lt.s32 %p(15+i*1), %r(1+i*1), (28+i*364);
and.pred %p1, %p15, %p16;
shl.b32 %r27, %r2, 2;
mov.u32 %r28, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r4, %r28, %r27;
for.size.2 i in range(0, 13, 1):
setp.lt.s32 %p(17+i*1), %r2, (391+i*-1);
and.pred %p(2+i*1), %p15, %p(17+i*1);
shl.b32 %r5, %r1, 2;
add.s32 %r6, %r5, 768;
shl.b32 %r29, %r1, 4;
mov.u32 %r30, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r31, %r30, %r29;
add.s32 %r7, %r31, 2048;
mov.f32 %f74, 0f00000000;
mov.u32 %r110, 0;
for.size.1 i in range(0, 14, 1):
not.pred %p(30+i*1), %p(1+i*1);
mov.f32 %f75, %f74;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r9, %r110, 392;
add.s32 %r32, %r9, %r2;
mul.wide.s32 %rd10, %r32, 4;
add.s64 %rd5, %rd1, %rd10;
@%p30 bra $L__BB0_3;
ld.global.nc.f32 %f11, [%rd5];
st.shared.f32 [%r4], %f11;
$L__BB0_3:
@%p31 bra $L__BB0_5;
ld.global.nc.f32 %f12, [%rd5+4];
st.shared.f32 [%r4+4], %f12;
$L__BB0_5:
@%p32 bra $L__BB0_7;
ld.global.nc.f32 %f13, [%rd5+8];
st.shared.f32 [%r4+8], %f13;
$L__BB0_7:
@%p33 bra $L__BB0_9;
ld.global.nc.f32 %f14, [%rd5+12];
st.shared.f32 [%r4+12], %f14;
$L__BB0_9:
@%p34 bra $L__BB0_11;
ld.global.nc.f32 %f15, [%rd5+16];
st.shared.f32 [%r4+16], %f15;
$L__BB0_11:
@%p35 bra $L__BB0_13;
ld.global.nc.f32 %f16, [%rd5+20];
st.shared.f32 [%r4+20], %f16;
$L__BB0_13:
@%p36 bra $L__BB0_15;
ld.global.nc.f32 %f17, [%rd5+24];
st.shared.f32 [%r4+24], %f17;
$L__BB0_15:
@%p37 bra $L__BB0_17;
ld.global.nc.f32 %f18, [%rd5+28];
st.shared.f32 [%r4+28], %f18;
$L__BB0_17:
@%p38 bra $L__BB0_19;
ld.global.nc.f32 %f19, [%rd5+32];
st.shared.f32 [%r4+32], %f19;
$L__BB0_19:
@%p39 bra $L__BB0_21;
ld.global.nc.f32 %f20, [%rd5+36];
st.shared.f32 [%r4+36], %f20;
$L__BB0_21:
@%p40 bra $L__BB0_23;
ld.global.nc.f32 %f21, [%rd5+40];
st.shared.f32 [%r4+40], %f21;
$L__BB0_23:
@%p41 bra $L__BB0_25;
ld.global.nc.f32 %f22, [%rd5+44];
st.shared.f32 [%r4+44], %f22;
$L__BB0_25:
@%p42 bra $L__BB0_27;
ld.global.nc.f32 %f23, [%rd5+48];
st.shared.f32 [%r4+48], %f23;
$L__BB0_27:
@%p43 bra $L__BB0_29;
ld.global.nc.f32 %f24, [%rd5+52];
st.shared.f32 [%r4+52], %f24;
$L__BB0_29:
mov.u32 %r114, 2048;
mov.u32 %r111, %r7;
mov.u32 %r112, %r5;
mov.u32 %r113, %r6;
$L__BB0_30:
for.size.13 i in range(0, 2, 1):
mul.hi.s32 %r(34+i*17), %r(112+i*-62), 1402438301;
shr.u32 %r(35+i*17), %r(34+i*17), 31;
shr.s32 %r(36+i*17), %r(34+i*17), 7;
add.s32 %r(37+i*17), %r(36+i*17), %r(35+i*17);
mul.lo.s32 %r(38+i*17), %r(37+i*17), 392;
sub.s32 %r(39+i*17), %r(112+i*-62), %r(38+i*17);
add.s32 %r(40+i*17), %r(39+i*17), %r9;
mad.lo.s32 %r(41+i*17), %r(37+i*17), 784, %r(40+i*17);
mul.wide.s32 %rd(11+i*2), %r(41+i*17), 4;
add.s64 %rd(12+i*2), %rd2, %rd(11+i*2);
ld.global.nc.v4.u32 {%r(42+i*17), %r(43+i*17), %r(44+i*17), %r(45+i*17)}, [%rd(12+i*2)+0];
st.shared.v4.u32 [%r111+(-2048+i*1024)], {%r(42+i*17), %r(43+i*17), %r(44+i*17), %r(45+i*17)};
add.s32 %r(50+i*17), %r113, (-512+i*256);
for.size.12 i in range(0, 2, 1):
mul.hi.s32 %r(68+i*16), %r(67+i*46), 1402438301;
shr.u32 %r(69+i*16), %r(68+i*16), 31;
shr.s32 %r(70+i*16), %r(68+i*16), 7;
add.s32 %r(71+i*16), %r(70+i*16), %r(69+i*16);
mul.lo.s32 %r(72+i*16), %r(71+i*16), 392;
sub.s32 %r(73+i*16), %r(67+i*46), %r(72+i*16);
add.s32 %r(74+i*16), %r(73+i*16), %r9;
mad.lo.s32 %r(75+i*16), %r(71+i*16), 784, %r(74+i*16);
mul.wide.s32 %rd(15+i*2), %r(75+i*16), 4;
add.s64 %rd(16+i*2), %rd2, %rd(15+i*2);
ld.global.nc.v4.u32 {%r(76+i*16), %r(77+i*16), %r(78+i*16), %r(79+i*16)}, [%rd(16+i*2)+0];
st.shared.v4.u32 [%r111+(0+i*1024)], {%r(76+i*16), %r(77+i*16), %r(78+i*16), %r(79+i*16)};
for.size.1 i in range(0, 2, 1):
add.s32 %r(111+i*3), %r(111+i*3), 4096;
setp.ne.s32 %p44, %r114, 202752;
for.size.1 i in range(0, 2, 1):
add.s32 %r(17+i*95), %r113, (1024+i*-768);
mov.u32 %r113, %r17;
@%p44 bra $L__BB0_30;
bar.sync 0;
mov.u32 %r100, 0;
mov.u32 %r115, %r100;
$L__BB0_32:
mul.lo.s32 %r20, %r115, 56;
add.s32 %r21, %r20, %r3;
mov.u32 %r116, %r100;
$L__BB0_33:
for.size.3 i in range(0, 2, 1):
add.s32 %r(102+i*4), %r(116+i*-95), %r(20+i*96);
shl.b32 %r(103+i*4), %r(102+i*4), 2;
add.s32 %r(105+i*4), %r(28+i*2), %r(103+i*4);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(25+i*1), [%r(109+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(27+i*2), %f26, %f(25+i*3), %f(74+i*1);
ld.shared.f32 %f(28+i*2), [%r109+(100352+i*-100348)];
for.size.5 i in range(0, 6, 1):
ld.shared.f32 %f(31+i*5), [%r105+(4+i*4)];
fma.rn.f32 %f(32+i*5), %f(31+i*5), %f(30+i*5), %f(27+i*5);
ld.shared.f32 %f(33+i*5), [%r109+(100356+i*4)];
fma.rn.f32 %f(34+i*5), %f(31+i*5), %f(33+i*5), %f(29+i*5);
ld.shared.f32 %f(35+i*5), [%r109+(8+i*4)];
ld.shared.f32 %f61, [%r105+28];
fma.rn.f32 %f74, %f61, %f60, %f57;
ld.shared.f32 %f62, [%r109+100380];
fma.rn.f32 %f75, %f61, %f62, %f59;
add.s32 %r116, %r116, 8;
setp.ne.s32 %p45, %r116, 56;
@%p45 bra $L__BB0_33;
add.s32 %r115, %r115, 1;
setp.lt.u32 %p46, %r115, 7;
@%p46 bra $L__BB0_32;
add.s32 %r110, %r110, 1;
setp.lt.u32 %p47, %r110, 2;
@%p47 bra $L__BB0_1;
mul.wide.s32 %rd19, %r1, 4;
add.s64 %rd20, %rd4, %rd19;
ld.global.nc.f32 %f63, [%rd20];
add.f32 %f64, %f74, %f63;
mov.f32 %f65, 0f00000000;
max.f32 %f66, %f64, %f65;
add.s64 %rd21, %rd3, %rd19;
st.global.f32 [%rd21], %f66;
ld.global.nc.f32 %f67, [%rd20+256];
add.f32 %f68, %f75, %f67;
max.f32 %f69, %f68, %f65;
st.global.f32 [%rd21+256], %f69;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[392];
__shared__ float placeholder_shared[50176];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 2; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_inner_s = 0; ax0_ax1_fused_inner_s < 14; ++ax0_ax1_fused_inner_s) {
if ((int)threadIdx.x * 14 + ax0_ax1_fused_inner_s < 392) {
if ((int)threadIdx.x < 28) {
int index0 = (int)threadIdx.x * 14 + ax0_ax1_fused_inner_s;
int index1 = k_outer_outer * 392 + (int)threadIdx.x * 14 + ax0_ax1_fused_inner_s;
placeholder_d_shared[index0] = placeholder[index1];
}
}
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 196; ++ax0_ax1_fused_outer_outer) {
((float4*)(placeholder_shared + ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 4))[0] = ((float4*)(placeholder1 + (ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 4) / 392 * 784 + k_outer_outer * 392 + (ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 4) % 392))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 7; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 56; ++k_inner) {
int index2 = k_outer_inner * 56 + k_inner;
int index3 = (int)threadIdx.x * 392 + k_outer_inner * 56 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 25088]);
}
}
}
int index4 = (int)threadIdx.x;
T_relu[index4] = max(T_dense[0] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[1] + placeholder2[index4 + 64], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpk8ekgztp | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<176>;
.reg .b32 %r<109>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[112];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[448];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd7, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd5;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd7;
mov.u32 %r1, %ctaid.x;
mov.u32 %r2, %tid.x;
shl.b32 %r17, %r2, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r18, %r17;
mov.u32 %r19, _ZZ24default_function_kernel0E18placeholder_shared;
mad.lo.s32 %r4, %r2, 112, %r19;
add.s32 %r5, %r2, 14;
add.s32 %r6, %r19, %r17;
mov.f32 %f174, 0f00000000;
mov.u32 %r105, 0;
mov.f32 %f175, %f174;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r21, %r105, 28;
add.s32 %r22, %r21, %r2;
mul.wide.s32 %rd8, %r22, 4;
add.s64 %rd9, %rd1, %rd8;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(7+i*1), [%rd9+(0+i*8)];
st.shared.f32 [%r3+(0+i*8)], %f(7+i*1);
ld.global.nc.f32 %f11, [%rd9+32];
mov.u32 %r108, 32;
for.size.2 i in range(0, 9, 1):
st.shared.f32 [%r3+(32+i*8)], %f(11+i*1);
ld.global.nc.f32 %f(12+i*1), [%rd9+(40+i*8)];
st.shared.f32 [%r3+104], %f20;
mad.lo.s32 %r8, %r1, 3136, %r21;
mov.u32 %r106, %r2;
mov.u32 %r107, %r5;
$L__BB0_2:
mul.hi.s32 %r24, %r106, -1840700269;
add.s32 %r25, %r24, %r106;
shr.u32 %r26, %r25, 31;
shr.s32 %r27, %r25, 4;
add.s32 %r28, %r27, %r26;
mul.lo.s32 %r29, %r28, 28;
sub.s32 %r30, %r106, %r29;
add.s32 %r31, %r8, %r30;
mad.lo.s32 %r32, %r28, 784, %r31;
mul.wide.s32 %rd10, %r32, 4;
add.s64 %rd11, %rd2, %rd10;
ld.global.nc.f32 %f21, [%rd11];
add.s32 %r33, %r6, %r108;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r33+(-32+i*8)], %f(21+i*1);
add.s32 %r(34+i*10), %r106, (2+i*2);
mul.hi.s32 %r(35+i*10), %r(34+i*10), -1840700269;
add.s32 %r(36+i*10), %r(35+i*10), %r(34+i*10);
shr.u32 %r(37+i*10), %r(36+i*10), 31;
shr.s32 %r(38+i*10), %r(36+i*10), 4;
add.s32 %r(39+i*10), %r(38+i*10), %r(37+i*10);
mul.lo.s32 %r(40+i*10), %r(39+i*10), 28;
sub.s32 %r(41+i*10), %r(34+i*10), %r(40+i*10);
add.s32 %r(42+i*10), %r8, %r(41+i*10);
mad.lo.s32 %r(43+i*10), %r(39+i*10), 784, %r(42+i*10);
mul.wide.s32 %rd(12+i*2), %r(43+i*10), 4;
add.s64 %rd(13+i*2), %rd2, %rd(12+i*2);
ld.global.nc.f32 %f(22+i*1), [%rd(13+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r33+(0+i*8)], %f(25+i*1);
add.s32 %r(74+i*10), %r107, (-4+i*2);
mul.hi.s32 %r(75+i*10), %r(74+i*10), -1840700269;
add.s32 %r(76+i*10), %r(75+i*10), %r(74+i*10);
shr.u32 %r(77+i*10), %r(76+i*10), 31;
shr.s32 %r(78+i*10), %r(76+i*10), 4;
add.s32 %r(79+i*10), %r(78+i*10), %r(77+i*10);
mul.lo.s32 %r(80+i*10), %r(79+i*10), 28;
sub.s32 %r(81+i*10), %r(74+i*10), %r(80+i*10);
add.s32 %r(82+i*10), %r8, %r(81+i*10);
mad.lo.s32 %r(83+i*10), %r(79+i*10), 784, %r(82+i*10);
mul.wide.s32 %rd(20+i*2), %r(83+i*10), 4;
add.s64 %rd(21+i*2), %rd2, %rd(20+i*2);
ld.global.nc.f32 %f(26+i*1), [%rd(21+i*2)+0];
st.shared.f32 [%r33+16], %f27;
mul.hi.s32 %r94, %r107, -1840700269;
add.s32 %r95, %r94, %r107;
shr.u32 %r96, %r95, 31;
shr.s32 %r97, %r95, 4;
add.s32 %r98, %r97, %r96;
mul.lo.s32 %r99, %r98, 28;
sub.s32 %r100, %r107, %r99;
add.s32 %r101, %r8, %r100;
mad.lo.s32 %r102, %r98, 784, %r101;
mul.wide.s32 %rd24, %r102, 4;
add.s64 %rd25, %rd2, %rd24;
ld.global.nc.f32 %f28, [%rd25];
st.shared.f32 [%r33+24], %f28;
for.size.1 i in range(0, 2, 1):
add.s32 %r(107+i*-1), %r(107+i*-1), 16;
add.s32 %r108, %r108, 64;
setp.ne.s32 %p1, %r108, 480;
@%p1 bra $L__BB0_2;
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(29+i*5), [%r4+(0+i*4)];
ld.shared.f32 %f(30+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(31+i*5), %f(30+i*5), %f(29+i*5), %f(174+i*-143);
ld.shared.f32 %f(32+i*5), [%r4+(224+i*4)];
fma.rn.f32 %f(33+i*5), %f(30+i*5), %f(32+i*5), %f(175+i*-142);
for.size.5 i in range(0, 25, 1):
ld.shared.f32 %f(39+i*5), [%r4+(8+i*4)];
ld.shared.f32 %f(40+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(41+i*5), %f(40+i*5), %f(39+i*5), %f(36+i*5);
ld.shared.f32 %f(42+i*5), [%r4+(232+i*4)];
fma.rn.f32 %f(43+i*5), %f(40+i*5), %f(42+i*5), %f(38+i*5);
ld.shared.f32 %f164, [%r4+108];
ld.shared.f32 %f165, [_ZZ24default_function_kernel0E20placeholder_d_shared+108];
fma.rn.f32 %f174, %f165, %f164, %f161;
ld.shared.f32 %f166, [%r4+332];
fma.rn.f32 %f175, %f165, %f166, %f163;
add.s32 %r105, %r105, 1;
setp.lt.u32 %p2, %r105, 28;
@%p2 bra $L__BB0_1;
shl.b32 %r103, %r1, 2;
add.s32 %r104, %r103, %r2;
mul.wide.s32 %rd26, %r104, 4;
add.s64 %rd27, %rd3, %rd26;
ld.global.nc.f32 %f167, [%rd27];
add.f32 %f168, %f174, %f167;
mov.f32 %f169, 0f00000000;
max.f32 %f170, %f168, %f169;
cvta.to.global.u64 %rd28, %rd4;
add.s64 %rd29, %rd28, %rd26;
st.global.f32 [%rd29], %f170;
ld.global.nc.f32 %f171, [%rd27+8];
add.f32 %f172, %f175, %f171;
max.f32 %f173, %f172, %f169;
st.global.f32 [%rd29+8], %f173;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[28];
__shared__ float placeholder_shared[112];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 28; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 14; ++ax0_ax1_fused_outer_outer) {
int index0 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index1 = k_outer_outer * 28 + ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 56; ++ax0_ax1_fused_outer_outer1) {
int index2 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 3136 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) / 28 * 784 + k_outer_outer * 28 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) % 28;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_inner = 0; k_inner < 28; ++k_inner) {
int index4 = k_inner;
int index5 = (int)threadIdx.x * 28 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 56]);
}
}
int index6 = (int)blockIdx.x * 4 + (int)threadIdx.x;
T_relu[index6] = max(T_dense[0] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 2] = max(T_dense[1] + placeholder2[index6 + 2], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpn7ligeau | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<13>;
.reg .f32 %f<150>;
.reg .b32 %r<129>;
.reg .b64 %rd<34>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[25088];
ld.param.u64 %rd8, [default_function_kernel0_param_0];
ld.param.u64 %rd9, [default_function_kernel0_param_1];
ld.param.u64 %rd10, [default_function_kernel0_param_2];
ld.param.u64 %rd11, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd8;
cvta.to.global.u64 %rd1, %rd9;
cvta.to.global.u64 %rd2, %rd10;
cvta.to.global.u64 %rd3, %rd11;
mov.u32 %r1, %ctaid.x;
mul.lo.s32 %r2, %r1, 25088;
mov.u32 %r3, %tid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(4+i*1), %r3, (196+i*-140);
or.b32 %r6, %r5, 7;
shl.b32 %r24, %r5, 2;
mov.u32 %r25, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r26, %r25, %r24;
add.s32 %r7, %r26, 16;
add.s64 %rd4, %rd12, 16;
mov.f32 %f134, 0f00000000;
mov.u32 %r123, 0;
for.size.1 i in range(0, 7, 1):
mov.f32 %f(135+i*1), %f134;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r10, %r123, 196;
add.s32 %r28, %r5, %r10;
mul.wide.s32 %rd13, %r28, 4;
add.s64 %rd33, %rd4, %rd13;
mov.u32 %r126, 7;
mov.u32 %r124, %r7;
mov.u32 %r125, %r6;
$L__BB0_2:
add.s32 %r29, %r125, -7;
setp.gt.s32 %p1, %r29, 195;
@%p1 bra $L__BB0_4;
ld.global.nc.f32 %f33, [%rd33+-16];
st.shared.f32 [%r124+-16], %f33;
$L__BB0_4:
add.s32 %r30, %r125, -6;
setp.gt.s32 %p2, %r30, 195;
@%p2 bra $L__BB0_6;
ld.global.nc.f32 %f34, [%rd33+-12];
st.shared.f32 [%r124+-12], %f34;
$L__BB0_6:
add.s32 %r31, %r125, -5;
setp.gt.s32 %p3, %r31, 195;
@%p3 bra $L__BB0_8;
ld.global.nc.f32 %f35, [%rd33+-8];
st.shared.f32 [%r124+-8], %f35;
$L__BB0_8:
add.s32 %r32, %r125, -4;
setp.gt.s32 %p4, %r32, 195;
@%p4 bra $L__BB0_10;
ld.global.nc.f32 %f36, [%rd33+-4];
st.shared.f32 [%r124+-4], %f36;
$L__BB0_10:
add.s32 %r33, %r125, -3;
setp.gt.s32 %p5, %r33, 195;
@%p5 bra $L__BB0_12;
ld.global.nc.f32 %f37, [%rd33];
st.shared.f32 [%r124], %f37;
$L__BB0_12:
add.s32 %r34, %r125, -2;
setp.gt.s32 %p6, %r34, 195;
@%p6 bra $L__BB0_14;
ld.global.nc.f32 %f38, [%rd33+4];
st.shared.f32 [%r124+4], %f38;
$L__BB0_14:
add.s32 %r35, %r125, -1;
setp.gt.s32 %p7, %r35, 195;
@%p7 bra $L__BB0_16;
ld.global.nc.f32 %f39, [%rd33+8];
st.shared.f32 [%r124+8], %f39;
$L__BB0_16:
setp.gt.s32 %p8, %r125, 195;
@%p8 bra $L__BB0_18;
ld.global.nc.f32 %f40, [%rd33+12];
st.shared.f32 [%r124+12], %f40;
$L__BB0_18:
for.size.1 i in range(0, 2, 1):
add.s32 %r(126+i*-1), %r(126+i*-1), 8;
add.s32 %r124, %r124, 32;
add.s64 %rd33, %rd33, 32;
setp.ne.s32 %p9, %r126, 63;
@%p9 bra $L__BB0_2;
add.s32 %r17, %r10, %r2;
mov.u32 %r127, 0;
$L__BB0_20:
shl.b32 %r37, %r127, 2;
add.s32 %r38, %r37, %r3;
mul.hi.s32 %r39, %r38, 1402438301;
shr.u32 %r40, %r39, 31;
shr.s32 %r41, %r39, 6;
add.s32 %r42, %r41, %r40;
mul.lo.s32 %r43, %r42, 196;
sub.s32 %r44, %r38, %r43;
add.s32 %r45, %r17, %r44;
mad.lo.s32 %r46, %r42, 784, %r45;
mul.wide.s32 %rd14, %r46, 4;
add.s64 %rd15, %rd1, %rd14;
ld.global.nc.f32 %f41, [%rd15];
shl.b32 %r47, %r38, 2;
mov.u32 %r48, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r49, %r48, %r47;
for.size.13 i in range(0, 7, 1):
st.shared.f32 [%r49+(0+i*16)], %f(41+i*1);
add.s32 %r(50+i*9), %r38, (4+i*4);
mul.hi.s32 %r(51+i*9), %r(50+i*9), 1402438301;
shr.u32 %r(52+i*9), %r(51+i*9), 31;
shr.s32 %r(53+i*9), %r(51+i*9), 6;
add.s32 %r(54+i*9), %r(53+i*9), %r(52+i*9);
mul.lo.s32 %r(55+i*9), %r(54+i*9), 196;
sub.s32 %r(56+i*9), %r(50+i*9), %r(55+i*9);
add.s32 %r(57+i*9), %r17, %r(56+i*9);
mad.lo.s32 %r(58+i*9), %r(54+i*9), 784, %r(57+i*9);
mul.wide.s32 %rd(16+i*2), %r(58+i*9), 4;
add.s64 %rd(17+i*2), %rd1, %rd(16+i*2);
ld.global.nc.f32 %f(42+i*1), [%rd(17+i*2)+0];
st.shared.f32 [%r49+112], %f48;
add.s32 %r127, %r127, 8;
setp.ne.s32 %p10, %r127, 1568;
@%p10 bra $L__BB0_20;
bar.sync 0;
mov.u32 %r128, 0;
$L__BB0_22:
add.s32 %r114, %r128, %r4;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(115+i*3), %r(114+i*14), 2;
add.s32 %r(117+i*3), %r(48+i*-23), %r(115+i*3);
ld.shared.f32 %f(49+i*1), [%r(117+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(51+i*2), %f50, %f(49+i*3), %f(134+i*1);
ld.shared.f32 %f(52+i*2), [%r117+(3136+i*3136)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(55+i*2), %f50, %f(54+i*2), %f(136+i*1);
ld.shared.f32 %f(56+i*2), [%r117+(9408+i*3136)];
for.size.17 i in range(0, 2, 1):
fma.rn.f32 %f(65+i*17), %f(50+i*17), %f(64+i*17), %f(141+i*-76);
ld.shared.f32 %f(66+i*17), [%r117+(4+i*4)];
ld.shared.f32 %f(67+i*17), [%r120+(4+i*4)];
fma.rn.f32 %f(68+i*17), %f(67+i*17), %f(66+i*17), %f(51+i*17);
ld.shared.f32 %f(69+i*17), [%r117+(3140+i*4)];
fma.rn.f32 %f(70+i*17), %f(67+i*17), %f(69+i*17), %f(53+i*17);
ld.shared.f32 %f(71+i*17), [%r117+(6276+i*4)];
fma.rn.f32 %f(72+i*17), %f(67+i*17), %f(71+i*17), %f(55+i*17);
ld.shared.f32 %f(73+i*17), [%r117+(9412+i*4)];
fma.rn.f32 %f(74+i*17), %f(67+i*17), %f(73+i*17), %f(57+i*17);
ld.shared.f32 %f(75+i*17), [%r117+(12548+i*4)];
fma.rn.f32 %f(76+i*17), %f(67+i*17), %f(75+i*17), %f(59+i*17);
ld.shared.f32 %f(77+i*17), [%r117+(15684+i*4)];
fma.rn.f32 %f(78+i*17), %f(67+i*17), %f(77+i*17), %f(61+i*17);
ld.shared.f32 %f(79+i*17), [%r117+(18820+i*4)];
fma.rn.f32 %f(80+i*17), %f(67+i*17), %f(79+i*17), %f(63+i*17);
ld.shared.f32 %f(81+i*17), [%r117+(21956+i*4)];
fma.rn.f32 %f99, %f84, %f98, %f82;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(100+i*1), [%r(117+i*3)+12];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(134+i*1), %f101, %f(100+i*2), %f(85+i*2);
ld.shared.f32 %f(102+i*1), [%r117+(3148+i*3136)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(136+i*1), %f101, %f(103+i*1), %f(89+i*2);
ld.shared.f32 %f(104+i*1), [%r117+(9420+i*3136)];
fma.rn.f32 %f141, %f101, %f108, %f99;
add.s32 %r128, %r128, 4;
setp.ne.s32 %p11, %r128, 196;
@%p11 bra $L__BB0_22;
add.s32 %r123, %r123, 1;
setp.lt.u32 %p12, %r123, 4;
@%p12 bra $L__BB0_1;
shl.b32 %r121, %r1, 5;
add.s32 %r122, %r121, %r3;
mul.wide.s32 %rd30, %r122, 4;
add.s64 %rd31, %rd3, %rd30;
ld.global.nc.f32 %f109, [%rd31];
add.f32 %f110, %f134, %f109;
mov.f32 %f111, 0f00000000;
max.f32 %f112, %f110, %f111;
add.s64 %rd32, %rd2, %rd30;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd32+(0+i*16)], %f(112+i*3);
ld.global.nc.f32 %f(113+i*3), [%rd31+(16+i*16)];
add.f32 %f(114+i*3), %f(135+i*1), %f(113+i*3);
max.f32 %f(115+i*3), %f(114+i*3), %f111;
st.global.f32 [%rd32+112], %f133;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[8];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[6272];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_inner_s = 0; ax0_ax1_fused_inner_s < 56; ++ax0_ax1_fused_inner_s) {
if ((int)threadIdx.x * 56 + ax0_ax1_fused_inner_s < 196) {
int index0 = (int)threadIdx.x * 56 + ax0_ax1_fused_inner_s;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 56 + ax0_ax1_fused_inner_s;
placeholder_d_shared[index0] = placeholder[index1];
}
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1568; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 25088 + (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) % 196;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 196 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1568]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2352]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3136]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3920]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4704]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 5488]);
}
}
int index6 = (int)blockIdx.x * 32 + (int)threadIdx.x;
T_relu[index6] = max(T_dense[0] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 4] = max(T_dense[1] + placeholder2[index6 + 4], 0.000000e+00f);
T_relu[index6 + 8] = max(T_dense[2] + placeholder2[index6 + 8], 0.000000e+00f);
T_relu[index6 + 12] = max(T_dense[3] + placeholder2[index6 + 12], 0.000000e+00f);
T_relu[index6 + 16] = max(T_dense[4] + placeholder2[index6 + 16], 0.000000e+00f);
T_relu[index6 + 20] = max(T_dense[5] + placeholder2[index6 + 20], 0.000000e+00f);
T_relu[index6 + 24] = max(T_dense[6] + placeholder2[index6 + 24], 0.000000e+00f);
T_relu[index6 + 28] = max(T_dense[7] + placeholder2[index6 + 28], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp44djmah_ | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<421>;
.reg .b32 %r<108>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[392];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[50176];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r19, %r18;
mul.lo.s32 %r3, %r1, 98;
add.s32 %r4, %r1, 28;
mov.u32 %r20, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r20, %r18;
mov.u32 %r103, 0;
mov.f32 %f357, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(358+i*1), %f357;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r7, %r103, 98;
add.s32 %r21, %r7, %r1;
mul.wide.s32 %rd10, %r21, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.2 i in range(0, 24, 1):
ld.global.nc.f32 %f(129+i*1), [%rd5+(0+i*16)];
st.shared.f32 [%r2+(0+i*16)], %f(129+i*1);
setp.gt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f153, [%rd5+384];
st.shared.f32 [%r2+384], %f153;
$L__BB0_3:
mov.u32 %r106, 64;
mov.u32 %r104, %r1;
mov.u32 %r105, %r4;
$L__BB0_4:
mul.hi.s32 %r23, %r104, 1402438301;
shr.u32 %r24, %r23, 31;
shr.s32 %r25, %r23, 5;
add.s32 %r26, %r25, %r24;
mul.lo.s32 %r27, %r26, 98;
sub.s32 %r28, %r104, %r27;
add.s32 %r29, %r28, %r7;
mad.lo.s32 %r30, %r26, 784, %r29;
mul.wide.s32 %rd11, %r30, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f154, [%rd12];
add.s32 %r31, %r5, %r106;
for.size.13 i in range(0, 4, 1):
st.shared.f32 [%r31+(-64+i*16)], %f(154+i*1);
add.s32 %r(32+i*9), %r104, (4+i*4);
mul.hi.s32 %r(33+i*9), %r(32+i*9), 1402438301;
shr.u32 %r(34+i*9), %r(33+i*9), 31;
shr.s32 %r(35+i*9), %r(33+i*9), 5;
add.s32 %r(36+i*9), %r(35+i*9), %r(34+i*9);
mul.lo.s32 %r(37+i*9), %r(36+i*9), 98;
sub.s32 %r(38+i*9), %r(32+i*9), %r(37+i*9);
add.s32 %r(39+i*9), %r(38+i*9), %r7;
mad.lo.s32 %r(40+i*9), %r(36+i*9), 784, %r(39+i*9);
mul.wide.s32 %rd(13+i*2), %r(40+i*9), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(155+i*1), [%rd(14+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.f32 [%r31+(0+i*16)], %f(158+i*1);
add.s32 %r(68+i*9), %r105, (-8+i*4);
mul.hi.s32 %r(69+i*9), %r(68+i*9), 1402438301;
shr.u32 %r(70+i*9), %r(69+i*9), 31;
shr.s32 %r(71+i*9), %r(69+i*9), 5;
add.s32 %r(72+i*9), %r(71+i*9), %r(70+i*9);
mul.lo.s32 %r(73+i*9), %r(72+i*9), 98;
sub.s32 %r(74+i*9), %r(68+i*9), %r(73+i*9);
add.s32 %r(75+i*9), %r(74+i*9), %r7;
mad.lo.s32 %r(76+i*9), %r(72+i*9), 784, %r(75+i*9);
mul.wide.s32 %rd(21+i*2), %r(76+i*9), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(159+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r31+32], %f160;
mul.hi.s32 %r86, %r105, 1402438301;
shr.u32 %r87, %r86, 31;
shr.s32 %r88, %r86, 5;
add.s32 %r89, %r88, %r87;
mul.lo.s32 %r90, %r89, 98;
sub.s32 %r91, %r105, %r90;
add.s32 %r92, %r91, %r7;
mad.lo.s32 %r93, %r89, 784, %r92;
mul.wide.s32 %rd25, %r93, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f161, [%rd26];
st.shared.f32 [%r31+48], %f161;
add.s32 %r106, %r106, 128;
setp.ne.s32 %p2, %r106, 50240;
for.size.1 i in range(0, 2, 1):
add.s32 %r(12+i*92), %r105, (32+i*-28);
mov.u32 %r105, %r12;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r107, 0;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(95+i*1), %r107, (1+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(98+i*1), %r(19+i*76), %r(96+i*-93);
shl.b32 %r100, %r99, 2;
add.s32 %r102, %r20, %r100;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(162+i*1), [%r(102+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(164+i*2), %f163, %f(162+i*3), %f(388+i*-1);
ld.shared.f32 %f(165+i*2), [%r102+(1568+i*1568)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(168+i*2), %f163, %f(167+i*2), %f(386+i*-1);
ld.shared.f32 %f(169+i*2), [%r102+(4704+i*1568)];
fma.rn.f32 %f226, %f163, %f225, %f357;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(227+i*1), [%r(102+i*-4)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(388+i*-1), %f228, %f(227+i*2), %f(164+i*2);
ld.shared.f32 %f(229+i*1), [%r102+(1572+i*1568)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(386+i*-1), %f228, %f(230+i*1), %f(168+i*2);
ld.shared.f32 %f(231+i*1), [%r102+(4708+i*1568)];
fma.rn.f32 %f357, %f228, %f259, %f226;
add.s32 %r107, %r107, 1;
setp.ne.s32 %p3, %r107, 49;
@%p3 bra $L__BB0_6;
add.s32 %r103, %r103, 1;
setp.lt.u32 %p4, %r103, 8;
@%p4 bra $L__BB0_1;
mul.wide.s32 %rd27, %r1, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f260, [%rd28];
add.f32 %f261, %f388, %f260;
mov.f32 %f262, 0f00000000;
max.f32 %f263, %f261, %f262;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 31, 1):
st.global.f32 [%rd29+(0+i*16)], %f(263+i*3);
ld.global.nc.f32 %f(264+i*3), [%rd28+(16+i*16)];
add.f32 %f(265+i*3), %f(387+i*-1), %f(264+i*3);
max.f32 %f(266+i*3), %f(265+i*3), %f262;
st.global.f32 [%rd29+496], %f356;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[98];
__shared__ float placeholder_shared[12544];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 8; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 98 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 12] = placeholder[index1 + 12];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 20] = placeholder[index1 + 20];
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
placeholder_d_shared[index0 + 28] = placeholder[index1 + 28];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 36] = placeholder[index1 + 36];
placeholder_d_shared[index0 + 40] = placeholder[index1 + 40];
placeholder_d_shared[index0 + 44] = placeholder[index1 + 44];
placeholder_d_shared[index0 + 48] = placeholder[index1 + 48];
placeholder_d_shared[index0 + 52] = placeholder[index1 + 52];
placeholder_d_shared[index0 + 56] = placeholder[index1 + 56];
placeholder_d_shared[index0 + 60] = placeholder[index1 + 60];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 68] = placeholder[index1 + 68];
placeholder_d_shared[index0 + 72] = placeholder[index1 + 72];
placeholder_d_shared[index0 + 76] = placeholder[index1 + 76];
placeholder_d_shared[index0 + 80] = placeholder[index1 + 80];
placeholder_d_shared[index0 + 84] = placeholder[index1 + 84];
placeholder_d_shared[index0 + 88] = placeholder[index1 + 88];
placeholder_d_shared[index0 + 92] = placeholder[index1 + 92];
if ((int)threadIdx.x < 2) {
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 3136; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) / 98 * 784 + k_outer_outer * 98 + (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) % 98;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 49; ++k_outer_inner) {
int index4 = k_outer_inner * 2;
int index5 = (int)threadIdx.x * 98 + k_outer_inner * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 392]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1176]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1568]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1960]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2352]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2744]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3136]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3528]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3920]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4312]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4704]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 5096]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 5488]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 5880]);
T_dense[16] = T_dense[16] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 6272]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 6664]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7056]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7448]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7840]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8232]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8624]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9016]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9408]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9800]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10192]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10584]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10976]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 11368]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 11760]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 12152]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 393]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 785]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1177]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1569]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1961]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 2353]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 2745]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 3137]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 3529]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 3921]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 4313]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 4705]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 5097]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 5489]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 5881]);
T_dense[16] = T_dense[16] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 6273]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 6665]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7057]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7449]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7841]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 8233]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 8625]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 9017]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 9409]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 9801]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 10193]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 10585]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 10977]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 11369]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 11761]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 12153]);
}
}
for (int vthread_s = 0; vthread_s < 32; ++vthread_s) {
int index6 = vthread_s * 4 + (int)threadIdx.x;
int index7 = vthread_s;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpz9o9w_yi | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<292>;
.reg .b32 %r<14>;
.reg .b64 %rd<22>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[4];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[512];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd1, %rd11;
mov.u32 %r1, %tid.x;
shl.b32 %r4, %r1, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r9, %r4;
mov.u32 %r10, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r10, %r4;
shl.b32 %r11, %r1, 4;
add.s32 %r5, %r10, %r11;
mul.lo.s32 %r12, %r1, 784;
cvta.to.global.u64 %rd13, %rd9;
mul.wide.s32 %rd14, %r12, 4;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(21+i*-6), %rd(13+i*-1), %rd14;
add.s64 %rd20, %rd15, 200704;
mov.u32 %r13, 0;
mov.f32 %f260, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(261+i*1), %f260;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f97, [%rd21];
st.shared.f32 [%r2], %f97;
$L__BB0_3:
for.size.2 i in range(0, 32, 1):
ld.global.nc.f32 %f(98+i*1), [%rd20+(-200704+i*12544)];
st.shared.f32 [%r3+(0+i*16)], %f(98+i*1);
bar.sync 0;
ld.shared.f32 %f130, [%r5];
ld.shared.f32 %f131, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(291+i*-1), %f131, %f(130+i*2), %f(291+i*-1);
ld.shared.f32 %f(132+i*1), [%r5+(64+i*64)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(289+i*-1), %f131, %f(133+i*1), %f(289+i*-1);
ld.shared.f32 %f(134+i*1), [%r5+(192+i*64)];
for.size.3 i in range(0, 3, 1):
for.size.2 j in range(0, 8, 1):
fma.rn.f32 %f(284+i*-8+j*-1), %f131, %f(138+i*8+j*1), %f(284+i*-8+j*-1);
ld.shared.f32 %f(139+i*8+j*1), [%r5+(4+i*4+j*64)];
fma.rn.f32 %f260, %f131, %f162, %f260;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(21+i*-1), %rd(21+i*-1), 4;
add.s32 %r13, %r13, 1;
setp.ne.s32 %p2, %r13, 784;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd16, %rd8;
mul.wide.s32 %rd17, %r4, 4;
add.s64 %rd18, %rd16, %rd17;
ld.global.nc.f32 %f163, [%rd18];
add.f32 %f164, %f291, %f163;
mov.f32 %f165, 0f00000000;
max.f32 %f166, %f164, %f165;
add.s64 %rd19, %rd1, %rd17;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd19+(0+i*64)], %f(166+i*3);
ld.global.nc.f32 %f(167+i*3), [%rd18+(64+i*64)];
add.f32 %f(168+i*3), %f(290+i*-1), %f(167+i*3);
max.f32 %f(169+i*3), %f(168+i*3), %f165;
for.size.10 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd19+(448+i*4+j*-444)], %f(187+i*24+j*3);
ld.global.nc.f32 %f(188+i*24+j*3), [%rd18+(4+i*4+j*64)];
add.f32 %f(189+i*24+j*3), %f(283+i*-8+j*-1), %f(188+i*24+j*3);
max.f32 %f(190+i*24+j*3), %f(189+i*24+j*3), %f165;
for.size.4 j in range(0, 6, 1):
st.global.f32 [%rd19+(68+i*4+j*64)], %f(193+i*24+j*3);
ld.global.nc.f32 %f(194+i*24+j*3), [%rd18+(132+i*4+j*64)];
add.f32 %f(195+i*24+j*3), %f(281+i*-8+j*-1), %f(194+i*24+j*3);
max.f32 %f(196+i*24+j*3), %f(195+i*24+j*3), %f165;
st.global.f32 [%rd19+460], %f259;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[128];
T_dense[0] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_shared[index0] = placeholder1[index1];
placeholder_shared[index0 + 4] = placeholder1[index1 + 3136];
placeholder_shared[index0 + 8] = placeholder1[index1 + 6272];
placeholder_shared[index0 + 12] = placeholder1[index1 + 9408];
placeholder_shared[index0 + 16] = placeholder1[index1 + 12544];
placeholder_shared[index0 + 20] = placeholder1[index1 + 15680];
placeholder_shared[index0 + 24] = placeholder1[index1 + 18816];
placeholder_shared[index0 + 28] = placeholder1[index1 + 21952];
placeholder_shared[index0 + 32] = placeholder1[index1 + 25088];
placeholder_shared[index0 + 36] = placeholder1[index1 + 28224];
placeholder_shared[index0 + 40] = placeholder1[index1 + 31360];
placeholder_shared[index0 + 44] = placeholder1[index1 + 34496];
placeholder_shared[index0 + 48] = placeholder1[index1 + 37632];
placeholder_shared[index0 + 52] = placeholder1[index1 + 40768];
placeholder_shared[index0 + 56] = placeholder1[index1 + 43904];
placeholder_shared[index0 + 60] = placeholder1[index1 + 47040];
placeholder_shared[index0 + 64] = placeholder1[index1 + 50176];
placeholder_shared[index0 + 68] = placeholder1[index1 + 53312];
placeholder_shared[index0 + 72] = placeholder1[index1 + 56448];
placeholder_shared[index0 + 76] = placeholder1[index1 + 59584];
placeholder_shared[index0 + 80] = placeholder1[index1 + 62720];
placeholder_shared[index0 + 84] = placeholder1[index1 + 65856];
placeholder_shared[index0 + 88] = placeholder1[index1 + 68992];
placeholder_shared[index0 + 92] = placeholder1[index1 + 72128];
placeholder_shared[index0 + 96] = placeholder1[index1 + 75264];
placeholder_shared[index0 + 100] = placeholder1[index1 + 78400];
placeholder_shared[index0 + 104] = placeholder1[index1 + 81536];
placeholder_shared[index0 + 108] = placeholder1[index1 + 84672];
placeholder_shared[index0 + 112] = placeholder1[index1 + 87808];
placeholder_shared[index0 + 116] = placeholder1[index1 + 90944];
placeholder_shared[index0 + 120] = placeholder1[index1 + 94080];
placeholder_shared[index0 + 124] = placeholder1[index1 + 97216];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 4];
int index2 = (int)threadIdx.x * 4 + 16;
T_dense[4] = T_dense[4] + placeholder_d_shared[0] * placeholder_shared[index2];
int index3 = (int)threadIdx.x * 4 + 32;
T_dense[8] = T_dense[8] + placeholder_d_shared[0] * placeholder_shared[index3];
int index4 = (int)threadIdx.x * 4 + 48;
T_dense[12] = T_dense[12] + placeholder_d_shared[0] * placeholder_shared[index4];
int index5 = (int)threadIdx.x * 4 + 64;
T_dense[16] = T_dense[16] + placeholder_d_shared[0] * placeholder_shared[index5];
int index6 = (int)threadIdx.x * 4 + 80;
T_dense[20] = T_dense[20] + placeholder_d_shared[0] * placeholder_shared[index6];
int index7 = (int)threadIdx.x * 4 + 96;
T_dense[24] = T_dense[24] + placeholder_d_shared[0] * placeholder_shared[index7];
int index8 = (int)threadIdx.x * 4 + 112;
T_dense[28] = T_dense[28] + placeholder_d_shared[0] * placeholder_shared[index8];
int index9 = (int)threadIdx.x * 4 + 1;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index9];
int index10 = (int)threadIdx.x * 4 + 17;
T_dense[5] = T_dense[5] + placeholder_d_shared[0] * placeholder_shared[index10];
int index11 = (int)threadIdx.x * 4 + 33;
T_dense[9] = T_dense[9] + placeholder_d_shared[0] * placeholder_shared[index11];
int index12 = (int)threadIdx.x * 4 + 49;
T_dense[13] = T_dense[13] + placeholder_d_shared[0] * placeholder_shared[index12];
int index13 = (int)threadIdx.x * 4 + 65;
T_dense[17] = T_dense[17] + placeholder_d_shared[0] * placeholder_shared[index13];
int index14 = (int)threadIdx.x * 4 + 81;
T_dense[21] = T_dense[21] + placeholder_d_shared[0] * placeholder_shared[index14];
int index15 = (int)threadIdx.x * 4 + 97;
T_dense[25] = T_dense[25] + placeholder_d_shared[0] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 4 + 113;
T_dense[29] = T_dense[29] + placeholder_d_shared[0] * placeholder_shared[index16];
int index17 = (int)threadIdx.x * 4 + 2;
T_dense[2] = T_dense[2] + placeholder_d_shared[0] * placeholder_shared[index17];
int index18 = (int)threadIdx.x * 4 + 18;
T_dense[6] = T_dense[6] + placeholder_d_shared[0] * placeholder_shared[index18];
int index19 = (int)threadIdx.x * 4 + 34;
T_dense[10] = T_dense[10] + placeholder_d_shared[0] * placeholder_shared[index19];
int index20 = (int)threadIdx.x * 4 + 50;
T_dense[14] = T_dense[14] + placeholder_d_shared[0] * placeholder_shared[index20];
int index21 = (int)threadIdx.x * 4 + 66;
T_dense[18] = T_dense[18] + placeholder_d_shared[0] * placeholder_shared[index21];
int index22 = (int)threadIdx.x * 4 + 82;
T_dense[22] = T_dense[22] + placeholder_d_shared[0] * placeholder_shared[index22];
int index23 = (int)threadIdx.x * 4 + 98;
T_dense[26] = T_dense[26] + placeholder_d_shared[0] * placeholder_shared[index23];
int index24 = (int)threadIdx.x * 4 + 114;
T_dense[30] = T_dense[30] + placeholder_d_shared[0] * placeholder_shared[index24];
int index25 = (int)threadIdx.x * 4 + 3;
T_dense[3] = T_dense[3] + placeholder_d_shared[0] * placeholder_shared[index25];
int index26 = (int)threadIdx.x * 4 + 19;
T_dense[7] = T_dense[7] + placeholder_d_shared[0] * placeholder_shared[index26];
int index27 = (int)threadIdx.x * 4 + 35;
T_dense[11] = T_dense[11] + placeholder_d_shared[0] * placeholder_shared[index27];
int index28 = (int)threadIdx.x * 4 + 51;
T_dense[15] = T_dense[15] + placeholder_d_shared[0] * placeholder_shared[index28];
int index29 = (int)threadIdx.x * 4 + 67;
T_dense[19] = T_dense[19] + placeholder_d_shared[0] * placeholder_shared[index29];
int index30 = (int)threadIdx.x * 4 + 83;
T_dense[23] = T_dense[23] + placeholder_d_shared[0] * placeholder_shared[index30];
int index31 = (int)threadIdx.x * 4 + 99;
T_dense[27] = T_dense[27] + placeholder_d_shared[0] * placeholder_shared[index31];
int index32 = (int)threadIdx.x * 4 + 115;
T_dense[31] = T_dense[31] + placeholder_d_shared[0] * placeholder_shared[index32];
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index33 = (int)threadIdx.x * 4 + ax1_inner;
int index34 = ax1_inner;
T_relu[index33] = max(T_dense[index34] + placeholder2[index33], 0.000000e+00f);
T_relu[index33 + 16] = max(T_dense[index34 + 4] + placeholder2[index33 + 16], 0.000000e+00f);
T_relu[index33 + 32] = max(T_dense[index34 + 8] + placeholder2[index33 + 32], 0.000000e+00f);
T_relu[index33 + 48] = max(T_dense[index34 + 12] + placeholder2[index33 + 48], 0.000000e+00f);
T_relu[index33 + 64] = max(T_dense[index34 + 16] + placeholder2[index33 + 64], 0.000000e+00f);
T_relu[index33 + 80] = max(T_dense[index34 + 20] + placeholder2[index33 + 80], 0.000000e+00f);
T_relu[index33 + 96] = max(T_dense[index34 + 24] + placeholder2[index33 + 96], 0.000000e+00f);
T_relu[index33 + 112] = max(T_dense[index34 + 28] + placeholder2[index33 + 112], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpimbn8k0p | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<400>;
.reg .b32 %r<116>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[56];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[7168];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r19, %r18;
mul.lo.s32 %r3, %r1, 14;
add.s32 %r4, %r1, 28;
mov.u32 %r20, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r20, %r18;
mov.u32 %r111, 0;
mov.f32 %f336, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(337+i*1), %f336;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r7, %r111, 14;
add.s32 %r21, %r7, %r1;
mul.wide.s32 %rd10, %r21, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.2 i in range(0, 3, 1):
ld.global.nc.f32 %f(129+i*1), [%rd5+(0+i*16)];
st.shared.f32 [%r2+(0+i*16)], %f(129+i*1);
setp.gt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f132, [%rd5+48];
st.shared.f32 [%r2+48], %f132;
$L__BB0_3:
mov.u32 %r114, 64;
mov.u32 %r112, %r1;
mov.u32 %r113, %r4;
$L__BB0_4:
mul.hi.s32 %r23, %r112, -1840700269;
add.s32 %r24, %r23, %r112;
shr.u32 %r25, %r24, 31;
shr.s32 %r26, %r24, 3;
add.s32 %r27, %r26, %r25;
mul.lo.s32 %r28, %r27, 14;
sub.s32 %r29, %r112, %r28;
add.s32 %r30, %r29, %r7;
mad.lo.s32 %r31, %r27, 784, %r30;
mul.wide.s32 %rd11, %r31, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f133, [%rd12];
add.s32 %r32, %r5, %r114;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r32+(-64+i*16)], %f(133+i*1);
add.s32 %r(33+i*10), %r112, (4+i*4);
mul.hi.s32 %r(34+i*10), %r(33+i*10), -1840700269;
add.s32 %r(35+i*10), %r(34+i*10), %r(33+i*10);
shr.u32 %r(36+i*10), %r(35+i*10), 31;
shr.s32 %r(37+i*10), %r(35+i*10), 3;
add.s32 %r(38+i*10), %r(37+i*10), %r(36+i*10);
mul.lo.s32 %r(39+i*10), %r(38+i*10), 14;
sub.s32 %r(40+i*10), %r(33+i*10), %r(39+i*10);
add.s32 %r(41+i*10), %r(40+i*10), %r7;
mad.lo.s32 %r(42+i*10), %r(38+i*10), 784, %r(41+i*10);
mul.wide.s32 %rd(13+i*2), %r(42+i*10), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(134+i*1), [%rd(14+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r32+(0+i*16)], %f(137+i*1);
add.s32 %r(73+i*10), %r113, (-8+i*4);
mul.hi.s32 %r(74+i*10), %r(73+i*10), -1840700269;
add.s32 %r(75+i*10), %r(74+i*10), %r(73+i*10);
shr.u32 %r(76+i*10), %r(75+i*10), 31;
shr.s32 %r(77+i*10), %r(75+i*10), 3;
add.s32 %r(78+i*10), %r(77+i*10), %r(76+i*10);
mul.lo.s32 %r(79+i*10), %r(78+i*10), 14;
sub.s32 %r(80+i*10), %r(73+i*10), %r(79+i*10);
add.s32 %r(81+i*10), %r(80+i*10), %r7;
mad.lo.s32 %r(82+i*10), %r(78+i*10), 784, %r(81+i*10);
mul.wide.s32 %rd(21+i*2), %r(82+i*10), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(138+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r32+32], %f139;
mul.hi.s32 %r93, %r113, -1840700269;
add.s32 %r94, %r93, %r113;
shr.u32 %r95, %r94, 31;
shr.s32 %r96, %r94, 3;
add.s32 %r97, %r96, %r95;
mul.lo.s32 %r98, %r97, 14;
sub.s32 %r99, %r113, %r98;
add.s32 %r100, %r99, %r7;
mad.lo.s32 %r101, %r97, 784, %r100;
mul.wide.s32 %rd25, %r101, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f140, [%rd26];
st.shared.f32 [%r32+48], %f140;
add.s32 %r114, %r114, 128;
setp.ne.s32 %p2, %r114, 7232;
for.size.1 i in range(0, 2, 1):
add.s32 %r(12+i*100), %r113, (32+i*-28);
mov.u32 %r113, %r12;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r115, 0;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(103+i*1), %r115, (1+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(106+i*1), %r(19+i*84), %r(104+i*-101);
shl.b32 %r108, %r107, 2;
add.s32 %r110, %r20, %r108;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(141+i*1), [%r(110+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(143+i*2), %f142, %f(141+i*3), %f(367+i*-1);
ld.shared.f32 %f(144+i*2), [%r110+(224+i*224)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(147+i*2), %f142, %f(146+i*2), %f(365+i*-1);
ld.shared.f32 %f(148+i*2), [%r110+(672+i*224)];
fma.rn.f32 %f205, %f142, %f204, %f336;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(206+i*1), [%r(110+i*-4)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(367+i*-1), %f207, %f(206+i*2), %f(143+i*2);
ld.shared.f32 %f(208+i*1), [%r110+(228+i*224)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(365+i*-1), %f207, %f(209+i*1), %f(147+i*2);
ld.shared.f32 %f(210+i*1), [%r110+(676+i*224)];
fma.rn.f32 %f336, %f207, %f238, %f205;
add.s32 %r115, %r115, 1;
setp.ne.s32 %p3, %r115, 7;
@%p3 bra $L__BB0_6;
add.s32 %r111, %r111, 1;
setp.lt.u32 %p4, %r111, 56;
@%p4 bra $L__BB0_1;
mul.wide.s32 %rd27, %r1, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f239, [%rd28];
add.f32 %f240, %f367, %f239;
mov.f32 %f241, 0f00000000;
max.f32 %f242, %f240, %f241;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 31, 1):
st.global.f32 [%rd29+(0+i*16)], %f(242+i*3);
ld.global.nc.f32 %f(243+i*3), [%rd28+(16+i*16)];
add.f32 %f(244+i*3), %f(366+i*-1), %f(243+i*3);
max.f32 %f(245+i*3), %f(244+i*3), %f241;
st.global.f32 [%rd29+496], %f335;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[14];
__shared__ float placeholder_shared[1792];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 56; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 14 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
if ((int)threadIdx.x < 2) {
placeholder_d_shared[index0 + 12] = placeholder[index1 + 12];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 448; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) / 14 * 784 + k_outer_outer * 14 + (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) % 14;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 7; ++k_outer_inner) {
int index4 = k_outer_inner * 2;
int index5 = (int)threadIdx.x * 14 + k_outer_inner * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 56]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 112]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 168]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 224]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 280]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 336]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 392]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 448]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 504]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 560]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 616]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 672]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 728]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 840]);
T_dense[16] = T_dense[16] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 896]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 952]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1008]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1064]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1120]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1176]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1232]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1288]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1344]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1400]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1456]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1512]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1568]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1624]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1680]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1736]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 57]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 113]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 169]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 225]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 281]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 337]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 393]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 449]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 505]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 561]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 617]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 673]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 729]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 785]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 841]);
T_dense[16] = T_dense[16] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 897]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 953]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1009]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1065]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1121]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1177]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1233]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1289]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1345]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1401]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1457]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1513]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1569]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1625]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1681]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1737]);
}
}
for (int vthread_s = 0; vthread_s < 32; ++vthread_s) {
int index6 = vthread_s * 4 + (int)threadIdx.x;
int index7 = vthread_s;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpy6xls5vo | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<41>;
.reg .b32 %r<20>;
.reg .b64 %rd<23>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[8];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[512];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd1, %rd11;
mov.u32 %r1, %tid.x;
shl.b32 %r9, %r1, 2;
mov.u32 %r10, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r10, %r9;
mul.lo.s32 %r11, %r1, 6272;
shl.b32 %r12, %r1, 6;
mov.u32 %r13, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r13, %r12;
shl.b32 %r14, %r1, 4;
add.s32 %r4, %r13, %r14;
cvta.to.global.u64 %rd13, %rd9;
mul.wide.s32 %rd14, %r1, 4;
add.s64 %rd22, %rd13, %rd14;
mov.u32 %r5, %ctaid.x;
mad.lo.s32 %r15, %r5, 50176, %r11;
mul.wide.s32 %rd15, %r15, 4;
add.s64 %rd16, %rd12, %rd15;
add.s64 %rd21, %rd16, 12544;
mov.u32 %r19, 0;
mov.f32 %f39, 0f00000000;
mov.f32 %f40, %f39;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f7, [%rd22];
st.shared.f32 [%r2], %f7;
$L__BB0_3:
setp.gt.s32 %p2, %r1, 7;
@%p2 bra $L__BB0_5;
for.size.3 i in range(0, 8, 1):
for.size.2 j in range(0, 2, 1):
ld.global.nc.f32 %f(8+i*2+j*1), [%rd21+(-12544+i*3136+j*4)];
st.shared.f32 [%r3+(0+i*8+j*4)], %f(8+i*2+j*1);
$L__BB0_5:
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(24+i*5), [%r4+(0+i*4)];
ld.shared.f32 %f(25+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(26+i*14), %f(25+i*5), %f(24+i*5), %f(40+i*-14);
ld.shared.f32 %f(27+i*4), [%r4+(8+i*4)];
fma.rn.f32 %f(28+i*11), %f(25+i*5), %f(27+i*4), %f(39+i*-11);
for.size.1 i in range(0, 2, 1):
add.s64 %rd(22+i*-1), %rd(22+i*-1), 8;
add.s32 %r19, %r19, 1;
setp.ne.s32 %p3, %r19, 392;
@%p3 bra $L__BB0_1;
cvta.to.global.u64 %rd17, %rd8;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(16+i*1), %r(1+i*4), (1+i*5);
add.s32 %r18, %r17, %r16;
mul.wide.s32 %rd18, %r18, 4;
add.s64 %rd19, %rd17, %rd18;
ld.global.nc.f32 %f32, [%rd19];
add.f32 %f33, %f40, %f32;
mov.f32 %f34, 0f00000000;
max.f32 %f35, %f33, %f34;
add.s64 %rd20, %rd1, %rd18;
st.global.f32 [%rd20], %f35;
ld.global.nc.f32 %f36, [%rd19+4];
add.f32 %f37, %f39, %f36;
max.f32 %f38, %f37, %f34;
st.global.f32 [%rd20+4], %f38;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[2];
__shared__ float placeholder_shared[128];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 392; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
if ((int)threadIdx.x < 8) {
int index0 = (int)threadIdx.x;
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
placeholder_shared[index0 * 16] = placeholder1[index2];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index3 = (int)threadIdx.x * 16 + 1;
placeholder_shared[index3] = placeholder1[index2 + 1];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index4 = (int)threadIdx.x * 16 + 2;
placeholder_shared[index4] = placeholder1[index2 + 784];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index5 = (int)threadIdx.x * 16 + 3;
placeholder_shared[index5] = placeholder1[index2 + 785];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index6 = (int)threadIdx.x * 16 + 4;
placeholder_shared[index6] = placeholder1[index2 + 1568];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index7 = (int)threadIdx.x * 16 + 5;
placeholder_shared[index7] = placeholder1[index2 + 1569];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index8 = (int)threadIdx.x * 16 + 6;
placeholder_shared[index8] = placeholder1[index2 + 2352];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index9 = (int)threadIdx.x * 16 + 7;
placeholder_shared[index9] = placeholder1[index2 + 2353];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index10 = (int)threadIdx.x * 16 + 8;
placeholder_shared[index10] = placeholder1[index2 + 3136];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index11 = (int)threadIdx.x * 16 + 9;
placeholder_shared[index11] = placeholder1[index2 + 3137];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index12 = (int)threadIdx.x * 16 + 10;
placeholder_shared[index12] = placeholder1[index2 + 3920];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index13 = (int)threadIdx.x * 16 + 11;
placeholder_shared[index13] = placeholder1[index2 + 3921];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index14 = (int)threadIdx.x * 16 + 12;
placeholder_shared[index14] = placeholder1[index2 + 4704];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index15 = (int)threadIdx.x * 16 + 13;
placeholder_shared[index15] = placeholder1[index2 + 4705];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index16 = (int)threadIdx.x * 16 + 14;
placeholder_shared[index16] = placeholder1[index2 + 5488];
}
if ((int)threadIdx.x < 8) {
int index2 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 6272 + k_outer_outer * 2;
int index17 = (int)threadIdx.x * 16 + 15;
placeholder_shared[index17] = placeholder1[index2 + 5489];
}
__syncthreads();
int index0 = (int)threadIdx.x;
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 4];
int index18 = (int)threadIdx.x * 4 + 2;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index18];
int index19 = (int)threadIdx.x * 4 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index19];
int index20 = (int)threadIdx.x * 4 + 3;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index20];
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index21 = (int)blockIdx.x * 64 + (int)threadIdx.x * 2 + ax1_inner;
int index22 = ax1_inner;
T_relu[index21] = max(T_dense[index22] + placeholder2[index21], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpkhtsols3 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<130>;
.reg .b32 %r<36>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[4096];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd1, %rd11;
mov.u32 %r1, %tid.x;
shl.b32 %r8, %r1, 5;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r9, %r8;
shr.s32 %r10, %r1, 3;
and.b32 %r11, %r1, 7;
shl.b32 %r12, %r1, 2;
mov.u32 %r13, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r13, %r12;
shl.b32 %r14, %r1, 6;
add.s32 %r4, %r13, %r14;
mul.lo.s32 %r15, %r1, 784;
cvta.to.global.u64 %rd13, %rd9;
mul.wide.s32 %rd14, %r15, 4;
add.s64 %rd15, %rd13, %rd14;
add.s64 %rd23, %rd15, 16;
mul.lo.s32 %r16, %r10, 784;
or.b32 %r17, %r16, %r11;
mul.wide.s32 %rd16, %r17, 4;
add.s64 %rd17, %rd12, %rd16;
add.s64 %rd22, %rd17, 200704;
mov.u32 %r35, 0;
mov.f32 %f126, 0f00000000;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(127+i*1), %f126;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
for.size.1 i in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(18+i*8), %r(19+i*8), %r(20+i*8), %r(21+i*8)}, [%rd23+(-16+i*16)];
for.size.1 i in range(0, 2, 1):
st.shared.v4.u32 [%r2+(0+i*16)], {%r(18+i*8), %r(19+i*8), %r(20+i*8), %r(21+i*8)};
$L__BB0_3:
for.size.2 i in range(0, 32, 1):
ld.global.nc.f32 %f(13+i*1), [%rd22+(-200704+i*12544)];
st.shared.f32 [%r3+(0+i*128)], %f(13+i*1);
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(45+i*5), [%r4+(0+i*4)];
ld.shared.f32 %f(46+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(47+i*5), %f(46+i*5), %f(45+i*5), %f(129+i*-82);
ld.shared.f32 %f(48+i*5), [%r4+(2048+i*4)];
fma.rn.f32 %f(49+i*5), %f(46+i*5), %f(48+i*5), %f(128+i*-79);
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(55+i*5), [%r4+(8+i*4)];
ld.shared.f32 %f(56+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(57+i*5), %f(56+i*5), %f(55+i*5), %f(52+i*5);
ld.shared.f32 %f(58+i*5), [%r4+(2056+i*4)];
fma.rn.f32 %f(59+i*5), %f(56+i*5), %f(58+i*5), %f(54+i*5);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(65+i*2), [%r4+(32+i*2048)];
fma.rn.f32 %f(66+i*2), %f46, %f(65+i*2), %f(127+i*-1);
for.size.3 i in range(0, 3, 1):
for.size.2 j in range(0, 2, 1):
ld.shared.f32 %f(69+i*4+j*2), [%r4+(36+i*4+j*2048)];
fma.rn.f32 %f(70+i*4+j*2), %f(51+j*5), %f(69+i*4+j*2), %f(66+i*4+j*2);
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(81+i*5), [%r4+(16+i*4)];
ld.shared.f32 %f(82+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(16+i*4)];
fma.rn.f32 %f(83+i*5), %f(82+i*5), %f(81+i*5), %f(62+i*21);
ld.shared.f32 %f(84+i*5), [%r4+(2064+i*4)];
fma.rn.f32 %f(85+i*5), %f(82+i*5), %f(84+i*5), %f(64+i*21);
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(91+i*5), [%r4+(24+i*4)];
ld.shared.f32 %f(92+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(24+i*4)];
fma.rn.f32 %f(93+i*36), %f(92+i*5), %f(91+i*5), %f(88+i*5);
ld.shared.f32 %f(94+i*4), [%r4+(2072+i*4)];
fma.rn.f32 %f(95+i*33), %f(92+i*5), %f(94+i*4), %f(90+i*5);
for.size.3 i in range(0, 2, 1):
for.size.2 j in range(0, 2, 1):
ld.shared.f32 %f(99+i*4+j*2), [%r4+(48+i*4+j*2048)];
fma.rn.f32 %f(100+i*4+j*2), %f(82+j*5), %f(99+i*4+j*2), %f(78+i*22+j*2);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(107+i*2), [%r4+(56+i*2048)];
fma.rn.f32 %f(108+i*2), %f92, %f(107+i*2), %f(104+i*2);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(111+i*1), [%r4+(60+i*2048)];
fma.rn.f32 %f(127+i*-1), %f97, %f(111+i*1), %f(108+i*2);
for.size.1 i in range(0, 2, 1):
add.s64 %rd(23+i*-1), %rd(23+i*-1), 32;
add.s32 %r35, %r35, 1;
setp.ne.s32 %p2, %r35, 98;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd18, %rd8;
shl.b32 %r34, %r1, 1;
mul.wide.s32 %rd19, %r34, 4;
add.s64 %rd20, %rd18, %rd19;
ld.global.nc.f32 %f113, [%rd20];
add.f32 %f114, %f129, %f113;
mov.f32 %f115, 0f00000000;
max.f32 %f116, %f114, %f115;
add.s64 %rd21, %rd1, %rd19;
for.size.4 i in range(0, 2, 1):
st.global.f32 [%rd21+(0+i*256)], %f(116+i*3);
ld.global.nc.f32 %f(117+i*3), [%rd20+(256+i*-252)];
add.f32 %f(118+i*3), %f(128+i*-1), %f(117+i*3);
max.f32 %f(119+i*3), %f(118+i*3), %f115;
st.global.f32 [%rd21+4], %f122;
ld.global.nc.f32 %f123, [%rd20+260];
add.f32 %f124, %f126, %f123;
max.f32 %f125, %f124, %f115;
st.global.f32 [%rd21+260], %f125;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[1024];
T_dense[0] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
ulonglong4*(placeholder_d_shared + (int)threadIdx.x * 8)[0] = ulonglong4*(placeholder + (int)threadIdx.x * 784 + k_outer_outer * 8)[0];
}
int index0 = (int)threadIdx.x;
int index1 = ((int)threadIdx.x >> 3) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 7);
placeholder_shared[index0] = placeholder1[index1];
placeholder_shared[index0 + 32] = placeholder1[index1 + 3136];
placeholder_shared[index0 + 64] = placeholder1[index1 + 6272];
placeholder_shared[index0 + 96] = placeholder1[index1 + 9408];
placeholder_shared[index0 + 128] = placeholder1[index1 + 12544];
placeholder_shared[index0 + 160] = placeholder1[index1 + 15680];
placeholder_shared[index0 + 192] = placeholder1[index1 + 18816];
placeholder_shared[index0 + 224] = placeholder1[index1 + 21952];
placeholder_shared[index0 + 256] = placeholder1[index1 + 25088];
placeholder_shared[index0 + 288] = placeholder1[index1 + 28224];
placeholder_shared[index0 + 320] = placeholder1[index1 + 31360];
placeholder_shared[index0 + 352] = placeholder1[index1 + 34496];
placeholder_shared[index0 + 384] = placeholder1[index1 + 37632];
placeholder_shared[index0 + 416] = placeholder1[index1 + 40768];
placeholder_shared[index0 + 448] = placeholder1[index1 + 43904];
placeholder_shared[index0 + 480] = placeholder1[index1 + 47040];
placeholder_shared[index0 + 512] = placeholder1[index1 + 50176];
placeholder_shared[index0 + 544] = placeholder1[index1 + 53312];
placeholder_shared[index0 + 576] = placeholder1[index1 + 56448];
placeholder_shared[index0 + 608] = placeholder1[index1 + 59584];
placeholder_shared[index0 + 640] = placeholder1[index1 + 62720];
placeholder_shared[index0 + 672] = placeholder1[index1 + 65856];
placeholder_shared[index0 + 704] = placeholder1[index1 + 68992];
placeholder_shared[index0 + 736] = placeholder1[index1 + 72128];
placeholder_shared[index0 + 768] = placeholder1[index1 + 75264];
placeholder_shared[index0 + 800] = placeholder1[index1 + 78400];
placeholder_shared[index0 + 832] = placeholder1[index1 + 81536];
placeholder_shared[index0 + 864] = placeholder1[index1 + 84672];
placeholder_shared[index0 + 896] = placeholder1[index1 + 87808];
placeholder_shared[index0 + 928] = placeholder1[index1 + 90944];
placeholder_shared[index0 + 960] = placeholder1[index1 + 94080];
placeholder_shared[index0 + 992] = placeholder1[index1 + 97216];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 16];
int index2 = (int)threadIdx.x * 16 + 512;
T_dense[2] = T_dense[2] + placeholder_d_shared[0] * placeholder_shared[index2];
int index3 = (int)threadIdx.x * 16 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index3];
int index4 = (int)threadIdx.x * 16 + 513;
T_dense[2] = T_dense[2] + placeholder_d_shared[1] * placeholder_shared[index4];
int index5 = (int)threadIdx.x * 16 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index5];
int index6 = (int)threadIdx.x * 16 + 514;
T_dense[2] = T_dense[2] + placeholder_d_shared[2] * placeholder_shared[index6];
int index7 = (int)threadIdx.x * 16 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index7];
int index8 = (int)threadIdx.x * 16 + 515;
T_dense[2] = T_dense[2] + placeholder_d_shared[3] * placeholder_shared[index8];
int index9 = (int)threadIdx.x * 16 + 8;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index9];
int index10 = (int)threadIdx.x * 16 + 520;
T_dense[3] = T_dense[3] + placeholder_d_shared[0] * placeholder_shared[index10];
int index11 = (int)threadIdx.x * 16 + 9;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index11];
int index12 = (int)threadIdx.x * 16 + 521;
T_dense[3] = T_dense[3] + placeholder_d_shared[1] * placeholder_shared[index12];
int index13 = (int)threadIdx.x * 16 + 10;
T_dense[1] = T_dense[1] + placeholder_d_shared[2] * placeholder_shared[index13];
int index14 = (int)threadIdx.x * 16 + 522;
T_dense[3] = T_dense[3] + placeholder_d_shared[2] * placeholder_shared[index14];
int index15 = (int)threadIdx.x * 16 + 11;
T_dense[1] = T_dense[1] + placeholder_d_shared[3] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 16 + 523;
T_dense[3] = T_dense[3] + placeholder_d_shared[3] * placeholder_shared[index16];
int index17 = (int)threadIdx.x * 16 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index17];
int index18 = (int)threadIdx.x * 16 + 516;
T_dense[2] = T_dense[2] + placeholder_d_shared[4] * placeholder_shared[index18];
int index19 = (int)threadIdx.x * 16 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index19];
int index20 = (int)threadIdx.x * 16 + 517;
T_dense[2] = T_dense[2] + placeholder_d_shared[5] * placeholder_shared[index20];
int index21 = (int)threadIdx.x * 16 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index21];
int index22 = (int)threadIdx.x * 16 + 518;
T_dense[2] = T_dense[2] + placeholder_d_shared[6] * placeholder_shared[index22];
int index23 = (int)threadIdx.x * 16 + 7;
T_dense[0] = T_dense[0] + placeholder_d_shared[7] * placeholder_shared[index23];
int index24 = (int)threadIdx.x * 16 + 519;
T_dense[2] = T_dense[2] + placeholder_d_shared[7] * placeholder_shared[index24];
int index25 = (int)threadIdx.x * 16 + 12;
T_dense[1] = T_dense[1] + placeholder_d_shared[4] * placeholder_shared[index25];
int index26 = (int)threadIdx.x * 16 + 524;
T_dense[3] = T_dense[3] + placeholder_d_shared[4] * placeholder_shared[index26];
int index27 = (int)threadIdx.x * 16 + 13;
T_dense[1] = T_dense[1] + placeholder_d_shared[5] * placeholder_shared[index27];
int index28 = (int)threadIdx.x * 16 + 525;
T_dense[3] = T_dense[3] + placeholder_d_shared[5] * placeholder_shared[index28];
int index29 = (int)threadIdx.x * 16 + 14;
T_dense[1] = T_dense[1] + placeholder_d_shared[6] * placeholder_shared[index29];
int index30 = (int)threadIdx.x * 16 + 526;
T_dense[3] = T_dense[3] + placeholder_d_shared[6] * placeholder_shared[index30];
int index31 = (int)threadIdx.x * 16 + 15;
T_dense[1] = T_dense[1] + placeholder_d_shared[7] * placeholder_shared[index31];
int index32 = (int)threadIdx.x * 16 + 527;
T_dense[3] = T_dense[3] + placeholder_d_shared[7] * placeholder_shared[index32];
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index33 = (int)threadIdx.x * 2 + ax1_inner;
int index34 = ax1_inner;
T_relu[index33] = max(T_dense[index34] + placeholder2[index33], 0.000000e+00f);
T_relu[index33 + 64] = max(T_dense[index34 + 2] + placeholder2[index33 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpqtf6r9yy | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<1291>;
.reg .b32 %r<282>;
.reg .b64 %rd<17>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[4096];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd5;
mov.u32 %r279, 0;
mov.f32 %f1035, 0f00000000;
for.size.1 i in range(0, 127, 1):
mov.f32 %f(1036+i*1), %f1035;
$L__BB0_1:
mov.u32 %r280, 0;
bar.sync 0;
shl.b32 %r2, %r279, 3;
mul.wide.s32 %rd9, %r2, 4;
add.s64 %rd10, %rd2, %rd9;
for.size.2 i in range(0, 8, 1):
ld.global.nc.f32 %f(513+i*1), [%rd10+(0+i*4)];
st.shared.f32 [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)], %f(513+i*1);
$L__BB0_2:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(276+i*-266), %r(279+i*1), (3+i*2);
mov.u32 %r11, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r12, %r11, %r10;
mad.lo.s32 %r13, %r280, 784, %r276;
mul.wide.s32 %rd11, %r13, 4;
add.s64 %rd12, %rd1, %rd11;
for.size.4 i in range(0, 16, 1):
for.size.1 j in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(14+i*16+j*8), %r(15+i*16+j*8), %r(16+i*16+j*8), %r(17+i*16+j*8)}, [%rd12+(0+i*3136+j*16)];
for.size.1 j in range(0, 2, 1):
st.shared.v4.u32 [%r12+(0+i*32+j*16)], {%r(14+i*16+j*8), %r(15+i*16+j*8), %r(16+i*16+j*8), %r(17+i*16+j*8)};
add.s32 %r280, %r280, 16;
setp.ne.s32 %p1, %r280, 128;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r281, 0;
$L__BB0_4:
mov.u32 %r277, _ZZ24default_function_kernel0E18placeholder_shared;
shl.b32 %r271, %r281, 2;
add.s32 %r273, %r277, %r271;
ld.shared.f32 %f521, [%r273];
mov.u32 %r274, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r275, %r274, %r271;
ld.shared.f32 %f522, [%r275];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(1162+i*-2), %f522, %f(521+i*2), %f(1162+i*-2);
ld.shared.f32 %f(523+i*1), [%r273+(64+i*64)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(1158+i*-2), %f522, %f(524+i*1), %f(1158+i*-2);
ld.shared.f32 %f(525+i*1), [%r273+(192+i*64)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(1036+i*125), %f522, %f(585+i*1), %f(1036+i*125);
ld.shared.f32 %f(586+i*1), [%r273+(32+i*64)];
for.size.2 i in range(0, 62, 1):
fma.rn.f32 %f(1159+i*-2), %f522, %f(587+i*1), %f(1159+i*-2);
ld.shared.f32 %f(588+i*1), [%r273+(160+i*64)];
fma.rn.f32 %f1035, %f522, %f649, %f1035;
add.s32 %r281, %r281, 1;
setp.ne.s32 %p2, %r281, 8;
@%p2 bra $L__BB0_4;
add.s32 %r279, %r279, 1;
setp.lt.u32 %p3, %r279, 98;
@%p3 bra $L__BB0_1;
ld.param.u64 %rd16, [default_function_kernel0_param_2];
cvta.to.global.u64 %rd15, %rd16;
ld.param.u64 %rd14, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd14;
ld.global.nc.f32 %f650, [%rd13];
add.f32 %f651, %f1162, %f650;
mov.f32 %f652, 0f00000000;
for.size.4 i in range(0, 2, 1):
max.f32 %f(653+i*3), %f(651+i*4), %f652;
st.global.f32 [%rd15+(0+i*8)], %f(653+i*3);
ld.global.nc.f32 %f(654+i*3), [%rd13+(8+i*8)];
add.f32 %f(655+i*3), %f(1160+i*-2), %f(654+i*3);
for.size.4 i in range(0, 61, 1):
max.f32 %f(659+i*3), %f(658+i*3), %f652;
st.global.f32 [%rd15+(16+i*8)], %f(659+i*3);
ld.global.nc.f32 %f(660+i*3), [%rd13+(24+i*8)];
add.f32 %f(661+i*3), %f(1156+i*-2), %f(660+i*3);
for.size.4 i in range(0, 2, 1):
max.f32 %f(842+i*3), %f(841+i*3), %f652;
st.global.f32 [%rd15+(504+i*-500)], %f(842+i*3);
ld.global.nc.f32 %f(843+i*3), [%rd13+(4+i*8)];
add.f32 %f(844+i*3), %f(1161+i*-2), %f(843+i*3);
for.size.4 i in range(0, 62, 1):
max.f32 %f(848+i*3), %f(847+i*3), %f652;
st.global.f32 [%rd15+(12+i*8)], %f(848+i*3);
ld.global.nc.f32 %f(849+i*3), [%rd13+(20+i*8)];
add.f32 %f(850+i*3), %f(1157+i*-2), %f(849+i*3);
max.f32 %f1034, %f1033, %f652;
st.global.f32 [%rd15+508], %f1034;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[128];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[1024];
for (int j_inner_init = 0; j_inner_init < 2; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
T_dense[index0 + 2] = 0.000000e+00f;
T_dense[index0 + 4] = 0.000000e+00f;
T_dense[index0 + 6] = 0.000000e+00f;
T_dense[index0 + 8] = 0.000000e+00f;
T_dense[index0 + 10] = 0.000000e+00f;
T_dense[index0 + 12] = 0.000000e+00f;
T_dense[index0 + 14] = 0.000000e+00f;
T_dense[index0 + 16] = 0.000000e+00f;
T_dense[index0 + 18] = 0.000000e+00f;
T_dense[index0 + 20] = 0.000000e+00f;
T_dense[index0 + 22] = 0.000000e+00f;
T_dense[index0 + 24] = 0.000000e+00f;
T_dense[index0 + 26] = 0.000000e+00f;
T_dense[index0 + 28] = 0.000000e+00f;
T_dense[index0 + 30] = 0.000000e+00f;
T_dense[index0 + 32] = 0.000000e+00f;
T_dense[index0 + 34] = 0.000000e+00f;
T_dense[index0 + 36] = 0.000000e+00f;
T_dense[index0 + 38] = 0.000000e+00f;
T_dense[index0 + 40] = 0.000000e+00f;
T_dense[index0 + 42] = 0.000000e+00f;
T_dense[index0 + 44] = 0.000000e+00f;
T_dense[index0 + 46] = 0.000000e+00f;
T_dense[index0 + 48] = 0.000000e+00f;
T_dense[index0 + 50] = 0.000000e+00f;
T_dense[index0 + 52] = 0.000000e+00f;
T_dense[index0 + 54] = 0.000000e+00f;
T_dense[index0 + 56] = 0.000000e+00f;
T_dense[index0 + 58] = 0.000000e+00f;
T_dense[index0 + 60] = 0.000000e+00f;
T_dense[index0 + 62] = 0.000000e+00f;
T_dense[index0 + 64] = 0.000000e+00f;
T_dense[index0 + 66] = 0.000000e+00f;
T_dense[index0 + 68] = 0.000000e+00f;
T_dense[index0 + 70] = 0.000000e+00f;
T_dense[index0 + 72] = 0.000000e+00f;
T_dense[index0 + 74] = 0.000000e+00f;
T_dense[index0 + 76] = 0.000000e+00f;
T_dense[index0 + 78] = 0.000000e+00f;
T_dense[index0 + 80] = 0.000000e+00f;
T_dense[index0 + 82] = 0.000000e+00f;
T_dense[index0 + 84] = 0.000000e+00f;
T_dense[index0 + 86] = 0.000000e+00f;
T_dense[index0 + 88] = 0.000000e+00f;
T_dense[index0 + 90] = 0.000000e+00f;
T_dense[index0 + 92] = 0.000000e+00f;
T_dense[index0 + 94] = 0.000000e+00f;
T_dense[index0 + 96] = 0.000000e+00f;
T_dense[index0 + 98] = 0.000000e+00f;
T_dense[index0 + 100] = 0.000000e+00f;
T_dense[index0 + 102] = 0.000000e+00f;
T_dense[index0 + 104] = 0.000000e+00f;
T_dense[index0 + 106] = 0.000000e+00f;
T_dense[index0 + 108] = 0.000000e+00f;
T_dense[index0 + 110] = 0.000000e+00f;
T_dense[index0 + 112] = 0.000000e+00f;
T_dense[index0 + 114] = 0.000000e+00f;
T_dense[index0 + 116] = 0.000000e+00f;
T_dense[index0 + 118] = 0.000000e+00f;
T_dense[index0 + 120] = 0.000000e+00f;
T_dense[index0 + 122] = 0.000000e+00f;
T_dense[index0 + 124] = 0.000000e+00f;
T_dense[index0 + 126] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
int index1 = k_outer_outer * 8;
placeholder_d_shared[0] = placeholder[index1];
placeholder_d_shared[1] = placeholder[index1 + 1];
placeholder_d_shared[2] = placeholder[index1 + 2];
placeholder_d_shared[3] = placeholder[index1 + 3];
placeholder_d_shared[4] = placeholder[index1 + 4];
placeholder_d_shared[5] = placeholder[index1 + 5];
placeholder_d_shared[6] = placeholder[index1 + 6];
placeholder_d_shared[7] = placeholder[index1 + 7];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 128; ++ax0_ax1_fused_outer_outer) {
ulonglong4*(placeholder_shared + ax0_ax1_fused_outer_outer * 8)[0] = ulonglong4*(placeholder1 + ax0_ax1_fused_outer_outer * 784 + k_outer_outer * 8)[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 8; ++k_outer_inner) {
for (int j_inner = 0; j_inner < 2; ++j_inner) {
int index2 = j_inner;
int index3 = k_outer_inner;
int index4 = j_inner * 8 + k_outer_inner;
T_dense[index2] = T_dense[index2] + placeholder_d_shared[index3] * placeholder_shared[index4];
T_dense[index2 + 2] = T_dense[index2 + 2] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 16]);
T_dense[index2 + 4] = T_dense[index2 + 4] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 32]);
T_dense[index2 + 6] = T_dense[index2 + 6] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 48]);
T_dense[index2 + 8] = T_dense[index2 + 8] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 64]);
T_dense[index2 + 10] = T_dense[index2 + 10] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 80]);
T_dense[index2 + 12] = T_dense[index2 + 12] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 96]);
T_dense[index2 + 14] = T_dense[index2 + 14] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 112]);
T_dense[index2 + 16] = T_dense[index2 + 16] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 128]);
T_dense[index2 + 18] = T_dense[index2 + 18] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 144]);
T_dense[index2 + 20] = T_dense[index2 + 20] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 160]);
T_dense[index2 + 22] = T_dense[index2 + 22] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 176]);
T_dense[index2 + 24] = T_dense[index2 + 24] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 192]);
T_dense[index2 + 26] = T_dense[index2 + 26] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 208]);
T_dense[index2 + 28] = T_dense[index2 + 28] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 224]);
T_dense[index2 + 30] = T_dense[index2 + 30] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 240]);
T_dense[index2 + 32] = T_dense[index2 + 32] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 256]);
T_dense[index2 + 34] = T_dense[index2 + 34] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 272]);
T_dense[index2 + 36] = T_dense[index2 + 36] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 288]);
T_dense[index2 + 38] = T_dense[index2 + 38] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 304]);
T_dense[index2 + 40] = T_dense[index2 + 40] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 320]);
T_dense[index2 + 42] = T_dense[index2 + 42] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 336]);
T_dense[index2 + 44] = T_dense[index2 + 44] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 352]);
T_dense[index2 + 46] = T_dense[index2 + 46] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 368]);
T_dense[index2 + 48] = T_dense[index2 + 48] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 384]);
T_dense[index2 + 50] = T_dense[index2 + 50] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 400]);
T_dense[index2 + 52] = T_dense[index2 + 52] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 416]);
T_dense[index2 + 54] = T_dense[index2 + 54] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 432]);
T_dense[index2 + 56] = T_dense[index2 + 56] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 448]);
T_dense[index2 + 58] = T_dense[index2 + 58] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 464]);
T_dense[index2 + 60] = T_dense[index2 + 60] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 480]);
T_dense[index2 + 62] = T_dense[index2 + 62] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 496]);
T_dense[index2 + 64] = T_dense[index2 + 64] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 512]);
T_dense[index2 + 66] = T_dense[index2 + 66] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 528]);
T_dense[index2 + 68] = T_dense[index2 + 68] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 544]);
T_dense[index2 + 70] = T_dense[index2 + 70] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 560]);
T_dense[index2 + 72] = T_dense[index2 + 72] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 576]);
T_dense[index2 + 74] = T_dense[index2 + 74] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 592]);
T_dense[index2 + 76] = T_dense[index2 + 76] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 608]);
T_dense[index2 + 78] = T_dense[index2 + 78] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 624]);
T_dense[index2 + 80] = T_dense[index2 + 80] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 640]);
T_dense[index2 + 82] = T_dense[index2 + 82] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 656]);
T_dense[index2 + 84] = T_dense[index2 + 84] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 672]);
T_dense[index2 + 86] = T_dense[index2 + 86] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 688]);
T_dense[index2 + 88] = T_dense[index2 + 88] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 704]);
T_dense[index2 + 90] = T_dense[index2 + 90] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 720]);
T_dense[index2 + 92] = T_dense[index2 + 92] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 736]);
T_dense[index2 + 94] = T_dense[index2 + 94] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 752]);
T_dense[index2 + 96] = T_dense[index2 + 96] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 768]);
T_dense[index2 + 98] = T_dense[index2 + 98] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 784]);
T_dense[index2 + 100] = T_dense[index2 + 100] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 800]);
T_dense[index2 + 102] = T_dense[index2 + 102] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 816]);
T_dense[index2 + 104] = T_dense[index2 + 104] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 832]);
T_dense[index2 + 106] = T_dense[index2 + 106] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 848]);
T_dense[index2 + 108] = T_dense[index2 + 108] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 864]);
T_dense[index2 + 110] = T_dense[index2 + 110] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 880]);
T_dense[index2 + 112] = T_dense[index2 + 112] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 896]);
T_dense[index2 + 114] = T_dense[index2 + 114] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 912]);
T_dense[index2 + 116] = T_dense[index2 + 116] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 928]);
T_dense[index2 + 118] = T_dense[index2 + 118] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 944]);
T_dense[index2 + 120] = T_dense[index2 + 120] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 960]);
T_dense[index2 + 122] = T_dense[index2 + 122] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 976]);
T_dense[index2 + 124] = T_dense[index2 + 124] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 992]);
T_dense[index2 + 126] = T_dense[index2 + 126] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 1008]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
for (int vthread_s = 0; vthread_s < 64; ++vthread_s) {
int index5 = vthread_s * 2 + ax1_inner;
T_relu[index5] = max(T_dense[index5] + placeholder2[index5], 0.000000e+00f);
}
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpyffb4r0w | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<6>;
.reg .b16 %rs<73>;
.reg .f32 %f<228>;
.reg .b32 %r<239>;
.reg .b64 %rd<34>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[1568];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[25088];
ld.param.u64 %rd8, [default_function_kernel0_param_0];
ld.param.u64 %rd9, [default_function_kernel0_param_1];
ld.param.u64 %rd10, [default_function_kernel0_param_2];
ld.param.u64 %rd11, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd8;
cvta.to.global.u64 %rd1, %rd9;
cvta.to.global.u64 %rd2, %rd10;
cvta.to.global.u64 %rd3, %rd11;
mov.u32 %r1, %ctaid.x;
mul.lo.s32 %r2, %r1, 12544;
add.s64 %rd4, %rd12, 16;
mov.u32 %r18, 0;
mov.f32 %f212, 0f00000000;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(213+i*1), %f212;
mov.u32 %r233, %r18;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r4, %r233, 392;
mul.wide.s32 %rd13, %r4, 4;
add.s64 %rd33, %rd4, %rd13;
mov.u32 %r234, %r19;
mov.u32 %r235, %r18;
$L__BB0_2:
for.size.2 i in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(21+i*8), %r(22+i*8), %r(23+i*8), %r(24+i*8)}, [%rd33+(-16+i*16)];
st.shared.v4.u32 [%r234+(0+i*16)], {%r(21+i*8), %r(22+i*8), %r(23+i*8), %r(24+i*8)};
add.s64 %rd33, %rd33, 32;
for.size.1 i in range(0, 2, 1):
add.s32 %r(234+i*1), %r(234+i*1), (32+i*-30);
setp.ne.s32 %p1, %r235, 98;
@%p1 bra $L__BB0_2;
add.s32 %r9, %r4, %r2;
mov.u32 %r236, 0;
$L__BB0_4:
shl.b32 %r38, %r236, 5;
mov.u32 %r39, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r40, %r39, %r38;
for.size.22 i in range(0, 2, 1):
cvt.u16.u32 %rs(1+i*9), %r(236+i*-173);
mul.wide.u16 %r(41+i*23), %rs(1+i*9), 20063;
shr.u32 %r(42+i*23), %r(41+i*23), 16;
cvt.u16.u32 %rs(2+i*9), %r(42+i*23);
sub.s16 %rs(3+i*9), %rs(1+i*9), %rs(2+i*9);
shr.u16 %rs(4+i*9), %rs(3+i*9), 1;
add.s16 %rs(5+i*9), %rs(4+i*9), %rs(2+i*9);
shr.u16 %rs(6+i*9), %rs(5+i*9), 5;
cvt.u32.u16 %r(43+i*23), %rs(6+i*9);
mul.lo.s16 %rs(7+i*9), %rs(6+i*9), 49;
sub.s16 %rs(8+i*9), %rs(1+i*9), %rs(7+i*9);
shl.b16 %rs(9+i*9), %rs(8+i*9), 3;
cvt.u32.u16 %r(44+i*23), %rs(9+i*9);
mad.lo.s32 %r(45+i*23), %r(43+i*23), 784, %r9;
add.s32 %r(46+i*23), %r(45+i*23), %r(44+i*23);
mul.wide.s32 %rd(14+i*2), %r(46+i*23), 4;
add.s64 %rd(15+i*2), %rd1, %rd(14+i*2);
ld.global.nc.v4.u32 {%r(47+i*23), %r(48+i*23), %r(49+i*23), %r(50+i*23)}, [%rd(15+i*2)+0];
ld.global.nc.v4.u32 {%r(55+i*23), %r(56+i*23), %r(57+i*23), %r(58+i*23)}, [%rd(15+i*2)+16];
st.shared.v4.u32 [%r40+(0+i*32)], {%r(47+i*23), %r(48+i*23), %r(49+i*23), %r(50+i*23)};
st.shared.v4.u32 [%r40+(16+i*32)], {%r(55+i*23), %r(56+i*23), %r(57+i*23), %r(58+i*23)};
add.s32 %r(63+i*23), %r236, (1+i*1);
for.size.22 i in range(0, 5, 1):
cvt.u16.u32 %rs(19+i*9), %r(86+i*23);
mul.wide.u16 %r(87+i*23), %rs(19+i*9), 20063;
shr.u32 %r(88+i*23), %r(87+i*23), 16;
cvt.u16.u32 %rs(20+i*9), %r(88+i*23);
sub.s16 %rs(21+i*9), %rs(19+i*9), %rs(20+i*9);
shr.u16 %rs(22+i*9), %rs(21+i*9), 1;
add.s16 %rs(23+i*9), %rs(22+i*9), %rs(20+i*9);
shr.u16 %rs(24+i*9), %rs(23+i*9), 5;
cvt.u32.u16 %r(89+i*23), %rs(24+i*9);
mul.lo.s16 %rs(25+i*9), %rs(24+i*9), 49;
sub.s16 %rs(26+i*9), %rs(19+i*9), %rs(25+i*9);
shl.b16 %rs(27+i*9), %rs(26+i*9), 3;
cvt.u32.u16 %r(90+i*23), %rs(27+i*9);
mad.lo.s32 %r(91+i*23), %r(89+i*23), 784, %r9;
add.s32 %r(92+i*23), %r(91+i*23), %r(90+i*23);
mul.wide.s32 %rd(18+i*2), %r(92+i*23), 4;
add.s64 %rd(19+i*2), %rd1, %rd(18+i*2);
ld.global.nc.v4.u32 {%r(93+i*23), %r(94+i*23), %r(95+i*23), %r(96+i*23)}, [%rd(19+i*2)+0];
ld.global.nc.v4.u32 {%r(101+i*23), %r(102+i*23), %r(103+i*23), %r(104+i*23)}, [%rd(19+i*2)+16];
st.shared.v4.u32 [%r40+(64+i*32)], {%r(93+i*23), %r(94+i*23), %r(95+i*23), %r(96+i*23)};
st.shared.v4.u32 [%r40+(80+i*32)], {%r(101+i*23), %r(102+i*23), %r(103+i*23), %r(104+i*23)};
add.s32 %r(109+i*23), %r236, (3+i*1);
cvt.u16.u32 %rs64, %r201;
mul.wide.u16 %r202, %rs64, 20063;
shr.u32 %r203, %r202, 16;
cvt.u16.u32 %rs65, %r203;
sub.s16 %rs66, %rs64, %rs65;
shr.u16 %rs67, %rs66, 1;
add.s16 %rs68, %rs67, %rs65;
shr.u16 %rs69, %rs68, 5;
cvt.u32.u16 %r204, %rs69;
mul.lo.s16 %rs70, %rs69, 49;
sub.s16 %rs71, %rs64, %rs70;
shl.b16 %rs72, %rs71, 3;
cvt.u32.u16 %r205, %rs72;
mad.lo.s32 %r206, %r204, 784, %r9;
add.s32 %r207, %r206, %r205;
mul.wide.s32 %rd28, %r207, 4;
add.s64 %rd29, %rd1, %rd28;
for.size.1 i in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(208+i*8), %r(209+i*8), %r(210+i*8), %r(211+i*8)}, [%rd29+(0+i*16)];
for.size.1 i in range(0, 2, 1):
st.shared.v4.u32 [%r40+(224+i*16)], {%r(208+i*8), %r(209+i*8), %r(210+i*8), %r(211+i*8)};
add.s32 %r236, %r236, 8;
setp.ne.s32 %p2, %r236, 784;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r224, 0;
mov.u32 %r237, %r224;
$L__BB0_6:
mul.lo.s32 %r13, %r237, 14;
mov.u32 %r238, %r224;
$L__BB0_7:
add.s32 %r226, %r238, %r13;
shl.b32 %r227, %r226, 2;
mov.u32 %r228, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(229+i*2), %r(228+i*-189), %r227;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(81+i*1), [%r(231+i*-2)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(83+i*2), %f82, %f(81+i*3), %f(227+i*-1);
ld.shared.f32 %f(84+i*2), [%r231+(12544+i*-10976)];
for.size.3 i in range(0, 6, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(87+i*4+j*2), %f82, %f(86+i*4+j*2), %f(225+i*-2+j*-1);
ld.shared.f32 %f(88+i*4+j*2), [%r231+(14112+i*1568+j*-10976)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(111+i*2), %f82, %f(110+i*2), %f(213+i*-1);
ld.shared.f32 %f(112+i*2), [%r231+(23520+i*-23516)];
ld.shared.f32 %f115, [%r229+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(227+i*-1), %f115, %f(114+i*2), %f(83+i*2);
ld.shared.f32 %f(116+i*1), [%r231+(12548+i*-10976)];
for.size.3 i in range(0, 6, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(225+i*-2+j*-1), %f115, %f(117+i*2+j*1), %f(87+i*4+j*2);
ld.shared.f32 %f(118+i*2+j*1), [%r231+(14116+i*1568+j*-10976)];
fma.rn.f32 %f213, %f115, %f129, %f111;
ld.shared.f32 %f130, [%r231+23524];
fma.rn.f32 %f212, %f115, %f130, %f113;
add.s32 %r238, %r238, 2;
setp.ne.s32 %p3, %r238, 14;
@%p3 bra $L__BB0_7;
add.s32 %r237, %r237, 1;
setp.lt.u32 %p4, %r237, 28;
@%p4 bra $L__BB0_6;
add.s32 %r233, %r233, 1;
setp.lt.u32 %p5, %r233, 2;
@%p5 bra $L__BB0_1;
shl.b32 %r232, %r1, 4;
mul.wide.s32 %rd30, %r232, 4;
add.s64 %rd31, %rd3, %rd30;
ld.global.nc.f32 %f131, [%rd31];
add.f32 %f132, %f227, %f131;
mov.f32 %f133, 0f00000000;
max.f32 %f134, %f132, %f133;
add.s64 %rd32, %rd2, %rd30;
for.size.5 i in range(0, 7, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd32+(0+i*4+j*32)], %f(134+i*6+j*3);
ld.global.nc.f32 %f(135+i*6+j*3), [%rd31+(32+i*4+j*-28)];
add.f32 %f(136+i*6+j*3), %f(226+i*-2+j*-1), %f(135+i*6+j*3);
max.f32 %f(137+i*6+j*3), %f(136+i*6+j*3), %f133;
st.global.f32 [%rd32+28], %f176;
ld.global.nc.f32 %f177, [%rd31+60];
add.f32 %f178, %f212, %f177;
max.f32 %f179, %f178, %f133;
st.global.f32 [%rd32+60], %f179;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[392];
__shared__ float placeholder_shared[6272];
T_dense[0] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 2; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
((float4*)(placeholder_d_shared + ax0_ax1_fused_outer_outer * 4))[0] = ((float4*)(placeholder + k_outer_outer * 392 + ax0_ax1_fused_outer_outer * 4))[0];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 784; ++ax0_ax1_fused_outer_outer1) {
ulonglong4*(placeholder_shared + ax0_ax1_fused_outer_outer1 * 8)[0] = ulonglong4*(placeholder1 + (int)blockIdx.x * 12544 + ax0_ax1_fused_outer_outer1 / 49 * 784 + k_outer_outer * 392 + ax0_ax1_fused_outer_outer1 % 49 * 8)[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 28; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 14; ++k_inner) {
int index0 = k_outer_inner * 14 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index0] * placeholder_shared[index0];
T_dense[8] = T_dense[8] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 3136]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 392]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 3528]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 784]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 3920]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 1176]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 4312]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 1568]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 4704]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 1960]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 5096]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 2352]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 5488]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 2744]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index0] * placeholder_shared[index0 + 5880]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 8; ++ax1_inner) {
int index1 = (int)blockIdx.x * 16 + ax1_inner;
int index2 = ax1_inner;
T_relu[index1] = max(T_dense[index2] + placeholder2[index1], 0.000000e+00f);
T_relu[index1 + 8] = max(T_dense[index2 + 8] + placeholder2[index1 + 8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpqmtaqn3t | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<102>;
.reg .b32 %r<81>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[4096];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd1, %rd11;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(8+i*1), %r1, (2+i*2);
mov.u32 %r10, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r10, %r9;
shl.b32 %r11, %r1, 5;
mov.u32 %r12, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r12, %r11;
shl.b32 %r13, %r1, 6;
add.s32 %r4, %r12, %r13;
cvta.to.global.u64 %rd13, %rd9;
mul.wide.s32 %rd14, %r8, 4;
add.s64 %rd15, %rd13, %rd14;
add.s64 %rd23, %rd15, 8;
mul.lo.s32 %r14, %r1, 784;
mul.wide.s32 %rd16, %r14, 4;
add.s64 %rd17, %rd12, %rd16;
add.s64 %rd22, %rd17, 200704;
mov.u32 %r80, 0;
mov.f32 %f98, 0f00000000;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(99+i*1), %f98;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_3;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(13+i*1), [%rd23+(-8+i*4)];
st.shared.f32 [%r2+(0+i*4)], %f(13+i*1);
$L__BB0_3:
for.size.4 i in range(0, 4, 1):
for.size.1 j in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(15+i*16+j*8), %r(16+i*16+j*8), %r(17+i*16+j*8), %r(18+i*16+j*8)}, [%rd22+(-200704+i*100352+j*16)];
for.size.1 j in range(0, 2, 1):
st.shared.v4.u32 [%r3+(0+i*1024+j*16)], {%r(15+i*16+j*8), %r(16+i*16+j*8), %r(17+i*16+j*8), %r(18+i*16+j*8)};
bar.sync 0;
for.size.9 i in range(0, 2, 1):
ld.shared.f32 %f(17+i*9), [%r4+(0+i*4)];
ld.shared.f32 %f(18+i*9), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(19+i*9), %f(18+i*9), %f(17+i*9), %f(101+i*-82);
ld.shared.f32 %f(20+i*9), [%r4+(2048+i*4)];
fma.rn.f32 %f(21+i*9), %f(18+i*9), %f(20+i*9), %f(100+i*-79);
ld.shared.f32 %f(22+i*9), [%r4+(32+i*4)];
fma.rn.f32 %f(23+i*9), %f(18+i*9), %f(22+i*9), %f(99+i*-76);
ld.shared.f32 %f(24+i*9), [%r4+(2080+i*4)];
fma.rn.f32 %f(25+i*9), %f(18+i*9), %f(24+i*9), %f(98+i*-73);
for.size.9 i in range(0, 5, 1):
ld.shared.f32 %f(35+i*9), [%r4+(8+i*4)];
ld.shared.f32 %f(36+i*9), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(37+i*9), %f(36+i*9), %f(35+i*9), %f(28+i*9);
ld.shared.f32 %f(38+i*9), [%r4+(2056+i*4)];
fma.rn.f32 %f(39+i*9), %f(36+i*9), %f(38+i*9), %f(30+i*9);
ld.shared.f32 %f(40+i*9), [%r4+(40+i*4)];
fma.rn.f32 %f(41+i*9), %f(36+i*9), %f(40+i*9), %f(32+i*9);
ld.shared.f32 %f(42+i*9), [%r4+(2088+i*4)];
fma.rn.f32 %f(43+i*9), %f(36+i*9), %f(42+i*9), %f(34+i*9);
ld.shared.f32 %f80, [%r4+28];
ld.shared.f32 %f81, [_ZZ24default_function_kernel0E20placeholder_d_shared+28];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(101+i*-1), %f81, %f(80+i*2), %f(73+i*2);
ld.shared.f32 %f(82+i*1), [%r4+(2076+i*-2016)];
fma.rn.f32 %f99, %f81, %f83, %f77;
ld.shared.f32 %f84, [%r4+2108];
fma.rn.f32 %f98, %f81, %f84, %f79;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(23+i*-1), %rd(23+i*-1), 32;
add.s32 %r80, %r80, 1;
setp.ne.s32 %p2, %r80, 98;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd18, %rd8;
shl.b32 %r79, %r1, 1;
mul.wide.s32 %rd19, %r79, 4;
add.s64 %rd20, %rd18, %rd19;
ld.global.nc.f32 %f85, [%rd20];
add.f32 %f86, %f101, %f85;
mov.f32 %f87, 0f00000000;
max.f32 %f88, %f86, %f87;
add.s64 %rd21, %rd1, %rd19;
for.size.4 i in range(0, 2, 1):
st.global.f32 [%rd21+(0+i*256)], %f(88+i*3);
ld.global.nc.f32 %f(89+i*3), [%rd20+(256+i*-252)];
add.f32 %f(90+i*3), %f(100+i*-1), %f(89+i*3);
max.f32 %f(91+i*3), %f(90+i*3), %f87;
st.global.f32 [%rd21+4], %f94;
ld.global.nc.f32 %f95, [%rd20+260];
add.f32 %f96, %f98, %f95;
max.f32 %f97, %f96, %f87;
st.global.f32 [%rd21+260], %f97;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[1024];
T_dense[0] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 8 + (int)threadIdx.x * 4;
placeholder_d_shared[index0] = placeholder[index1];
}
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 8 + (int)threadIdx.x * 4;
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
}
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 8 + (int)threadIdx.x * 4;
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
}
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 8 + (int)threadIdx.x * 4;
placeholder_d_shared[index0 + 3] = placeholder[index1 + 3];
}
ulonglong4*(placeholder_shared + (int)threadIdx.x * 8)[0] = ulonglong4*(placeholder1 + (int)threadIdx.x * 784 + k_outer_outer * 8)[0];
ulonglong4*(placeholder_shared + (int)threadIdx.x * 8 + 256)[0] = ulonglong4*(placeholder1 + (int)threadIdx.x * 784 + k_outer_outer * 8 + 25088)[0];
ulonglong4*(placeholder_shared + (int)threadIdx.x * 8 + 512)[0] = ulonglong4*(placeholder1 + (int)threadIdx.x * 784 + k_outer_outer * 8 + 50176)[0];
ulonglong4*(placeholder_shared + (int)threadIdx.x * 8 + 768)[0] = ulonglong4*(placeholder1 + (int)threadIdx.x * 784 + k_outer_outer * 8 + 75264)[0];
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 8; ++k_outer_inner) {
int index2 = k_outer_inner;
int index3 = (int)threadIdx.x * 16 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 512]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 8]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 520]);
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index4 = (int)threadIdx.x * 2 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[index5 + 2] + placeholder2[index4 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpufy11w5t | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<478>;
.reg .b32 %r<41>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd11, [default_function_kernel0_param_1];
ld.param.u64 %rd12, [default_function_kernel0_param_2];
ld.param.u64 %rd13, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd14, %rd10;
cvta.to.global.u64 %rd1, %rd11;
cvta.to.global.u64 %rd2, %rd12;
cvta.to.global.u64 %rd3, %rd13;
mov.u32 %r1, %tid.x;
shl.b32 %r19, %r1, 2;
mov.u32 %r20, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r21, %r20, %r19;
add.s32 %r35, %r21, 32;
mul.wide.s32 %rd15, %r1, 4;
add.s64 %rd16, %rd14, %rd15;
add.s64 %rd22, %rd16, 32;
mov.u32 %r36, 0;
$L__BB0_1:
for.size.2 i in range(0, 8, 1):
ld.global.nc.f32 %f(65+i*1), [%rd22+(-32+i*8)];
st.shared.f32 [%r35+(-32+i*8)], %f(65+i*1);
add.s32 %r35, %r35, 64;
add.s64 %rd22, %rd22, 64;
add.s32 %r36, %r36, 8;
setp.ne.s32 %p1, %r36, 392;
@%p1 bra $L__BB0_1;
mov.u32 %r7, %ctaid.x;
mad.lo.s32 %r23, %r7, 50176, %r1;
mul.wide.s32 %rd17, %r23, 4;
add.s64 %rd18, %rd1, %rd17;
add.s64 %rd23, %rd18, 64;
mov.u32 %r25, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r26, %r25, %r19;
add.s32 %r37, %r26, 64;
mov.u32 %r38, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(73+i*1), [%rd23+(-64+i*8)];
st.shared.f32 [%r37+(-64+i*8)], %f(73+i*1);
add.s64 %rd23, %rd23, 128;
for.size.1 i in range(0, 2, 1):
add.s32 %r(37+i*1), %r(37+i*1), (128+i*-112);
setp.ne.s32 %p2, %r38, 25088;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r29, %r1, 6272, %r25;
add.s32 %r39, %r29, 100352;
mov.u32 %r40, 100352;
mov.f32 %f446, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(447+i*1), %f446;
$L__BB0_5:
add.s32 %r31, %r20, %r40;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(121+i*1), [%r(39+i*-8)+-100352];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(123+i*2), %f122, %f(121+i*3), %f(477+i*-2);
ld.shared.f32 %f(124+i*2), [%r39+(-87808+i*12544)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(127+i*2), %f122, %f(126+i*2), %f(473+i*-2);
ld.shared.f32 %f(128+i*2), [%r39+(-62720+i*12544)];
for.size.33 i in range(0, 2, 1):
fma.rn.f32 %f(153+i*33), %f(122+i*33), %f(152+i*33), %f(447+i*-294);
ld.shared.f32 %f(154+i*33), [%r39+(-100348+i*4)];
ld.shared.f32 %f(155+i*33), [%r31+(-100348+i*4)];
fma.rn.f32 %f(156+i*33), %f(155+i*33), %f(154+i*33), %f(123+i*33);
ld.shared.f32 %f(157+i*33), [%r39+(-87804+i*4)];
fma.rn.f32 %f(158+i*33), %f(155+i*33), %f(157+i*33), %f(125+i*33);
ld.shared.f32 %f(159+i*33), [%r39+(-75260+i*4)];
fma.rn.f32 %f(160+i*33), %f(155+i*33), %f(159+i*33), %f(127+i*33);
ld.shared.f32 %f(161+i*33), [%r39+(-62716+i*4)];
fma.rn.f32 %f(162+i*33), %f(155+i*33), %f(161+i*33), %f(129+i*33);
ld.shared.f32 %f(163+i*33), [%r39+(-50172+i*4)];
fma.rn.f32 %f(164+i*33), %f(155+i*33), %f(163+i*33), %f(131+i*33);
ld.shared.f32 %f(165+i*33), [%r39+(-37628+i*4)];
fma.rn.f32 %f(166+i*33), %f(155+i*33), %f(165+i*33), %f(133+i*33);
ld.shared.f32 %f(167+i*33), [%r39+(-25084+i*4)];
fma.rn.f32 %f(168+i*33), %f(155+i*33), %f(167+i*33), %f(135+i*33);
ld.shared.f32 %f(169+i*33), [%r39+(-12540+i*4)];
fma.rn.f32 %f(170+i*33), %f(155+i*33), %f(169+i*33), %f(137+i*33);
ld.shared.f32 %f(171+i*33), [%r39+(4+i*4)];
fma.rn.f32 %f(172+i*33), %f(155+i*33), %f(171+i*33), %f(139+i*33);
ld.shared.f32 %f(173+i*33), [%r39+(12548+i*4)];
fma.rn.f32 %f(174+i*33), %f(155+i*33), %f(173+i*33), %f(141+i*33);
ld.shared.f32 %f(175+i*33), [%r39+(25092+i*4)];
fma.rn.f32 %f(176+i*33), %f(155+i*33), %f(175+i*33), %f(143+i*33);
ld.shared.f32 %f(177+i*33), [%r39+(37636+i*4)];
fma.rn.f32 %f(178+i*33), %f(155+i*33), %f(177+i*33), %f(145+i*33);
ld.shared.f32 %f(179+i*33), [%r39+(50180+i*4)];
fma.rn.f32 %f(180+i*33), %f(155+i*33), %f(179+i*33), %f(147+i*33);
ld.shared.f32 %f(181+i*33), [%r39+(62724+i*4)];
fma.rn.f32 %f(182+i*33), %f(155+i*33), %f(181+i*33), %f(149+i*33);
ld.shared.f32 %f(183+i*33), [%r39+(75268+i*4)];
fma.rn.f32 %f(184+i*33), %f(155+i*33), %f(183+i*33), %f(151+i*33);
ld.shared.f32 %f(185+i*33), [%r39+(87812+i*4)];
fma.rn.f32 %f219, %f188, %f218, %f186;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(220+i*1), [%r(39+i*-8)+-100340];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(477+i*-2), %f221, %f(220+i*2), %f(189+i*2);
ld.shared.f32 %f(222+i*1), [%r39+(-87796+i*12544)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(473+i*-2), %f221, %f(223+i*1), %f(193+i*2);
ld.shared.f32 %f(224+i*1), [%r39+(-62708+i*12544)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(447+i*-209), %f(221+i*-99), %f(236+i*1), %f(219+i*257);
ld.shared.f32 %f(237+i*2), [%r39+(-97216+i*12544)];
for.size.2 i in range(0, 14, 1):
fma.rn.f32 %f(240+i*2), %f122, %f(239+i*2), %f(474+i*-2);
ld.shared.f32 %f(241+i*2), [%r39+(-72128+i*12544)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(268+i*2), %f(122+i*33), %f(267+i*2), %f(446+i*-208);
ld.shared.f32 %f(269+i*2), [%r39+(-97212+i*12544)];
for.size.2 i in range(0, 14, 1):
fma.rn.f32 %f(272+i*2), %f155, %f(271+i*2), %f(240+i*2);
ld.shared.f32 %f(273+i*2), [%r39+(-72124+i*12544)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(300+i*2), %f(155+i*33), %f(299+i*2), %f(268+i*2);
ld.shared.f32 %f(301+i*2), [%r39+(-97208+i*12544)];
for.size.2 i in range(0, 14, 1):
fma.rn.f32 %f(304+i*2), %f188, %f(303+i*2), %f(272+i*2);
ld.shared.f32 %f(305+i*2), [%r39+(-72120+i*12544)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(332+i*144), %f(188+i*33), %f(331+i*2), %f(300+i*2);
ld.shared.f32 %f(333+i*1), [%r39+(-97204+i*12544)];
for.size.2 i in range(0, 14, 1):
fma.rn.f32 %f(474+i*-2), %f221, %f(334+i*1), %f(304+i*2);
ld.shared.f32 %f(335+i*1), [%r39+(-72116+i*12544)];
fma.rn.f32 %f446, %f221, %f348, %f332;
for.size.1 i in range(0, 2, 1):
add.s32 %r(39+i*1), %r(39+i*1), 16;
setp.ne.s32 %p3, %r40, 103488;
@%p3 bra $L__BB0_5;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(32+i*1), %r(7+i*-6), (6+i*-5);
add.s32 %r34, %r33, %r32;
mul.wide.s32 %rd19, %r34, 4;
add.s64 %rd20, %rd3, %rd19;
ld.global.nc.f32 %f349, [%rd20];
add.f32 %f350, %f477, %f349;
mov.f32 %f351, 0f00000000;
max.f32 %f352, %f350, %f351;
add.s64 %rd21, %rd2, %rd19;
for.size.4 i in range(0, 15, 1):
st.global.f32 [%rd21+(0+i*16)], %f(352+i*3);
ld.global.nc.f32 %f(353+i*3), [%rd20+(16+i*16)];
add.f32 %f(354+i*3), %f(475+i*-2), %f(353+i*3);
max.f32 %f(355+i*3), %f(354+i*3), %f351;
for.size.4 i in range(0, 2, 1):
st.global.f32 [%rd21+(240+i*-236)], %f(397+i*3);
ld.global.nc.f32 %f(398+i*3), [%rd20+(4+i*16)];
add.f32 %f(399+i*3), %f(476+i*-2), %f(398+i*3);
max.f32 %f(400+i*3), %f(399+i*3), %f351;
for.size.4 i in range(0, 14, 1):
st.global.f32 [%rd21+(20+i*16)], %f(403+i*3);
ld.global.nc.f32 %f(404+i*3), [%rd20+(36+i*16)];
add.f32 %f(405+i*3), %f(472+i*-2), %f(404+i*3);
max.f32 %f(406+i*3), %f(405+i*3), %f351;
st.global.f32 [%rd21+244], %f445;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[50176];
for (int j_outer_inner_init = 0; j_outer_inner_init < 2; ++j_outer_inner_init) {
int index0 = j_outer_inner_init;
T_dense[index0] = 0.000000e+00f;
T_dense[index0 + 2] = 0.000000e+00f;
T_dense[index0 + 4] = 0.000000e+00f;
T_dense[index0 + 6] = 0.000000e+00f;
T_dense[index0 + 8] = 0.000000e+00f;
T_dense[index0 + 10] = 0.000000e+00f;
T_dense[index0 + 12] = 0.000000e+00f;
T_dense[index0 + 14] = 0.000000e+00f;
T_dense[index0 + 16] = 0.000000e+00f;
T_dense[index0 + 18] = 0.000000e+00f;
T_dense[index0 + 20] = 0.000000e+00f;
T_dense[index0 + 22] = 0.000000e+00f;
T_dense[index0 + 24] = 0.000000e+00f;
T_dense[index0 + 26] = 0.000000e+00f;
T_dense[index0 + 28] = 0.000000e+00f;
T_dense[index0 + 30] = 0.000000e+00f;
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 392; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index1];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 25088; ++ax0_ax1_fused_outer_outer1) {
int index2 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 50176 + ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
for (int j_outer_inner = 0; j_outer_inner < 2; ++j_outer_inner) {
for (int k_inner = 0; k_inner < 4; ++k_inner) {
int index4 = j_outer_inner;
int index5 = k_outer_inner * 4 + k_inner;
int index6 = (int)threadIdx.x * 1568 + j_outer_inner * 784 + k_outer_inner * 4 + k_inner;
T_dense[index4] = T_dense[index4] + placeholder_d_shared[index5] * placeholder_shared[index6];
T_dense[index4 + 2] = T_dense[index4 + 2] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 3136]);
T_dense[index4 + 4] = T_dense[index4 + 4] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 6272]);
T_dense[index4 + 6] = T_dense[index4 + 6] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 9408]);
T_dense[index4 + 8] = T_dense[index4 + 8] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 12544]);
T_dense[index4 + 10] = T_dense[index4 + 10] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 15680]);
T_dense[index4 + 12] = T_dense[index4 + 12] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 18816]);
T_dense[index4 + 14] = T_dense[index4 + 14] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 21952]);
T_dense[index4 + 16] = T_dense[index4 + 16] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 25088]);
T_dense[index4 + 18] = T_dense[index4 + 18] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 28224]);
T_dense[index4 + 20] = T_dense[index4 + 20] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 31360]);
T_dense[index4 + 22] = T_dense[index4 + 22] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 34496]);
T_dense[index4 + 24] = T_dense[index4 + 24] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 37632]);
T_dense[index4 + 26] = T_dense[index4 + 26] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 40768]);
T_dense[index4 + 28] = T_dense[index4 + 28] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 43904]);
T_dense[index4 + 30] = T_dense[index4 + 30] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 47040]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
for (int vthread_s = 0; vthread_s < 16; ++vthread_s) {
int index7 = (int)blockIdx.x * 64 + vthread_s * 4 + (int)threadIdx.x * 2 + ax1_inner;
int index8 = vthread_s * 2 + ax1_inner;
T_relu[index7] = max(T_dense[index8] + placeholder2[index7], 0.000000e+00f);
}
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpm2cq_yky | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<79>;
.reg .b32 %r<30>;
.reg .b64 %rd<21>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[401408];
ld.param.u64 %rd11, [default_function_kernel0_param_0];
ld.param.u64 %rd8, [default_function_kernel0_param_1];
ld.param.u64 %rd9, [default_function_kernel0_param_2];
ld.param.u64 %rd10, [default_function_kernel0_param_3];
mov.u32 %r12, %tid.x;
cvt.s64.s32 %rd1, %r12;
cvta.to.global.u64 %rd12, %rd11;
mul.wide.s32 %rd13, %r12, 4;
add.s64 %rd2, %rd12, %rd13;
ld.global.nc.f32 %f3, [%rd2];
shl.b32 %r13, %r12, 2;
mov.u32 %r14, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r1, %r14, %r13;
for.size.2 i in range(0, 5, 1):
st.shared.f32 [%r1+(0+i*512)], %f(3+i*1);
ld.global.nc.f32 %f(4+i*1), [%rd2+(512+i*512)];
st.shared.f32 [%r1+2560], %f8;
setp.gt.s32 %p1, %r12, 15;
@%p1 bra $L__BB0_2;
ld.global.nc.f32 %f9, [%rd2+3072];
st.shared.f32 [%r1+3072], %f9;
$L__BB0_2:
cvt.u32.u64 %r16, %rd1;
shl.b32 %r17, %r16, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r19, %r18, %r17;
add.s32 %r26, %r19, 4096;
cvta.to.global.u64 %rd14, %rd8;
shl.b64 %rd15, %rd1, 2;
add.s64 %rd16, %rd14, %rd15;
add.s64 %rd20, %rd16, 4096;
cvta.to.global.u64 %rd4, %rd10;
cvta.to.global.u64 %rd5, %rd9;
mov.u32 %r27, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(10+i*1), [%rd20+(-4096+i*512)];
st.shared.f32 [%r26+(-4096+i*512)], %f(10+i*1);
add.s32 %r26, %r26, 8192;
add.s64 %rd20, %rd20, 8192;
add.s32 %r27, %r27, 16;
setp.ne.s32 %p2, %r27, 784;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r23, %r16, 3136, %r18;
add.s32 %r28, %r23, 32;
mov.u32 %r29, 32;
mov.f32 %f78, 0f00000000;
$L__BB0_5:
ld.shared.f32 %f27, [%r28+-32];
add.s32 %r25, %r14, %r29;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(28+i*3), [%r25+(-32+i*4)];
fma.rn.f32 %f(29+i*3), %f(28+i*3), %f(27+i*3), %f(78+i*-49);
ld.shared.f32 %f(30+i*3), [%r28+(-28+i*4)];
for.size.3 i in range(0, 13, 1):
ld.shared.f32 %f(34+i*3), [%r25+(-24+i*4)];
fma.rn.f32 %f(35+i*3), %f(34+i*3), %f(33+i*3), %f(32+i*3);
ld.shared.f32 %f(36+i*3), [%r28+(-20+i*4)];
ld.shared.f32 %f73, [%r25+28];
fma.rn.f32 %f78, %f73, %f72, %f71;
for.size.1 i in range(0, 2, 1):
add.s32 %r(28+i*1), %r(28+i*1), 64;
setp.ne.s32 %p3, %r29, 3168;
@%p3 bra $L__BB0_5;
add.s64 %rd18, %rd4, %rd15;
ld.global.nc.f32 %f74, [%rd18];
add.f32 %f75, %f78, %f74;
mov.f32 %f76, 0f00000000;
max.f32 %f77, %f75, %f76;
add.s64 %rd19, %rd5, %rd15;
st.global.f32 [%rd19], %f77;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[100352];
T_dense[0] = 0.000000e+00f;
int index0 = (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index0];
placeholder_d_shared[index0 + 128] = placeholder[index0 + 128];
placeholder_d_shared[index0 + 256] = placeholder[index0 + 256];
placeholder_d_shared[index0 + 384] = placeholder[index0 + 384];
placeholder_d_shared[index0 + 512] = placeholder[index0 + 512];
placeholder_d_shared[index0 + 640] = placeholder[index0 + 640];
if ((int)threadIdx.x < 16) {
placeholder_d_shared[index0 + 768] = placeholder[index0 + 768];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 784; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 128 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index1];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 784; ++k_outer_inner) {
int index2 = k_outer_inner;
int index3 = (int)threadIdx.x * 784 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
}
T_relu[index0] = max(T_dense[0] + placeholder2[index0], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp5utetogw | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<580>;
.reg .b32 %r<22>;
.reg .b64 %rd<20>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[4];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[512];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd3, %rd5;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 784;
shl.b32 %r11, %r1, 2;
mov.u32 %r12, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r12, %r11;
mul.lo.s32 %r4, %r1, 1568;
shl.b32 %r13, %r1, 3;
mov.u32 %r14, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r6, %r14, %r13;
shl.b32 %r15, %r1, 8;
add.s32 %r7, %r14, %r15;
mov.u32 %r21, 0;
mov.f32 %f516, 0f00000000;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(517+i*1), %f516;
$L__BB0_1:
mov.u32 %r20, %tid.x;
bar.sync 0;
setp.gt.s32 %p1, %r20, 0;
@%p1 bra $L__BB0_3;
add.s32 %r16, %r21, %r2;
mul.wide.s32 %rd9, %r16, 4;
add.s64 %rd10, %rd3, %rd9;
ld.global.nc.f32 %f193, [%rd10];
st.shared.f32 [%r3], %f193;
$L__BB0_3:
add.s32 %r17, %r21, %r4;
mul.wide.s32 %rd11, %r17, 4;
add.s64 %rd12, %rd1, %rd11;
for.size.1 i in range(0, 2, 1):
ld.global.nc.f32 %f(194+i*1), [%rd12+(3136+i*-3136)];
for.size.3 i in range(0, 31, 1):
st.shared.v2.f32 [%r6+(0+i*16)], {%f(195+i*2), %f(194+i*2)};
ld.global.nc.f32 %f(196+i*2), [%rd12+(15680+i*12544)];
ld.global.nc.f32 %f(197+i*2), [%rd12+(12544+i*12544)];
st.shared.v2.f32 [%r6+496], {%f257, %f256};
bar.sync 0;
ld.shared.f32 %f258, [%r7];
ld.shared.f32 %f259, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(579+i*-1), %f259, %f(258+i*2), %f(579+i*-1);
ld.shared.f32 %f(260+i*1), [%r7+(4+i*4)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(577+i*-1), %f259, %f(261+i*1), %f(577+i*-1);
ld.shared.f32 %f(262+i*1), [%r7+(12+i*4)];
fma.rn.f32 %f516, %f259, %f322, %f516;
add.s32 %r21, %r21, 1;
setp.lt.u32 %p2, %r21, 784;
@%p2 bra $L__BB0_1;
ld.param.u64 %rd19, [default_function_kernel0_param_2];
cvta.to.global.u64 %rd18, %rd19;
ld.param.u64 %rd17, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd16, %rd17;
mov.u32 %r19, %tid.x;
shl.b32 %r18, %r19, 6;
mul.wide.s32 %rd13, %r18, 4;
add.s64 %rd14, %rd16, %rd13;
ld.global.nc.f32 %f323, [%rd14];
add.f32 %f324, %f579, %f323;
mov.f32 %f325, 0f00000000;
max.f32 %f326, %f324, %f325;
add.s64 %rd15, %rd18, %rd13;
for.size.4 i in range(0, 63, 1):
st.global.f32 [%rd15+(0+i*4)], %f(326+i*3);
ld.global.nc.f32 %f(327+i*3), [%rd14+(4+i*4)];
add.f32 %f(328+i*3), %f(578+i*-1), %f(327+i*3);
max.f32 %f(329+i*3), %f(328+i*3), %f325;
st.global.f32 [%rd15+252], %f515;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[128];
for (int j_inner_init = 0; j_inner_init < 64; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
int index1 = (int)threadIdx.x;
int index2 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 32; ++ax0_ax1_fused_outer_outer) {
int2 _1 = make_int2(ax0_ax1_fused_outer_outer * 3136 + (int)threadIdx.x * 1568 + k_outer_outer+784*0, ax0_ax1_fused_outer_outer * 3136 + (int)threadIdx.x * 1568 + k_outer_outer+784*1);
int index3 = _1.x;
int index4 = _1.y;
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x * 2))[0] = make_float2(placeholder1[index3],placeholder1[index4]);
}
__syncthreads();
for (int j_inner = 0; j_inner < 64; ++j_inner) {
int index5 = j_inner;
int index6 = (int)threadIdx.x * 64 + j_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[0] * placeholder_shared[index6];
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index7 = (int)threadIdx.x * 64 + ax1_inner;
int index8 = ax1_inner;
T_relu[index7] = max(T_dense[index8] + placeholder2[index7], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpggjpq1o7 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<182>;
.reg .b32 %r<37>;
.reg .b64 %rd<25>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[401408];
ld.param.u64 %rd12, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd13, [default_function_kernel0_param_2];
ld.param.u64 %rd11, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd14, %rd12;
cvta.to.global.u64 %rd1, %rd13;
mov.u32 %r15, %tid.x;
cvt.s64.s32 %rd2, %r15;
mul.wide.s32 %rd15, %r15, 4;
add.s64 %rd16, %rd14, %rd15;
add.s64 %rd23, %rd16, 32;
shl.b32 %r16, %r15, 2;
mov.u32 %r17, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r18, %r17, %r16;
add.s32 %r32, %r18, 32;
mov.u32 %r33, 0;
$L__BB0_1:
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(33+i*1), [%rd23+(-32+i*32)];
st.shared.f32 [%r32+(-32+i*32)], %f(33+i*1);
add.s64 %rd23, %rd23, 64;
for.size.1 i in range(0, 2, 1):
add.s32 %r(32+i*1), %r(32+i*1), (64+i*-62);
setp.ne.s32 %p1, %r33, 98;
@%p1 bra $L__BB0_1;
cvt.u32.u64 %r20, %rd2;
cvta.to.global.u64 %rd6, %rd11;
shl.b32 %r21, %r20, 2;
mov.u32 %r22, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r23, %r22, %r21;
add.s32 %r34, %r23, 256;
cvta.to.global.u64 %rd17, %rd10;
shl.b64 %rd18, %rd2, 2;
add.s64 %rd19, %rd17, %rd18;
add.s64 %rd24, %rd19, 256;
mov.u32 %r35, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(35+i*1), [%rd24+(-256+i*32)];
st.shared.f32 [%r34+(-256+i*32)], %f(35+i*1);
add.s32 %r34, %r34, 512;
add.s64 %rd24, %rd24, 512;
add.s32 %r35, %r35, 16;
setp.ne.s32 %p2, %r35, 12544;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r11, %r20, 25088, %r22;
mov.f32 %f166, 0f00000000;
mov.u32 %r36, 200704;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(167+i*1), %f166;
$L__BB0_5:
for.size.1 i in range(0, 2, 1):
add.s32 %r(28+i*1), %r(17+i*-6), %r36;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(67+i*1), [%r(29+i*-1)+-200704];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(69+i*2), %f68, %f(67+i*3), %f(181+i*-1);
ld.shared.f32 %f(70+i*2), [%r29+(0+i*-200700)];
ld.shared.f32 %f73, [%r28+-200700];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(181+i*-1), %f73, %f(72+i*2), %f(69+i*2);
ld.shared.f32 %f(74+i*1), [%r29+(4+i*-197572)];
for.size.6 i in range(0, 6, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(76+i*6+j*2), %f68, %f(75+i*6+j*2), %f(179+i*-2+j*-1);
ld.shared.f32 %f(77+i*6+j*2), [%r29+(3136+i*3136+j*-200700)];
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(179+i*-2+j*-1), %f73, %f(79+i*6+j*1), %f(76+i*6+j*2);
ld.shared.f32 %f(80+i*6+j*1), [%r29+(3140+i*3136+j*-197572)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(112+i*2), %f68, %f(111+i*2), %f(167+i*-1);
ld.shared.f32 %f(113+i*2), [%r29+(21952+i*-200700)];
fma.rn.f32 %f167, %f73, %f115, %f112;
ld.shared.f32 %f116, [%r29+21956];
fma.rn.f32 %f166, %f73, %f116, %f114;
add.s32 %r36, %r36, 8;
setp.ne.s32 %p3, %r36, 203840;
@%p3 bra $L__BB0_5;
shl.b32 %r31, %r20, 3;
mul.wide.s32 %rd20, %r31, 4;
add.s64 %rd21, %rd6, %rd20;
ld.global.nc.f32 %f117, [%rd21];
add.f32 %f118, %f181, %f117;
mov.f32 %f119, 0f00000000;
max.f32 %f120, %f118, %f119;
add.s64 %rd22, %rd1, %rd20;
for.size.5 i in range(0, 7, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd22+(0+i*4+j*256)], %f(120+i*6+j*3);
ld.global.nc.f32 %f(121+i*6+j*3), [%rd21+(256+i*4+j*-252)];
add.f32 %f(122+i*6+j*3), %f(180+i*-2+j*-1), %f(121+i*6+j*3);
max.f32 %f(123+i*6+j*3), %f(122+i*6+j*3), %f119;
st.global.f32 [%rd22+28], %f162;
ld.global.nc.f32 %f163, [%rd21+284];
add.f32 %f164, %f166, %f163;
max.f32 %f165, %f164, %f119;
st.global.f32 [%rd22+284], %f165;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[100352];
T_dense[0] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
int index0 = ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index0];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 12544; ++ax0_ax1_fused_outer_outer1) {
int index1 = ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index1];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 392; ++k_outer_inner) {
int index2 = k_outer_inner * 2;
int index3 = (int)threadIdx.x * 6272 + k_outer_inner * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[8] = T_dense[8] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 50176]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 1]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 50177]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 784]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 50960]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 785]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 50961]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 1568]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 51744]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 1569]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 51745]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 2352]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 52528]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 2353]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 52529]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 3136]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 53312]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 3137]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 53313]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 3920]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 54096]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 3921]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 54097]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 4704]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 54880]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 4705]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 54881]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 5488]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 55664]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 5489]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 55665]);
}
for (int ax1_inner = 0; ax1_inner < 8; ++ax1_inner) {
int index4 = (int)threadIdx.x * 8 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[index5 + 8] + placeholder2[index4 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp7vrzxud4 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<82>;
.reg .b32 %r<118>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[448];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[57344];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd7, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd5;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd7;
mov.u32 %r1, %tid.x;
shl.b32 %r2, %r1, 4;
mul.lo.s32 %r3, %r1, 224;
shl.b32 %r19, %r1, 6;
mov.u32 %r20, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r4, %r20, %r19;
add.s32 %r5, %r1, 448;
shl.b32 %r21, %r1, 2;
mov.u32 %r22, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r6, %r22, %r21;
mov.u32 %r113, 0;
mov.f32 %f78, 0f00000000;
mov.f32 %f79, %f78;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r113, 112;
setp.gt.s32 %p1, %r1, 6;
@%p1 bra $L__BB0_3;
add.s32 %r23, %r8, %r2;
mul.wide.s32 %rd8, %r23, 4;
add.s64 %rd9, %rd1, %rd8;
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(9+i*1), [%rd9+(0+i*4)];
st.shared.f32 [%r4+(0+i*4)], %f(9+i*1);
$L__BB0_3:
mov.u32 %r116, 1024;
mov.u32 %r114, %r1;
mov.u32 %r115, %r5;
$L__BB0_4:
mul.hi.s32 %r25, %r114, -1840700269;
add.s32 %r26, %r25, %r114;
shr.u32 %r27, %r26, 31;
shr.s32 %r28, %r26, 6;
add.s32 %r29, %r28, %r27;
mul.lo.s32 %r30, %r29, 112;
sub.s32 %r31, %r114, %r30;
add.s32 %r32, %r31, %r8;
mad.lo.s32 %r33, %r29, 784, %r32;
mul.wide.s32 %rd10, %r33, 4;
add.s64 %rd11, %rd2, %rd10;
ld.global.nc.f32 %f25, [%rd11];
add.s32 %r34, %r6, %r116;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r34+(-1024+i*256)], %f(25+i*1);
add.s32 %r(35+i*10), %r114, (64+i*64);
mul.hi.s32 %r(36+i*10), %r(35+i*10), -1840700269;
add.s32 %r(37+i*10), %r(36+i*10), %r(35+i*10);
shr.u32 %r(38+i*10), %r(37+i*10), 31;
shr.s32 %r(39+i*10), %r(37+i*10), 6;
add.s32 %r(40+i*10), %r(39+i*10), %r(38+i*10);
mul.lo.s32 %r(41+i*10), %r(40+i*10), 112;
sub.s32 %r(42+i*10), %r(35+i*10), %r(41+i*10);
add.s32 %r(43+i*10), %r(42+i*10), %r8;
mad.lo.s32 %r(44+i*10), %r(40+i*10), 784, %r(43+i*10);
mul.wide.s32 %rd(12+i*2), %r(44+i*10), 4;
add.s64 %rd(13+i*2), %rd2, %rd(12+i*2);
ld.global.nc.f32 %f(26+i*1), [%rd(13+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r34+(0+i*256)], %f(29+i*1);
add.s32 %r(75+i*10), %r115, (-128+i*64);
mul.hi.s32 %r(76+i*10), %r(75+i*10), -1840700269;
add.s32 %r(77+i*10), %r(76+i*10), %r(75+i*10);
shr.u32 %r(78+i*10), %r(77+i*10), 31;
shr.s32 %r(79+i*10), %r(77+i*10), 6;
add.s32 %r(80+i*10), %r(79+i*10), %r(78+i*10);
mul.lo.s32 %r(81+i*10), %r(80+i*10), 112;
sub.s32 %r(82+i*10), %r(75+i*10), %r(81+i*10);
add.s32 %r(83+i*10), %r(82+i*10), %r8;
mad.lo.s32 %r(84+i*10), %r(80+i*10), 784, %r(83+i*10);
mul.wide.s32 %rd(20+i*2), %r(84+i*10), 4;
add.s64 %rd(21+i*2), %rd2, %rd(20+i*2);
ld.global.nc.f32 %f(30+i*1), [%rd(21+i*2)+0];
st.shared.f32 [%r34+512], %f31;
mul.hi.s32 %r95, %r115, -1840700269;
add.s32 %r96, %r95, %r115;
shr.u32 %r97, %r96, 31;
shr.s32 %r98, %r96, 6;
add.s32 %r99, %r98, %r97;
mul.lo.s32 %r100, %r99, 112;
sub.s32 %r101, %r115, %r100;
add.s32 %r102, %r101, %r8;
mad.lo.s32 %r103, %r99, 784, %r102;
mul.wide.s32 %rd24, %r103, 4;
add.s64 %rd25, %rd2, %rd24;
ld.global.nc.f32 %f32, [%rd25];
st.shared.f32 [%r34+768], %f32;
for.size.1 i in range(0, 2, 1):
add.s32 %r(115+i*-1), %r(115+i*-1), 512;
add.s32 %r116, %r116, 2048;
setp.ne.s32 %p2, %r116, 58368;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r117, 0;
$L__BB0_6:
add.s32 %r105, %r117, %r3;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(106+i*3), %r(105+i*12), 2;
add.s32 %r(108+i*3), %r(22+i*-2), %r(106+i*3);
ld.shared.f32 %f(33+i*1), [%r(108+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(35+i*2), %f34, %f(33+i*3), %f(79+i*-1);
ld.shared.f32 %f(36+i*2), [%r108+(448+i*-444)];
for.size.5 i in range(0, 6, 1):
ld.shared.f32 %f(39+i*5), [%r111+(4+i*4)];
fma.rn.f32 %f(40+i*5), %f(39+i*5), %f(38+i*5), %f(35+i*5);
ld.shared.f32 %f(41+i*5), [%r108+(452+i*4)];
fma.rn.f32 %f(42+i*5), %f(39+i*5), %f(41+i*5), %f(37+i*5);
ld.shared.f32 %f(43+i*5), [%r108+(8+i*4)];
ld.shared.f32 %f69, [%r111+28];
fma.rn.f32 %f79, %f69, %f68, %f65;
ld.shared.f32 %f70, [%r108+476];
fma.rn.f32 %f78, %f69, %f70, %f67;
add.s32 %r117, %r117, 8;
setp.ne.s32 %p3, %r117, 112;
@%p3 bra $L__BB0_6;
add.s32 %r113, %r113, 1;
setp.lt.u32 %p4, %r113, 7;
@%p4 bra $L__BB0_1;
shl.b32 %r112, %r1, 1;
mul.wide.s32 %rd26, %r112, 4;
add.s64 %rd27, %rd3, %rd26;
ld.global.nc.f32 %f71, [%rd27];
add.f32 %f72, %f79, %f71;
mov.f32 %f73, 0f00000000;
max.f32 %f74, %f72, %f73;
cvta.to.global.u64 %rd28, %rd4;
add.s64 %rd29, %rd28, %rd26;
st.global.f32 [%rd29], %f74;
ld.global.nc.f32 %f75, [%rd27+4];
add.f32 %f76, %f78, %f75;
max.f32 %f77, %f76, %f73;
st.global.f32 [%rd29+4], %f77;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[112];
__shared__ float placeholder_shared[14336];
for (int j_inner_init = 0; j_inner_init < 2; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 7; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_inner_s = 0; ax0_ax1_fused_inner_s < 16; ++ax0_ax1_fused_inner_s) {
if ((int)threadIdx.x * 16 + ax0_ax1_fused_inner_s < 112) {
if ((int)threadIdx.x < 7) {
int index1 = (int)threadIdx.x * 16 + ax0_ax1_fused_inner_s;
int index2 = k_outer_outer * 112 + (int)threadIdx.x * 16 + ax0_ax1_fused_inner_s;
placeholder_d_shared[index1] = placeholder[index2];
}
}
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 224; ++ax0_ax1_fused_outer_outer) {
int index3 = ax0_ax1_fused_outer_outer * 64 + (int)threadIdx.x;
int index4 = (ax0_ax1_fused_outer_outer * 64 + (int)threadIdx.x) / 112 * 784 + k_outer_outer * 112 + (ax0_ax1_fused_outer_outer * 64 + (int)threadIdx.x) % 112;
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 112; ++k_outer_inner) {
for (int j_inner = 0; j_inner < 2; ++j_inner) {
int index5 = j_inner;
int index6 = k_outer_inner;
int index7 = (int)threadIdx.x * 224 + j_inner * 112 + k_outer_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index8 = (int)threadIdx.x * 2 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpw7ftouhp | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<86>;
.reg .b32 %r<28>;
.reg .b64 %rd<21>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[401408];
ld.param.u64 %rd11, [default_function_kernel0_param_0];
ld.param.u64 %rd8, [default_function_kernel0_param_1];
ld.param.u64 %rd9, [default_function_kernel0_param_2];
ld.param.u64 %rd10, [default_function_kernel0_param_3];
mov.u32 %r1, %tid.x;
shl.b32 %r13, %r1, 3;
cvta.to.global.u64 %rd12, %rd11;
mul.wide.s32 %rd13, %r13, 4;
add.s64 %rd1, %rd12, %rd13;
ld.global.nc.f32 %f5, [%rd1];
shl.b32 %r14, %r1, 5;
mov.u32 %r26, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r26, %r14;
for.size.2 i in range(0, 7, 1):
st.shared.f32 [%r2+(0+i*4)], %f(5+i*1);
ld.global.nc.f32 %f(6+i*1), [%rd1+(4+i*4)];
st.shared.f32 [%r2+28], %f12;
setp.gt.s32 %p1, %r1, 33;
@%p1 bra $L__BB0_2;
for.size.2 i in range(0, 8, 1):
ld.global.nc.f32 %f(13+i*1), [%rd1+(2048+i*4)];
st.shared.f32 [%r2+(2048+i*4)], %f(13+i*1);
$L__BB0_2:
shl.b32 %r17, %r1, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r19, %r18, %r17;
add.s32 %r24, %r19, 2048;
cvt.s64.s32 %rd2, %r1;
cvta.to.global.u64 %rd14, %rd8;
mul.wide.s32 %rd15, %r1, 4;
add.s64 %rd16, %rd14, %rd15;
add.s64 %rd20, %rd16, 2048;
cvta.to.global.u64 %rd4, %rd10;
cvta.to.global.u64 %rd5, %rd9;
mov.u32 %r25, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(21+i*1), [%rd20+(-2048+i*256)];
st.shared.f32 [%r24+(-2048+i*256)], %f(21+i*1);
add.s32 %r24, %r24, 4096;
add.s64 %rd20, %rd20, 4096;
add.s32 %r25, %r25, 16;
setp.ne.s32 %p2, %r25, 1568;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r8, %r1, 3136, %r18;
mov.f32 %f84, 0f00000000;
mov.u32 %r27, 200704;
mov.f32 %f85, %f84;
$L__BB0_5:
add.s32 %r23, %r8, %r27;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(39+i*5), [%r23+(-200704+i*4)];
ld.shared.f32 %f(40+i*5), [%r26+(0+i*4)];
fma.rn.f32 %f(41+i*5), %f(40+i*5), %f(39+i*5), %f(84+i*-43);
ld.shared.f32 %f(42+i*5), [%r23+(0+i*4)];
fma.rn.f32 %f(43+i*5), %f(40+i*5), %f(42+i*5), %f(85+i*-42);
for.size.5 i in range(0, 5, 1):
ld.shared.f32 %f(49+i*5), [%r23+(-200696+i*4)];
ld.shared.f32 %f(50+i*5), [%r26+(8+i*4)];
fma.rn.f32 %f(51+i*5), %f(50+i*5), %f(49+i*5), %f(46+i*5);
ld.shared.f32 %f(52+i*5), [%r23+(8+i*4)];
fma.rn.f32 %f(53+i*5), %f(50+i*5), %f(52+i*5), %f(48+i*5);
ld.shared.f32 %f74, [%r23+-200676];
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(75+i*1), [%r(26+i*-3)+28];
fma.rn.f32 %f(84+i*1), %f75, %f(74+i*2), %f(71+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(26+i*1), %r(26+i*1), 32;
setp.ne.s32 %p3, %r27, 203840;
@%p3 bra $L__BB0_5;
shl.b64 %rd17, %rd2, 2;
add.s64 %rd18, %rd4, %rd17;
ld.global.nc.f32 %f77, [%rd18];
add.f32 %f78, %f84, %f77;
mov.f32 %f79, 0f00000000;
max.f32 %f80, %f78, %f79;
add.s64 %rd19, %rd5, %rd17;
st.global.f32 [%rd19], %f80;
ld.global.nc.f32 %f81, [%rd18+256];
add.f32 %f82, %f85, %f81;
max.f32 %f83, %f82, %f79;
st.global.f32 [%rd19+256], %f83;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[100352];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
int index0 = (int)threadIdx.x * 8;
placeholder_d_shared[index0] = placeholder[index0];
placeholder_d_shared[index0 + 1] = placeholder[index0 + 1];
placeholder_d_shared[index0 + 2] = placeholder[index0 + 2];
placeholder_d_shared[index0 + 3] = placeholder[index0 + 3];
placeholder_d_shared[index0 + 4] = placeholder[index0 + 4];
placeholder_d_shared[index0 + 5] = placeholder[index0 + 5];
placeholder_d_shared[index0 + 6] = placeholder[index0 + 6];
placeholder_d_shared[index0 + 7] = placeholder[index0 + 7];
if ((int)threadIdx.x < 34) {
placeholder_d_shared[index0 + 512] = placeholder[index0 + 512];
}
if ((int)threadIdx.x < 34) {
placeholder_d_shared[index0 + 513] = placeholder[index0 + 513];
}
if ((int)threadIdx.x < 34) {
placeholder_d_shared[index0 + 514] = placeholder[index0 + 514];
}
if ((int)threadIdx.x < 34) {
placeholder_d_shared[index0 + 515] = placeholder[index0 + 515];
}
if ((int)threadIdx.x < 34) {
placeholder_d_shared[index0 + 516] = placeholder[index0 + 516];
}
if ((int)threadIdx.x < 34) {
placeholder_d_shared[index0 + 517] = placeholder[index0 + 517];
}
if ((int)threadIdx.x < 34) {
placeholder_d_shared[index0 + 518] = placeholder[index0 + 518];
}
if ((int)threadIdx.x < 34) {
placeholder_d_shared[index0 + 519] = placeholder[index0 + 519];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1568; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 64 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index1];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 98; ++k_outer_inner) {
int index2 = k_outer_inner * 8;
int index3 = (int)threadIdx.x * 784 + k_outer_inner * 8;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 50176]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 50177]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 2]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 50178]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 3]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 50179]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 4] * placeholder_shared[index3 + 4]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 4] * placeholder_shared[index3 + 50180]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 5] * placeholder_shared[index3 + 5]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 5] * placeholder_shared[index3 + 50181]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 6] * placeholder_shared[index3 + 6]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 6] * placeholder_shared[index3 + 50182]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 7] * placeholder_shared[index3 + 7]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 7] * placeholder_shared[index3 + 50183]);
}
int index4 = (int)threadIdx.x;
T_relu[index4] = max(T_dense[0] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[1] + placeholder2[index4 + 64], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpejxpp24h | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<29>;
.reg .f32 %f<212>;
.reg .b32 %r<133>;
.reg .b64 %rd<34>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[1568];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd8, [default_function_kernel0_param_0];
ld.param.u64 %rd9, [default_function_kernel0_param_1];
ld.param.u64 %rd10, [default_function_kernel0_param_2];
ld.param.u64 %rd11, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd8;
cvta.to.global.u64 %rd1, %rd10;
cvta.to.global.u64 %rd2, %rd9;
cvta.to.global.u64 %rd3, %rd11;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 6272;
add.s32 %r3, %r1, 24;
shl.b32 %r4, %r1, 1;
add.s32 %r5, %r1, 8;
add.s64 %rd4, %rd12, 128;
add.s32 %r6, %r4, 48;
shl.b32 %r30, %r1, 3;
mov.u32 %r31, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r32, %r31, %r30;
add.s32 %r7, %r32, 128;
mov.u32 %r29, 0;
mov.f32 %f180, 0f00000000;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(181+i*1), %f180;
mov.u32 %r126, %r29;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r10, %r126, 392;
add.s32 %r34, %r4, %r10;
mul.wide.s32 %rd13, %r34, 4;
add.s64 %rd33, %rd4, %rd13;
mov.u32 %r127, %r7;
mov.u32 %r128, %r6;
mov.u32 %r129, %r29;
mov.u32 %r130, %r3;
bra.uni $L__BB0_2;
$L__BB0_19:
for.size.1 i in range(0, 2, 1):
add.s32 %r(130+i*-1), %r(130+i*-1), 32;
add.s64 %rd33, %rd33, 256;
for.size.1 i in range(0, 2, 1):
add.s32 %r(128+i*-1), %r(128+i*-1), (64+i*192);
$L__BB0_2:
for.size.2 i in range(0, 2, 1):
add.s32 %r(35+i*1), %r(130+i*-2), (-24+i*-24);
setp.gt.s32 %p(1+i*1), %r(35+i*1), (195+i*196);
or.pred %p3, %p2, %p1;
@%p3 bra $L__BB0_4;
ld.global.nc.f32 %f65, [%rd33+-128];
st.shared.f32 [%r127+-128], %f65;
$L__BB0_4:
setp.gt.s32 %p5, %r36, 390;
or.pred %p6, %p5, %p1;
@%p6 bra $L__BB0_6;
ld.global.nc.f32 %f66, [%rd33+-124];
st.shared.f32 [%r127+-124], %f66;
$L__BB0_6:
setp.eq.s32 %p7, %r129, 192;
@%p7 bra $L__BB0_20;
add.s32 %r15, %r128, -32;
add.s32 %r16, %r5, %r129;
for.size.1 i in range(0, 2, 1):
setp.gt.s32 %p(8+i*1), %r(16+i*-1), (195+i*196);
or.pred %p10, %p9, %p8;
@%p10 bra $L__BB0_9;
ld.global.nc.f32 %f67, [%rd33+-64];
st.shared.f32 [%r127+-64], %f67;
$L__BB0_9:
setp.gt.s32 %p12, %r15, 390;
or.pred %p13, %p12, %p8;
@%p13 bra $L__BB0_11;
ld.global.nc.f32 %f68, [%rd33+-60];
st.shared.f32 [%r127+-60], %f68;
$L__BB0_11:
for.size.1 i in range(0, 2, 1):
add.s32 %r(17+i*1), %r(128+i*2), (-16+i*8);
for.size.1 i in range(0, 2, 1):
setp.gt.s32 %p(14+i*1), %r(18+i*-1), (195+i*196);
or.pred %p16, %p15, %p14;
@%p16 bra $L__BB0_13;
ld.global.nc.f32 %f69, [%rd33];
st.shared.f32 [%r127], %f69;
$L__BB0_13:
setp.gt.s32 %p18, %r17, 390;
or.pred %p19, %p18, %p14;
@%p19 bra $L__BB0_15;
ld.global.nc.f32 %f70, [%rd33+4];
st.shared.f32 [%r127+4], %f70;
$L__BB0_15:
for.size.1 i in range(0, 2, 1):
setp.gt.s32 %p(20+i*1), %r(128+i*2), (391+i*-196);
or.pred %p22, %p20, %p21;
@%p22 bra $L__BB0_17;
ld.global.nc.f32 %f71, [%rd33+64];
st.shared.f32 [%r127+64], %f71;
$L__BB0_17:
setp.gt.s32 %p24, %r128, 390;
or.pred %p25, %p24, %p21;
@%p25 bra $L__BB0_19;
ld.global.nc.f32 %f72, [%rd33+68];
st.shared.f32 [%r127+68], %f72;
bra.uni $L__BB0_19;
$L__BB0_20:
mov.u32 %r131, 0;
$L__BB0_21:
shl.b32 %r40, %r131, 3;
add.s32 %r41, %r40, %r1;
mul.hi.s32 %r42, %r41, 1402438301;
shr.u32 %r43, %r42, 31;
shr.s32 %r44, %r42, 7;
add.s32 %r45, %r44, %r43;
mul.lo.s32 %r46, %r45, 392;
sub.s32 %r47, %r41, %r46;
add.s32 %r48, %r47, %r10;
mad.lo.s32 %r49, %r45, 784, %r48;
mul.wide.s32 %rd14, %r49, 4;
add.s64 %rd15, %rd2, %rd14;
ld.global.nc.f32 %f73, [%rd15];
shl.b32 %r50, %r41, 2;
mov.u32 %r51, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r52, %r51, %r50;
for.size.13 i in range(0, 7, 1):
st.shared.f32 [%r52+(0+i*32)], %f(73+i*1);
add.s32 %r(53+i*9), %r41, (8+i*8);
mul.hi.s32 %r(54+i*9), %r(53+i*9), 1402438301;
shr.u32 %r(55+i*9), %r(54+i*9), 31;
shr.s32 %r(56+i*9), %r(54+i*9), 7;
add.s32 %r(57+i*9), %r(56+i*9), %r(55+i*9);
mul.lo.s32 %r(58+i*9), %r(57+i*9), 392;
sub.s32 %r(59+i*9), %r(53+i*9), %r(58+i*9);
add.s32 %r(60+i*9), %r(59+i*9), %r10;
mad.lo.s32 %r(61+i*9), %r(57+i*9), 784, %r(60+i*9);
mul.wide.s32 %rd(16+i*2), %r(61+i*9), 4;
add.s64 %rd(17+i*2), %rd2, %rd(16+i*2);
ld.global.nc.f32 %f(74+i*1), [%rd(17+i*2)+0];
st.shared.f32 [%r52+224], %f80;
add.s32 %r131, %r131, 8;
setp.ne.s32 %p26, %r131, 6272;
@%p26 bra $L__BB0_21;
bar.sync 0;
mov.u32 %r132, 0;
$L__BB0_23:
for.size.2 i in range(0, 2, 1):
shl.b32 %r(117+i*2), %r132, (1+i*2);
add.s32 %r(118+i*3), %r(117+i*-86), %r(2+i*117);
shl.b32 %r122, %r118, 2;
add.s32 %r124, %r51, %r122;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(81+i*1), [%r(124+i*-3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(83+i*2), %f82, %f(81+i*3), %f(195+i*-1);
ld.shared.f32 %f(84+i*2), [%r124+(1568+i*1568)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(87+i*2), %f82, %f(86+i*2), %f(193+i*-1);
ld.shared.f32 %f(88+i*2), [%r124+(4704+i*1568)];
fma.rn.f32 %f113, %f82, %f112, %f180;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(114+i*1), [%r(124+i*-3)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(195+i*-1), %f115, %f(114+i*2), %f(83+i*2);
ld.shared.f32 %f(116+i*1), [%r124+(1572+i*1568)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(193+i*-1), %f115, %f(117+i*1), %f(87+i*2);
ld.shared.f32 %f(118+i*1), [%r124+(4708+i*1568)];
fma.rn.f32 %f180, %f115, %f130, %f113;
add.s32 %r132, %r132, 1;
setp.ne.s32 %p27, %r132, 196;
@%p27 bra $L__BB0_23;
add.s32 %r126, %r126, 1;
setp.lt.u32 %p28, %r126, 2;
@%p28 bra $L__BB0_1;
shl.b32 %r125, %r1, 4;
mul.wide.s32 %rd30, %r125, 4;
add.s64 %rd31, %rd3, %rd30;
ld.global.nc.f32 %f131, [%rd31];
add.f32 %f132, %f195, %f131;
mov.f32 %f133, 0f00000000;
max.f32 %f134, %f132, %f133;
add.s64 %rd32, %rd1, %rd30;
for.size.4 i in range(0, 15, 1):
st.global.f32 [%rd32+(0+i*4)], %f(134+i*3);
ld.global.nc.f32 %f(135+i*3), [%rd31+(4+i*4)];
add.f32 %f(136+i*3), %f(194+i*-1), %f(135+i*3);
max.f32 %f(137+i*3), %f(136+i*3), %f133;
st.global.f32 [%rd32+60], %f179;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[392];
__shared__ float placeholder_shared[50176];
for (int j_inner_init = 0; j_inner_init < 16; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 2; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 25; ++ax0_ax1_fused_outer_outer) {
for (int ax0_ax1_fused_inner_s = 0; ax0_ax1_fused_inner_s < 2; ++ax0_ax1_fused_inner_s) {
if (ax0_ax1_fused_outer_outer * 16 + (int)threadIdx.x * 2 + ax0_ax1_fused_inner_s < 392) {
if (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x < 196) {
int index1 = ax0_ax1_fused_outer_outer * 16 + (int)threadIdx.x * 2 + ax0_ax1_fused_inner_s;
int index2 = k_outer_outer * 392 + ax0_ax1_fused_outer_outer * 16 + (int)threadIdx.x * 2 + ax0_ax1_fused_inner_s;
placeholder_d_shared[index1] = placeholder[index2];
}
}
}
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 6272; ++ax0_ax1_fused_outer_outer1) {
int index3 = ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x;
int index4 = (ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x) / 392 * 784 + k_outer_outer * 392 + (ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x) % 392;
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 2; ++k_inner) {
for (int j_inner = 0; j_inner < 16; ++j_inner) {
int index5 = j_inner;
int index6 = k_outer_inner * 2 + k_inner;
int index7 = (int)threadIdx.x * 6272 + j_inner * 392 + k_outer_inner * 2 + k_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
}
for (int ax1_inner = 0; ax1_inner < 16; ++ax1_inner) {
int index8 = (int)threadIdx.x * 16 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmplj6xzgtb | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<128>;
.reg .b32 %r<115>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[1568];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd3, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd5, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 1568;
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r19, %r18;
mov.u32 %r110, 0;
mov.f32 %f120, 0f00000000;
cvta.to.global.u64 %rd7, %rd3;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(121+i*1), %f120;
$L__BB0_1:
setp.gt.s32 %p1, %r1, 23;
bar.sync 0;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(21+i*-16), %r(1+i*109), (7+i*385);
add.s32 %r22, %r5, %r21;
mul.wide.s32 %rd8, %r22, 4;
add.s64 %rd2, %rd7, %rd8;
ld.global.nc.f32 %f17, [%rd2];
shl.b32 %r23, %r21, 2;
mov.u32 %r24, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r25, %r24, %r23;
for.size.2 i in range(0, 6, 1):
st.shared.f32 [%r25+(0+i*4)], %f(17+i*1);
ld.global.nc.f32 %f(18+i*1), [%rd2+(4+i*4)];
st.shared.f32 [%r25+24], %f23;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f24, [%rd2+896];
mov.u32 %r26, %tid.x;
mad.lo.s32 %r28, %r26, 28, %r24;
for.size.2 i in range(0, 6, 1):
st.shared.f32 [%r28+(896+i*4)], %f(24+i*1);
ld.global.nc.f32 %f(25+i*1), [%rd2+(900+i*4)];
st.shared.f32 [%r28+920], %f30;
$L__BB0_3:
add.s32 %r112, %r1, 224;
mov.u32 %r113, 512;
mov.u32 %r111, %r1;
$L__BB0_4:
mul.hi.s32 %r30, %r111, 1402438301;
shr.u32 %r31, %r30, 31;
shr.s32 %r32, %r30, 7;
add.s32 %r33, %r32, %r31;
mul.lo.s32 %r34, %r33, 392;
sub.s32 %r35, %r111, %r34;
add.s32 %r36, %r35, %r5;
mad.lo.s32 %r37, %r33, 784, %r36;
mul.wide.s32 %rd9, %r37, 4;
add.s64 %rd10, %rd1, %rd9;
ld.global.nc.f32 %f31, [%rd10];
add.s32 %r38, %r3, %r113;
for.size.13 i in range(0, 4, 1):
st.shared.f32 [%r38+(-512+i*128)], %f(31+i*1);
add.s32 %r(39+i*9), %r111, (32+i*32);
mul.hi.s32 %r(40+i*9), %r(39+i*9), 1402438301;
shr.u32 %r(41+i*9), %r(40+i*9), 31;
shr.s32 %r(42+i*9), %r(40+i*9), 7;
add.s32 %r(43+i*9), %r(42+i*9), %r(41+i*9);
mul.lo.s32 %r(44+i*9), %r(43+i*9), 392;
sub.s32 %r(45+i*9), %r(39+i*9), %r(44+i*9);
add.s32 %r(46+i*9), %r(45+i*9), %r5;
mad.lo.s32 %r(47+i*9), %r(43+i*9), 784, %r(46+i*9);
mul.wide.s32 %rd(11+i*2), %r(47+i*9), 4;
add.s64 %rd(12+i*2), %rd1, %rd(11+i*2);
ld.global.nc.f32 %f(32+i*1), [%rd(12+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.f32 [%r38+(0+i*128)], %f(35+i*1);
add.s32 %r(75+i*9), %r112, (-64+i*32);
mul.hi.s32 %r(76+i*9), %r(75+i*9), 1402438301;
shr.u32 %r(77+i*9), %r(76+i*9), 31;
shr.s32 %r(78+i*9), %r(76+i*9), 7;
add.s32 %r(79+i*9), %r(78+i*9), %r(77+i*9);
mul.lo.s32 %r(80+i*9), %r(79+i*9), 392;
sub.s32 %r(81+i*9), %r(75+i*9), %r(80+i*9);
add.s32 %r(82+i*9), %r(81+i*9), %r5;
mad.lo.s32 %r(83+i*9), %r(79+i*9), 784, %r(82+i*9);
mul.wide.s32 %rd(19+i*2), %r(83+i*9), 4;
add.s64 %rd(20+i*2), %rd1, %rd(19+i*2);
ld.global.nc.f32 %f(36+i*1), [%rd(20+i*2)+0];
st.shared.f32 [%r38+256], %f37;
mul.hi.s32 %r93, %r112, 1402438301;
shr.u32 %r94, %r93, 31;
shr.s32 %r95, %r93, 7;
add.s32 %r96, %r95, %r94;
mul.lo.s32 %r97, %r96, 392;
sub.s32 %r98, %r112, %r97;
add.s32 %r99, %r98, %r5;
mad.lo.s32 %r100, %r96, 784, %r99;
mul.wide.s32 %rd23, %r100, 4;
add.s64 %rd24, %rd1, %rd23;
ld.global.nc.f32 %f38, [%rd24];
st.shared.f32 [%r38+384], %f38;
for.size.1 i in range(0, 2, 1):
add.s32 %r(112+i*-1), %r(112+i*-1), 256;
add.s32 %r113, %r113, 1024;
setp.ne.s32 %p2, %r113, 201216;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r114, 0;
$L__BB0_6:
add.s32 %r102, %r114, %r2;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(103+i*3), %r(102+i*12), 2;
add.s32 %r(105+i*3), %r(19+i*5), %r(103+i*3);
ld.shared.f32 %f(39+i*1), [%r(105+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(41+i*2), %f40, %f(39+i*3), %f(123+i*-1);
ld.shared.f32 %f(42+i*2), [%r105+(1568+i*1568)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(45+i*2), %f40, %f(44+i*2), %f(121+i*-1);
ld.shared.f32 %f(46+i*2), [%r105+(4704+i*-4700)];
for.size.9 i in range(0, 6, 1):
ld.shared.f32 %f(49+i*9), [%r108+(4+i*4)];
fma.rn.f32 %f(50+i*9), %f(49+i*9), %f(48+i*9), %f(41+i*9);
ld.shared.f32 %f(51+i*9), [%r105+(1572+i*4)];
fma.rn.f32 %f(52+i*9), %f(49+i*9), %f(51+i*9), %f(43+i*9);
ld.shared.f32 %f(53+i*9), [%r105+(3140+i*4)];
fma.rn.f32 %f(54+i*9), %f(49+i*9), %f(53+i*9), %f(45+i*9);
ld.shared.f32 %f(55+i*9), [%r105+(4708+i*4)];
fma.rn.f32 %f(56+i*9), %f(49+i*9), %f(55+i*9), %f(47+i*9);
ld.shared.f32 %f(57+i*9), [%r105+(8+i*4)];
ld.shared.f32 %f103, [%r108+28];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(123+i*-1), %f103, %f(102+i*2), %f(95+i*2);
ld.shared.f32 %f(104+i*1), [%r105+(1596+i*1568)];
fma.rn.f32 %f121, %f103, %f105, %f99;
ld.shared.f32 %f106, [%r105+4732];
fma.rn.f32 %f120, %f103, %f106, %f101;
add.s32 %r114, %r114, 8;
setp.ne.s32 %p3, %r114, 392;
@%p3 bra $L__BB0_6;
add.s32 %r110, %r110, 1;
setp.lt.u32 %p4, %r110, 2;
@%p4 bra $L__BB0_1;
cvta.to.global.u64 %rd25, %rd5;
mul.wide.s32 %rd26, %r18, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f107, [%rd27];
add.f32 %f108, %f123, %f107;
mov.f32 %f109, 0f00000000;
max.f32 %f110, %f108, %f109;
cvta.to.global.u64 %rd28, %rd4;
add.s64 %rd29, %rd28, %rd26;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd29+(0+i*4)], %f(110+i*3);
ld.global.nc.f32 %f(111+i*3), [%rd27+(4+i*4)];
add.f32 %f(112+i*3), %f(122+i*-1), %f(111+i*3);
max.f32 %f(113+i*3), %f(112+i*3), %f109;
st.global.f32 [%rd29+12], %f119;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[392];
__shared__ float placeholder_shared[50176];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 2; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x * 7;
int index1 = k_outer_outer * 392 + (int)threadIdx.x * 7;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
placeholder_d_shared[index0 + 3] = placeholder[index1 + 3];
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
placeholder_d_shared[index0 + 5] = placeholder[index1 + 5];
placeholder_d_shared[index0 + 6] = placeholder[index1 + 6];
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 224] = placeholder[index1 + 224];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 225] = placeholder[index1 + 225];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 226] = placeholder[index1 + 226];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 227] = placeholder[index1 + 227];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 228] = placeholder[index1 + 228];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 229] = placeholder[index1 + 229];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 230] = placeholder[index1 + 230];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1568; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) / 392 * 784 + k_outer_outer * 392 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) % 392;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 392; ++k_outer_inner) {
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 1568 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 392]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1176]);
}
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index6 = (int)threadIdx.x * 4 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpjm382yyx | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<345>;
.reg .b32 %r<110>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[196];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[25088];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 7;
shl.b32 %r19, %r2, 2;
mov.u32 %r20, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r20, %r19;
mul.lo.s32 %r4, %r1, 196;
add.s32 %r5, %r1, 28;
shl.b32 %r21, %r1, 2;
mov.u32 %r22, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r6, %r22, %r21;
mov.u32 %r105, 0;
mov.f32 %f281, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(282+i*1), %f281;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r105, 49;
add.s32 %r23, %r8, %r2;
mul.wide.s32 %rd10, %r23, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.2 i in range(0, 7, 1):
ld.global.nc.f32 %f(129+i*1), [%rd5+(0+i*4)];
st.shared.f32 [%r3+(0+i*4)], %f(129+i*1);
setp.gt.s32 %p1, %r1, 2;
@%p1 bra $L__BB0_3;
for.size.2 i in range(0, 7, 1):
ld.global.nc.f32 %f(136+i*1), [%rd5+(112+i*4)];
st.shared.f32 [%r3+(112+i*4)], %f(136+i*1);
$L__BB0_3:
mov.u32 %r108, 64;
mov.u32 %r106, %r1;
mov.u32 %r107, %r5;
$L__BB0_4:
mul.hi.s32 %r25, %r106, 1402438301;
shr.u32 %r26, %r25, 31;
shr.s32 %r27, %r25, 4;
add.s32 %r28, %r27, %r26;
mul.lo.s32 %r29, %r28, 49;
sub.s32 %r30, %r106, %r29;
add.s32 %r31, %r30, %r8;
mad.lo.s32 %r32, %r28, 784, %r31;
mul.wide.s32 %rd11, %r32, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f143, [%rd12];
add.s32 %r33, %r6, %r108;
for.size.13 i in range(0, 4, 1):
st.shared.f32 [%r33+(-64+i*16)], %f(143+i*1);
add.s32 %r(34+i*9), %r106, (4+i*4);
mul.hi.s32 %r(35+i*9), %r(34+i*9), 1402438301;
shr.u32 %r(36+i*9), %r(35+i*9), 31;
shr.s32 %r(37+i*9), %r(35+i*9), 4;
add.s32 %r(38+i*9), %r(37+i*9), %r(36+i*9);
mul.lo.s32 %r(39+i*9), %r(38+i*9), 49;
sub.s32 %r(40+i*9), %r(34+i*9), %r(39+i*9);
add.s32 %r(41+i*9), %r(40+i*9), %r8;
mad.lo.s32 %r(42+i*9), %r(38+i*9), 784, %r(41+i*9);
mul.wide.s32 %rd(13+i*2), %r(42+i*9), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(144+i*1), [%rd(14+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.f32 [%r33+(0+i*16)], %f(147+i*1);
add.s32 %r(70+i*9), %r107, (-8+i*4);
mul.hi.s32 %r(71+i*9), %r(70+i*9), 1402438301;
shr.u32 %r(72+i*9), %r(71+i*9), 31;
shr.s32 %r(73+i*9), %r(71+i*9), 4;
add.s32 %r(74+i*9), %r(73+i*9), %r(72+i*9);
mul.lo.s32 %r(75+i*9), %r(74+i*9), 49;
sub.s32 %r(76+i*9), %r(70+i*9), %r(75+i*9);
add.s32 %r(77+i*9), %r(76+i*9), %r8;
mad.lo.s32 %r(78+i*9), %r(74+i*9), 784, %r(77+i*9);
mul.wide.s32 %rd(21+i*2), %r(78+i*9), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(148+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r33+32], %f149;
mul.hi.s32 %r88, %r107, 1402438301;
shr.u32 %r89, %r88, 31;
shr.s32 %r90, %r88, 4;
add.s32 %r91, %r90, %r89;
mul.lo.s32 %r92, %r91, 49;
sub.s32 %r93, %r107, %r92;
add.s32 %r94, %r93, %r8;
mad.lo.s32 %r95, %r91, 784, %r94;
mul.wide.s32 %rd25, %r95, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f150, [%rd26];
st.shared.f32 [%r33+48], %f150;
add.s32 %r108, %r108, 128;
setp.ne.s32 %p2, %r108, 25152;
for.size.1 i in range(0, 2, 1):
add.s32 %r(13+i*93), %r107, (32+i*-28);
mov.u32 %r107, %r13;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r109, 0;
$L__BB0_6:
add.s32 %r97, %r109, %r4;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(98+i*3), %r(97+i*12), 2;
add.s32 %r(100+i*3), %r(22+i*-2), %r(98+i*3);
ld.shared.f32 %f(151+i*1), [%r(100+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(312+i*-1), %f152, %f(151+i*2), %f(312+i*-1);
ld.shared.f32 %f(153+i*1), [%r100+(3136+i*3136)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(310+i*-1), %f152, %f(154+i*1), %f(310+i*-1);
ld.shared.f32 %f(155+i*1), [%r100+(9408+i*3136)];
for.size.3 i in range(0, 3, 1):
for.size.2 j in range(0, 8, 1):
fma.rn.f32 %f(305+i*-8+j*-1), %f152, %f(159+i*8+j*1), %f(305+i*-8+j*-1);
ld.shared.f32 %f(160+i*8+j*1), [%r100+(196+i*196+j*3136)];
fma.rn.f32 %f281, %f152, %f183, %f281;
add.s32 %r109, %r109, 1;
setp.ne.s32 %p3, %r109, 49;
@%p3 bra $L__BB0_6;
add.s32 %r105, %r105, 1;
setp.lt.u32 %p4, %r105, 16;
@%p4 bra $L__BB0_1;
mul.wide.s32 %rd27, %r21, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f184, [%rd28];
add.f32 %f185, %f312, %f184;
mov.f32 %f186, 0f00000000;
max.f32 %f187, %f185, %f186;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd29+(0+i*64)], %f(187+i*3);
ld.global.nc.f32 %f(188+i*3), [%rd28+(64+i*64)];
add.f32 %f(189+i*3), %f(311+i*-1), %f(188+i*3);
max.f32 %f(190+i*3), %f(189+i*3), %f186;
for.size.10 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(448+i*4+j*-444)], %f(208+i*24+j*3);
ld.global.nc.f32 %f(209+i*24+j*3), [%rd28+(4+i*4+j*64)];
add.f32 %f(210+i*24+j*3), %f(304+i*-8+j*-1), %f(209+i*24+j*3);
max.f32 %f(211+i*24+j*3), %f(210+i*24+j*3), %f186;
for.size.4 j in range(0, 6, 1):
st.global.f32 [%rd29+(68+i*4+j*64)], %f(214+i*24+j*3);
ld.global.nc.f32 %f(215+i*24+j*3), [%rd28+(132+i*4+j*64)];
add.f32 %f(216+i*24+j*3), %f(302+i*-8+j*-1), %f(215+i*24+j*3);
max.f32 %f(217+i*24+j*3), %f(216+i*24+j*3), %f186;
st.global.f32 [%rd29+460], %f280;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[49];
__shared__ float placeholder_shared[6272];
T_dense[0] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 16; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x * 7;
int index1 = k_outer_outer * 49 + (int)threadIdx.x * 7;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
placeholder_d_shared[index0 + 3] = placeholder[index1 + 3];
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
placeholder_d_shared[index0 + 5] = placeholder[index1 + 5];
placeholder_d_shared[index0 + 6] = placeholder[index1 + 6];
if ((int)threadIdx.x < 3) {
placeholder_d_shared[index0 + 28] = placeholder[index1 + 28];
}
if ((int)threadIdx.x < 3) {
placeholder_d_shared[index0 + 29] = placeholder[index1 + 29];
}
if ((int)threadIdx.x < 3) {
placeholder_d_shared[index0 + 30] = placeholder[index1 + 30];
}
if ((int)threadIdx.x < 3) {
placeholder_d_shared[index0 + 31] = placeholder[index1 + 31];
}
if ((int)threadIdx.x < 3) {
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
}
if ((int)threadIdx.x < 3) {
placeholder_d_shared[index0 + 33] = placeholder[index1 + 33];
}
if ((int)threadIdx.x < 3) {
placeholder_d_shared[index0 + 34] = placeholder[index1 + 34];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1568; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) / 49 * 784 + k_outer_outer * 49 + (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x) % 49;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 49; ++k_outer_inner) {
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 196 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1568]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2352]);
T_dense[16] = T_dense[16] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3136]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3920]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4704]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 5488]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 49]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 833]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1617]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2401]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3185]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3969]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4753]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 5537]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 98]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 882]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1666]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2450]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3234]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4018]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4802]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 5586]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 147]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 931]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1715]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2499]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3283]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4067]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4851]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 5635]);
}
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index6 = (int)threadIdx.x * 4 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 16] = max(T_dense[index7 + 4] + placeholder2[index6 + 16], 0.000000e+00f);
T_relu[index6 + 32] = max(T_dense[index7 + 8] + placeholder2[index6 + 32], 0.000000e+00f);
T_relu[index6 + 48] = max(T_dense[index7 + 12] + placeholder2[index6 + 48], 0.000000e+00f);
T_relu[index6 + 64] = max(T_dense[index7 + 16] + placeholder2[index6 + 64], 0.000000e+00f);
T_relu[index6 + 80] = max(T_dense[index7 + 20] + placeholder2[index6 + 80], 0.000000e+00f);
T_relu[index6 + 96] = max(T_dense[index7 + 24] + placeholder2[index6 + 96], 0.000000e+00f);
T_relu[index6 + 112] = max(T_dense[index7 + 28] + placeholder2[index6 + 112], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpyh9mmiws | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<1283>;
.reg .b32 %r<298>;
.reg .b64 %rd<17>;
// demoted variable
.shared .align 16 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[4096];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd5;
mov.u32 %r295, 0;
mov.f32 %f1027, 0f00000000;
for.size.1 i in range(0, 127, 1):
mov.f32 %f(1028+i*1), %f1027;
$L__BB0_1:
mov.u32 %r296, 0;
bar.sync 0;
shl.b32 %r2, %r295, 3;
mul.wide.s32 %rd9, %r2, 4;
add.s64 %rd10, %rd2, %rd9;
for.size.2 i in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(10+i*8), %r(11+i*8), %r(12+i*8), %r(13+i*8)}, [%rd10+(0+i*16)];
st.shared.v4.u32 [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*16)], {%r(10+i*8), %r(11+i*8), %r(12+i*8), %r(13+i*8)};
$L__BB0_2:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(292+i*-266), %r(295+i*1), (3+i*2);
mov.u32 %r27, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r28, %r27, %r26;
mad.lo.s32 %r29, %r296, 784, %r292;
mul.wide.s32 %rd11, %r29, 4;
add.s64 %rd12, %rd1, %rd11;
for.size.4 i in range(0, 16, 1):
for.size.1 j in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(30+i*16+j*8), %r(31+i*16+j*8), %r(32+i*16+j*8), %r(33+i*16+j*8)}, [%rd12+(0+i*3136+j*16)];
for.size.1 j in range(0, 2, 1):
st.shared.v4.u32 [%r28+(0+i*32+j*16)], {%r(30+i*16+j*8), %r(31+i*16+j*8), %r(32+i*16+j*8), %r(33+i*16+j*8)};
add.s32 %r296, %r296, 16;
setp.ne.s32 %p1, %r296, 128;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r297, 0;
$L__BB0_4:
mov.u32 %r293, _ZZ24default_function_kernel0E18placeholder_shared;
shl.b32 %r287, %r297, 2;
add.s32 %r289, %r293, %r287;
ld.shared.f32 %f513, [%r289];
mov.u32 %r290, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r291, %r290, %r287;
ld.shared.f32 %f514, [%r291];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(1154+i*-2), %f514, %f(513+i*2), %f(1154+i*-2);
ld.shared.f32 %f(515+i*1), [%r289+(64+i*64)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(1150+i*-2), %f514, %f(516+i*1), %f(1150+i*-2);
ld.shared.f32 %f(517+i*1), [%r289+(192+i*64)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(1028+i*125), %f514, %f(577+i*1), %f(1028+i*125);
ld.shared.f32 %f(578+i*1), [%r289+(32+i*64)];
for.size.2 i in range(0, 62, 1):
fma.rn.f32 %f(1151+i*-2), %f514, %f(579+i*1), %f(1151+i*-2);
ld.shared.f32 %f(580+i*1), [%r289+(160+i*64)];
fma.rn.f32 %f1027, %f514, %f641, %f1027;
add.s32 %r297, %r297, 1;
setp.ne.s32 %p2, %r297, 8;
@%p2 bra $L__BB0_4;
add.s32 %r295, %r295, 1;
setp.lt.u32 %p3, %r295, 98;
@%p3 bra $L__BB0_1;
ld.param.u64 %rd16, [default_function_kernel0_param_2];
cvta.to.global.u64 %rd15, %rd16;
ld.param.u64 %rd14, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd14;
ld.global.nc.f32 %f642, [%rd13];
add.f32 %f643, %f1154, %f642;
mov.f32 %f644, 0f00000000;
for.size.4 i in range(0, 2, 1):
max.f32 %f(645+i*3), %f(643+i*4), %f644;
st.global.f32 [%rd15+(0+i*8)], %f(645+i*3);
ld.global.nc.f32 %f(646+i*3), [%rd13+(8+i*8)];
add.f32 %f(647+i*3), %f(1152+i*-2), %f(646+i*3);
for.size.4 i in range(0, 61, 1):
max.f32 %f(651+i*3), %f(650+i*3), %f644;
st.global.f32 [%rd15+(16+i*8)], %f(651+i*3);
ld.global.nc.f32 %f(652+i*3), [%rd13+(24+i*8)];
add.f32 %f(653+i*3), %f(1148+i*-2), %f(652+i*3);
for.size.4 i in range(0, 2, 1):
max.f32 %f(834+i*3), %f(833+i*3), %f644;
st.global.f32 [%rd15+(504+i*-500)], %f(834+i*3);
ld.global.nc.f32 %f(835+i*3), [%rd13+(4+i*8)];
add.f32 %f(836+i*3), %f(1153+i*-2), %f(835+i*3);
for.size.4 i in range(0, 62, 1):
max.f32 %f(840+i*3), %f(839+i*3), %f644;
st.global.f32 [%rd15+(12+i*8)], %f(840+i*3);
ld.global.nc.f32 %f(841+i*3), [%rd13+(20+i*8)];
add.f32 %f(842+i*3), %f(1149+i*-2), %f(841+i*3);
max.f32 %f1026, %f1025, %f644;
st.global.f32 [%rd15+508], %f1026;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[128];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[1024];
for (int j_inner_init = 0; j_inner_init < 2; ++j_inner_init) {
for (int vthread_s = 0; vthread_s < 64; ++vthread_s) {
int index0 = vthread_s * 2 + j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
}
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
((float4*)(placeholder_d_shared + 0))[0] = ((float4*)(placeholder + k_outer_outer * 8))[0];
((float4*)(placeholder_d_shared + 4))[0] = ((float4*)(placeholder + k_outer_outer * 8 + 4))[0];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 128; ++ax0_ax1_fused_outer_outer) {
ulonglong4*(placeholder_shared + ax0_ax1_fused_outer_outer * 8)[0] = ulonglong4*(placeholder1 + ax0_ax1_fused_outer_outer * 784 + k_outer_outer * 8)[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 8; ++k_outer_inner) {
for (int j_inner = 0; j_inner < 2; ++j_inner) {
for (int vthread_s1 = 0; vthread_s1 < 64; ++vthread_s1) {
int index1 = vthread_s1 * 2 + j_inner;
int index2 = k_outer_inner;
int index3 = vthread_s1 * 16 + j_inner * 8 + k_outer_inner;
T_dense[index1] = T_dense[index1] + placeholder_d_shared[index2] * placeholder_shared[index3];
}
}
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
for (int vthread_s2 = 0; vthread_s2 < 64; ++vthread_s2) {
int index4 = vthread_s2 * 2 + ax1_inner;
T_relu[index4] = max(T_dense[index4] + placeholder2[index4], 0.000000e+00f);
}
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpaegymf72 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<6>;
.reg .f32 %f<781>;
.reg .b32 %r<127>;
.reg .b64 %rd<32>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[100352];
ld.param.u64 %rd1, [default_function_kernel0_param_0];
ld.param.u64 %rd2, [default_function_kernel0_param_1];
mov.u32 %r15, %tid.x;
mul.lo.s32 %r1, %r15, 784;
mov.u32 %r122, 0;
mov.f32 %f717, 0f00000000;
cvta.to.global.u64 %rd5, %rd1;
cvta.to.global.u64 %rd8, %rd2;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(718+i*1), %f717;
$L__BB0_1:
mov.u32 %r123, 0;
bar.sync 0;
$L__BB0_2:
mul.lo.s32 %r119, %r122, 196;
add.s32 %r18, %r119, %r15;
shl.b32 %r19, %r123, 1;
add.s32 %r20, %r18, %r19;
mul.wide.s32 %rd6, %r20, 4;
add.s64 %rd7, %rd5, %rd6;
ld.global.nc.f32 %f321, [%rd7];
add.s32 %r21, %r19, %r15;
shl.b32 %r22, %r21, 2;
mov.u32 %r23, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r24, %r23, %r22;
st.shared.f32 [%r24], %f321;
ld.global.nc.f32 %f322, [%rd7+8];
st.shared.f32 [%r24+8], %f322;
add.s32 %r123, %r123, 2;
setp.ne.s32 %p1, %r123, 98;
@%p1 bra $L__BB0_2;
mov.u32 %r124, 0;
$L__BB0_4:
mul.lo.s32 %r115, %r122, 196;
shl.b32 %r27, %r124, 1;
add.s32 %r28, %r27, %r15;
mul.hi.s32 %r29, %r28, 1402438301;
shr.u32 %r30, %r29, 31;
shr.s32 %r31, %r29, 6;
add.s32 %r32, %r31, %r30;
mul.lo.s32 %r33, %r32, 196;
sub.s32 %r34, %r28, %r33;
add.s32 %r35, %r34, %r115;
mad.lo.s32 %r36, %r32, 784, %r35;
mul.wide.s32 %rd9, %r36, 4;
add.s64 %rd10, %rd8, %rd9;
ld.global.nc.f32 %f323, [%rd10];
shl.b32 %r37, %r28, 2;
mov.u32 %r38, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r39, %r38, %r37;
for.size.13 i in range(0, 7, 1):
st.shared.f32 [%r39+(0+i*8)], %f(323+i*1);
add.s32 %r(40+i*9), %r28, (2+i*2);
mul.hi.s32 %r(41+i*9), %r(40+i*9), 1402438301;
shr.u32 %r(42+i*9), %r(41+i*9), 31;
shr.s32 %r(43+i*9), %r(41+i*9), 6;
add.s32 %r(44+i*9), %r(43+i*9), %r(42+i*9);
mul.lo.s32 %r(45+i*9), %r(44+i*9), 196;
sub.s32 %r(46+i*9), %r(40+i*9), %r(45+i*9);
add.s32 %r(47+i*9), %r(46+i*9), %r115;
mad.lo.s32 %r(48+i*9), %r(44+i*9), 784, %r(47+i*9);
mul.wide.s32 %rd(11+i*2), %r(48+i*9), 4;
add.s64 %rd(12+i*2), %rd8, %rd(11+i*2);
ld.global.nc.f32 %f(324+i*1), [%rd(12+i*2)+0];
st.shared.f32 [%r39+56], %f330;
add.s32 %r124, %r124, 8;
setp.ne.s32 %p2, %r124, 12544;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r125, 0;
$L__BB0_6:
mov.u32 %r126, 0;
$L__BB0_7:
shl.b32 %r120, %r125, 2;
mov.u32 %r117, _ZZ24default_function_kernel0E18placeholder_shared;
mov.u32 %r116, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.3 i in range(0, 2, 1):
add.s32 %r(105+i*4), %r(126+i*-21), %r(120+i*-119);
shl.b32 %r(106+i*4), %r(105+i*4), 2;
add.s32 %r(108+i*4), %r(116+i*1), %r(106+i*4);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(331+i*1), [%r(112+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(780+i*-4), %f332, %f(331+i*2), %f(780+i*-4);
ld.shared.f32 %f(333+i*1), [%r112+(6272+i*6272)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(772+i*-4), %f332, %f(334+i*1), %f(772+i*-4);
ld.shared.f32 %f(335+i*1), [%r112+(18816+i*6272)];
for.size.6 i in range(0, 3, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(720+i*-1+j*59), %f332, %f(347+i*16+j*1), %f(720+i*-1+j*59);
ld.shared.f32 %f(348+i*16+j*1), [%r112+(784+i*784+j*6272)];
for.size.2 j in range(0, 14, 1):
fma.rn.f32 %f(775+i*-1+j*-4), %f332, %f(349+i*16+j*1), %f(775+i*-1+j*-4);
ld.shared.f32 %f(350+i*16+j*1), [%r112+(13328+i*784+j*6272)];
fma.rn.f32 %f717, %f332, %f395, %f717;
add.s32 %r126, %r126, 1;
setp.ne.s32 %p3, %r126, 4;
@%p3 bra $L__BB0_7;
add.s32 %r125, %r125, 1;
setp.lt.u32 %p4, %r125, 49;
@%p4 bra $L__BB0_6;
add.s32 %r122, %r122, 1;
setp.lt.u32 %p5, %r122, 4;
@%p5 bra $L__BB0_1;
ld.param.u64 %rd31, [default_function_kernel0_param_2];
ld.param.u64 %rd30, [default_function_kernel0_param_3];
shl.b32 %r114, %r15, 2;
cvta.to.global.u64 %rd25, %rd30;
mul.wide.s32 %rd26, %r114, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f396, [%rd27];
add.f32 %f397, %f780, %f396;
mov.f32 %f398, 0f00000000;
max.f32 %f399, %f397, %f398;
cvta.to.global.u64 %rd28, %rd31;
add.s64 %rd29, %rd28, %rd26;
for.size.4 i in range(0, 15, 1):
st.global.f32 [%rd29+(0+i*32)], %f(399+i*3);
ld.global.nc.f32 %f(400+i*3), [%rd27+(32+i*32)];
add.f32 %f(401+i*3), %f(776+i*-4), %f(400+i*3);
max.f32 %f(402+i*3), %f(401+i*3), %f398;
for.size.10 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(480+i*4+j*-476)], %f(444+i*48+j*3);
ld.global.nc.f32 %f(445+i*48+j*3), [%rd27+(4+i*4+j*32)];
add.f32 %f(446+i*48+j*3), %f(779+i*-1+j*-4), %f(445+i*48+j*3);
max.f32 %f(447+i*48+j*3), %f(446+i*48+j*3), %f398;
for.size.4 j in range(0, 14, 1):
st.global.f32 [%rd29+(36+i*4+j*32)], %f(450+i*48+j*3);
ld.global.nc.f32 %f(451+i*48+j*3), [%rd27+(68+i*4+j*32)];
add.f32 %f(452+i*48+j*3), %f(771+i*-1+j*-4), %f(451+i*48+j*3);
max.f32 %f(453+i*48+j*3), %f(452+i*48+j*3), %f398;
st.global.f32 [%rd29+492], %f588;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[25088];
for (int j_inner_init = 0; j_inner_init < 4; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
T_dense[index0 + 4] = 0.000000e+00f;
T_dense[index0 + 8] = 0.000000e+00f;
T_dense[index0 + 12] = 0.000000e+00f;
T_dense[index0 + 16] = 0.000000e+00f;
T_dense[index0 + 20] = 0.000000e+00f;
T_dense[index0 + 24] = 0.000000e+00f;
T_dense[index0 + 28] = 0.000000e+00f;
T_dense[index0 + 32] = 0.000000e+00f;
T_dense[index0 + 36] = 0.000000e+00f;
T_dense[index0 + 40] = 0.000000e+00f;
T_dense[index0 + 44] = 0.000000e+00f;
T_dense[index0 + 48] = 0.000000e+00f;
T_dense[index0 + 52] = 0.000000e+00f;
T_dense[index0 + 56] = 0.000000e+00f;
T_dense[index0 + 60] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index2 = k_outer_outer * 196 + ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 12544; ++ax0_ax1_fused_outer_outer1) {
int index3 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
int index4 = (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) % 196;
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 49; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 4; ++k_inner) {
for (int j_inner = 0; j_inner < 4; ++j_inner) {
int index5 = j_inner;
int index6 = k_outer_inner * 4 + k_inner;
int index7 = (int)threadIdx.x * 784 + j_inner * 196 + k_outer_inner * 4 + k_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
T_dense[index5 + 4] = T_dense[index5 + 4] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 1568]);
T_dense[index5 + 8] = T_dense[index5 + 8] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 3136]);
T_dense[index5 + 12] = T_dense[index5 + 12] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 4704]);
T_dense[index5 + 16] = T_dense[index5 + 16] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 6272]);
T_dense[index5 + 20] = T_dense[index5 + 20] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 7840]);
T_dense[index5 + 24] = T_dense[index5 + 24] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 9408]);
T_dense[index5 + 28] = T_dense[index5 + 28] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 10976]);
T_dense[index5 + 32] = T_dense[index5 + 32] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 12544]);
T_dense[index5 + 36] = T_dense[index5 + 36] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 14112]);
T_dense[index5 + 40] = T_dense[index5 + 40] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 15680]);
T_dense[index5 + 44] = T_dense[index5 + 44] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 17248]);
T_dense[index5 + 48] = T_dense[index5 + 48] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 18816]);
T_dense[index5 + 52] = T_dense[index5 + 52] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 20384]);
T_dense[index5 + 56] = T_dense[index5 + 56] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 21952]);
T_dense[index5 + 60] = T_dense[index5 + 60] + (placeholder_d_shared[index6] * placeholder_shared[index7 + 23520]);
}
}
}
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
for (int vthread_s = 0; vthread_s < 16; ++vthread_s) {
int index8 = vthread_s * 8 + (int)threadIdx.x * 4 + ax1_inner;
int index9 = vthread_s * 4 + ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp0t283_el | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<6>;
.reg .f32 %f<50>;
.reg .b32 %r<28>;
.reg .b64 %rd<26>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[8];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[512];
ld.param.u64 %rd11, [default_function_kernel0_param_0];
ld.param.u64 %rd12, [default_function_kernel0_param_1];
ld.param.u64 %rd13, [default_function_kernel0_param_2];
ld.param.u64 %rd10, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd14, %rd12;
cvta.to.global.u64 %rd1, %rd13;
mov.u32 %r1, %tid.x;
shl.b32 %r10, %r1, 3;
mov.u32 %r11, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r11, %r10;
mul.lo.s32 %r12, %r1, 3136;
shl.b32 %r13, %r1, 5;
mov.u32 %r14, _ZZ24default_function_kernel0E18placeholder_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(3+i*1), %r14, %r(13+i*-3);
mul.lo.s32 %r15, %r1, 784;
cvta.to.global.u64 %rd15, %rd11;
mul.wide.s32 %rd16, %r15, 4;
add.s64 %rd17, %rd15, %rd16;
add.s64 %rd25, %rd17, 8;
mov.u32 %r5, %ctaid.x;
mad.lo.s32 %r16, %r5, 50176, %r12;
mul.wide.s32 %rd18, %r16, 4;
add.s64 %rd19, %rd14, %rd18;
add.s64 %rd24, %rd19, 6272;
mov.f32 %f48, 0f00000000;
mov.u32 %r27, 0;
mov.f32 %f49, %f48;
$L__BB0_1:
bar.sync 0;
add.s64 %rd6, %rd25, -8;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.v2.u32 {%r17, %r18}, [%rd6];
st.shared.v2.u32 [%r2], {%r17, %r18};
$L__BB0_3:
setp.gt.s32 %p2, %r1, 15;
add.s64 %rd7, %rd24, -6272;
@%p2 bra $L__BB0_5;
for.size.3 i in range(0, 4, 1):
for.size.2 j in range(0, 2, 1):
ld.global.nc.f32 %f(9+i*2+j*1), [%rd7+(0+i*3136+j*4)];
st.shared.f32 [%r3+(0+i*8+j*4)], %f(9+i*2+j*1);
$L__BB0_5:
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(17+i*5), [%r4+(0+i*4)];
ld.shared.f32 %f(18+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(19+i*-16), %f(18+i*5), %f(17+i*5), %f(48+i*-29);
ld.shared.f32 %f(20+i*4), [%r4+(256+i*4)];
fma.rn.f32 %f(21+i*-17), %f(18+i*5), %f(20+i*4), %f(49+i*-28);
bar.sync 0;
@%p1 bra $L__BB0_7;
ld.global.nc.v2.u32 {%r21, %r22}, [%rd6+8];
st.shared.v2.u32 [%r2], {%r21, %r22};
$L__BB0_7:
@%p2 bra $L__BB0_9;
for.size.3 i in range(0, 4, 1):
for.size.2 j in range(0, 2, 1):
ld.global.nc.f32 %f(25+i*2+j*1), [%rd7+(8+i*3136+j*4)];
st.shared.f32 [%r3+(0+i*8+j*4)], %f(25+i*2+j*1);
$L__BB0_9:
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(33+i*5), [%r4+(0+i*4)];
ld.shared.f32 %f(34+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(35+i*13), %f(34+i*5), %f(33+i*5), %f(3+i*32);
ld.shared.f32 %f(36+i*4), [%r4+(256+i*4)];
fma.rn.f32 %f(37+i*12), %f(34+i*5), %f(36+i*4), %f(4+i*33);
for.size.1 i in range(0, 2, 1):
add.s64 %rd(25+i*-1), %rd(25+i*-1), 16;
add.s32 %r27, %r27, 2;
setp.ne.s32 %p5, %r27, 392;
@%p5 bra $L__BB0_1;
cvta.to.global.u64 %rd20, %rd10;
shl.b32 %r25, %r5, 6;
add.s32 %r26, %r25, %r1;
mul.wide.s32 %rd21, %r26, 4;
add.s64 %rd22, %rd20, %rd21;
ld.global.nc.f32 %f41, [%rd22];
add.f32 %f42, %f48, %f41;
mov.f32 %f43, 0f00000000;
max.f32 %f44, %f42, %f43;
add.s64 %rd23, %rd1, %rd21;
st.global.f32 [%rd23], %f44;
ld.global.nc.f32 %f45, [%rd22+128];
add.f32 %f46, %f49, %f45;
max.f32 %f47, %f46, %f43;
st.global.f32 [%rd23+128], %f47;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[2];
__shared__ float placeholder_shared[128];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 392; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
((float2*)(placeholder_d_shared + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder + (int)threadIdx.x * 784 + k_outer_outer * 2))[0];
}
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x * 8;
int index1 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 3136 + k_outer_outer * 2;
placeholder_shared[index0] = placeholder1[index1];
}
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x * 8;
int index1 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 3136 + k_outer_outer * 2;
placeholder_shared[index0 + 1] = placeholder1[index1 + 1];
}
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x * 8;
int index1 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 3136 + k_outer_outer * 2;
placeholder_shared[index0 + 2] = placeholder1[index1 + 784];
}
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x * 8;
int index1 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 3136 + k_outer_outer * 2;
placeholder_shared[index0 + 3] = placeholder1[index1 + 785];
}
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x * 8;
int index1 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 3136 + k_outer_outer * 2;
placeholder_shared[index0 + 4] = placeholder1[index1 + 1568];
}
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x * 8;
int index1 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 3136 + k_outer_outer * 2;
placeholder_shared[index0 + 5] = placeholder1[index1 + 1569];
}
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x * 8;
int index1 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 3136 + k_outer_outer * 2;
placeholder_shared[index0 + 6] = placeholder1[index1 + 2352];
}
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x * 8;
int index1 = (int)blockIdx.x * 50176 + (int)threadIdx.x * 3136 + k_outer_outer * 2;
placeholder_shared[index0 + 7] = placeholder1[index1 + 2353];
}
__syncthreads();
int index2 = (int)threadIdx.x * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index2];
T_dense[1] = T_dense[1] + (placeholder_d_shared[0] * placeholder_shared[index2 + 64]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[1] * placeholder_shared[index2 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[1] * placeholder_shared[index2 + 65]);
}
int index3 = (int)blockIdx.x * 64 + (int)threadIdx.x;
T_relu[index3] = max(T_dense[0] + placeholder2[index3], 0.000000e+00f);
T_relu[index3 + 32] = max(T_dense[1] + placeholder2[index3 + 32], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpcypax66w | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<19>;
.reg .f32 %f<211>;
.reg .b32 %r<146>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[196];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[25088];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd8;
cvta.to.global.u64 %rd3, %rd7;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 784;
shl.b32 %r24, %r1, 2;
mov.u32 %r25, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r25, %r24;
shl.b32 %r4, %r1, 1;
add.s32 %r5, %r4, 49;
shl.b32 %r26, %r1, 3;
mov.u32 %r27, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r28, %r27, %r26;
add.s32 %r6, %r28, 128;
mov.u32 %r140, 0;
mov.f32 %f179, 0f00000000;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(180+i*1), %f179;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r140, 49;
add.s32 %r29, %r8, %r1;
mul.wide.s32 %rd10, %r29, 4;
add.s64 %rd5, %rd1, %rd10;
setp.gt.s32 %p1, %r1, 48;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f81, [%rd5];
st.shared.f32 [%r3], %f81;
$L__BB0_3:
setp.gt.s32 %p2, %r1, 40;
@%p2 bra $L__BB0_5;
ld.global.nc.f32 %f82, [%rd5+32];
st.shared.f32 [%r3+32], %f82;
$L__BB0_5:
setp.gt.s32 %p3, %r1, 32;
@%p3 bra $L__BB0_7;
ld.global.nc.f32 %f83, [%rd5+64];
st.shared.f32 [%r3+64], %f83;
$L__BB0_7:
setp.gt.s32 %p4, %r1, 24;
@%p4 bra $L__BB0_9;
ld.global.nc.f32 %f84, [%rd5+96];
st.shared.f32 [%r3+96], %f84;
$L__BB0_9:
setp.gt.s32 %p5, %r1, 16;
@%p5 bra $L__BB0_11;
ld.global.nc.f32 %f85, [%rd5+128];
st.shared.f32 [%r3+128], %f85;
$L__BB0_11:
setp.gt.s32 %p6, %r1, 8;
@%p6 bra $L__BB0_13;
ld.global.nc.f32 %f86, [%rd5+160];
st.shared.f32 [%r3+160], %f86;
$L__BB0_13:
setp.gt.s32 %p7, %r1, 0;
@%p7 bra $L__BB0_15;
ld.global.nc.f32 %f87, [%rd5+192];
st.shared.f32 [%r3+192], %f87;
$L__BB0_15:
mov.u32 %r144, 128;
mov.u32 %r141, %r6;
mov.u32 %r142, %r4;
mov.u32 %r143, %r5;
$L__BB0_16:
add.s32 %r31, %r142, 1;
for.size.6 i in range(0, 2, 1):
mul.hi.s32 %r(32+i*6), %r(31+i*111), 1402438301;
shr.u32 %r(33+i*6), %r(32+i*6), 31;
shr.s32 %r(34+i*6), %r(32+i*6), 4;
add.s32 %r(35+i*6), %r(34+i*6), %r(33+i*6);
mul.lo.s32 %r(36+i*6), %r(35+i*6), 49;
sub.s32 %r(37+i*6), %r(31+i*111), %r(36+i*6);
for.size.1 i in range(0, 2, 1):
setp.gt.s32 %p(8+i*1), %r(43+i*-6), -1;
for.size.2 i in range(0, 2, 1):
shr.s32 %r(44+i*2), %r(43+i*-6), 31;
add.s32 %r(45+i*2), %r(41+i*-6), %r(44+i*2);
for.size.1 i in range(0, 2, 1):
selp.b32 %r(48+i*1), 0, 49, %p(8+i*1);
for.size.1 i in range(0, 2, 1):
add.s32 %r(50+i*1), %r(43+i*7), %r(8+i*40);
for.size.35 i in range(0, 2, 1):
mad.lo.s32 %r(52+i*26), %r(45+i*26), 784, %r(51+i*26);
add.s32 %r(53+i*26), %r(37+i*32), %r8;
add.s32 %r(54+i*26), %r(53+i*26), %r(49+i*26);
mad.lo.s32 %r(55+i*26), %r(47+i*26), 784, %r(54+i*26);
mul.wide.s32 %rd(11+i*4), %r(52+i*26), 4;
add.s64 %rd(12+i*4), %rd3, %rd(11+i*4);
mul.wide.s32 %rd(13+i*4), %r(55+i*26), 4;
add.s64 %rd(14+i*4), %rd3, %rd(13+i*4);
ld.global.nc.f32 %f(88+i*2), [%rd(12+i*6)+0];
ld.global.nc.f32 %f(89+i*2), [%rd(14+i*2)+0];
st.shared.v2.f32 [%r141+(-128+i*64)], {%f(88+i*3), %f(89+i*1)};
add.s32 %r(56+i*26), %r142, (16+i*16);
mul.hi.s32 %r(57+i*26), %r(56+i*26), 1402438301;
shr.u32 %r(58+i*26), %r(57+i*26), 31;
shr.s32 %r(59+i*26), %r(57+i*26), 4;
add.s32 %r(60+i*26), %r(59+i*26), %r(58+i*26);
mul.lo.s32 %r(61+i*26), %r(60+i*26), 49;
sub.s32 %r(62+i*26), %r(56+i*26), %r(61+i*26);
add.s32 %r(63+i*26), %r(142+i*1), (17+i*-33);
mul.hi.s32 %r(64+i*26), %r(63+i*26), 1402438301;
shr.u32 %r(65+i*26), %r(64+i*26), 31;
shr.s32 %r(66+i*26), %r(64+i*26), 4;
add.s32 %r(67+i*26), %r(66+i*26), %r(65+i*26);
mul.lo.s32 %r(68+i*26), %r(67+i*26), 49;
sub.s32 %r(69+i*26), %r(63+i*26), %r(68+i*26);
setp.gt.s32 %p(10+i*2), %r(62+i*26), -1;
setp.gt.s32 %p(11+i*2), %r(69+i*26), -1;
shr.s32 %r(70+i*26), %r(62+i*26), 31;
add.s32 %r(71+i*26), %r(60+i*26), %r(70+i*26);
shr.s32 %r(72+i*26), %r(69+i*26), 31;
add.s32 %r(73+i*26), %r(67+i*26), %r(72+i*26);
selp.b32 %r(74+i*26), 0, 49, %p(10+i*2);
selp.b32 %r(75+i*26), 0, 49, %p(11+i*2);
add.s32 %r(76+i*26), %r(62+i*26), %r8;
add.s32 %r(77+i*26), %r(76+i*26), %r(74+i*26);
mad.lo.s32 %r104, %r97, 784, %r103;
for.size.1 i in range(0, 2, 1):
add.s32 %r(105+i*1), %r(95+i*10), %r(8+i*93);
mad.lo.s32 %r107, %r99, 784, %r106;
for.size.2 i in range(0, 2, 1):
mul.wide.s32 %rd(19+i*2), %r(104+i*3), 4;
add.s64 %rd(20+i*2), %rd3, %rd(19+i*2);
for.size.1 i in range(0, 2, 1):
ld.global.nc.f32 %f(92+i*1), [%rd(22+i*-2)+0];
st.shared.v2.f32 [%r141], {%f93, %f92};
add.s32 %r108, %r143, -1;
for.size.7 i in range(0, 2, 1):
mul.hi.s32 %r(109+i*6), %r(108+i*35), 1402438301;
shr.u32 %r(110+i*6), %r(109+i*6), 31;
shr.s32 %r(111+i*6), %r(109+i*6), 4;
add.s32 %r(112+i*6), %r(111+i*6), %r(110+i*6);
mul.lo.s32 %r(113+i*6), %r(112+i*6), 49;
sub.s32 %r(114+i*6), %r(108+i*35), %r(113+i*6);
setp.gt.s32 %p(14+i*1), %r(114+i*6), -1;
for.size.2 i in range(0, 2, 1):
shr.s32 %r(121+i*2), %r(114+i*6), 31;
add.s32 %r(122+i*2), %r(112+i*6), %r(121+i*2);
for.size.1 i in range(0, 2, 1):
selp.b32 %r(125+i*1), 0, 49, %p(14+i*1);
for.size.1 i in range(0, 2, 1):
add.s32 %r(127+i*1), %r(114+i*13), %r(8+i*117);
mad.lo.s32 %r129, %r122, 784, %r128;
for.size.1 i in range(0, 2, 1):
add.s32 %r(130+i*1), %r(120+i*10), %r(8+i*118);
mad.lo.s32 %r132, %r124, 784, %r131;
for.size.2 i in range(0, 2, 1):
mul.wide.s32 %rd(23+i*2), %r(129+i*3), 4;
add.s64 %rd(24+i*2), %rd3, %rd(23+i*2);
for.size.1 i in range(0, 2, 1):
ld.global.nc.f32 %f(94+i*1), [%rd(26+i*-2)+0];
st.shared.v2.f32 [%r141+64], {%f95, %f94};
for.size.1 i in range(0, 2, 1):
add.s32 %r(141+i*3), %r(141+i*3), 256;
setp.ne.s32 %p16, %r144, 25216;
for.size.1 i in range(0, 2, 1):
add.s32 %r(16+i*126), %r143, (64+i*-49);
mov.u32 %r143, %r16;
@%p16 bra $L__BB0_16;
bar.sync 0;
mov.u32 %r145, 0;
bra.uni $L__BB0_18;
$L__BB0_19:
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(113+i*1), [%r(19+i*1)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(194+i*-1), %f114, %f(113+i*2), %f(194+i*-1);
ld.shared.f32 %f(115+i*1), [%r19+(200+i*196)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(192+i*-1), %f114, %f(116+i*1), %f(192+i*-1);
ld.shared.f32 %f(117+i*1), [%r19+(592+i*196)];
fma.rn.f32 %f179, %f114, %f129, %f179;
add.s32 %r145, %r145, 2;
$L__BB0_18:
add.s32 %r134, %r145, %r2;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(135+i*2), %r(134+i*11), 2;
add.s32 %r(19+i*1), %r(27+i*-2), %r(135+i*2);
ld.shared.f32 %f(96+i*1), [%r(19+i*1)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(194+i*-1), %f97, %f(96+i*2), %f(194+i*-1);
ld.shared.f32 %f(98+i*1), [%r19+(196+i*196)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(192+i*-1), %f97, %f(99+i*1), %f(192+i*-1);
ld.shared.f32 %f(100+i*1), [%r19+(588+i*196)];
fma.rn.f32 %f179, %f97, %f112, %f179;
setp.eq.s32 %p17, %r145, 48;
@%p17 bra $L__BB0_20;
bra.uni $L__BB0_19;
$L__BB0_20:
add.s32 %r140, %r140, 1;
setp.lt.u32 %p18, %r140, 16;
@%p18 bra $L__BB0_1;
shl.b32 %r139, %r1, 4;
mul.wide.s32 %rd27, %r139, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f130, [%rd28];
add.f32 %f131, %f194, %f130;
mov.f32 %f132, 0f00000000;
max.f32 %f133, %f131, %f132;
add.s64 %rd29, %rd2, %rd27;
for.size.4 i in range(0, 15, 1):
st.global.f32 [%rd29+(0+i*4)], %f(133+i*3);
ld.global.nc.f32 %f(134+i*3), [%rd28+(4+i*4)];
add.f32 %f(135+i*3), %f(193+i*-1), %f(134+i*3);
max.f32 %f(136+i*3), %f(135+i*3), %f132;
st.global.f32 [%rd29+60], %f178;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[49];
__shared__ float placeholder_shared[6272];
for (int j_inner_init = 0; j_inner_init < 16; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 16; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 7; ++ax0_ax1_fused_outer_outer) {
if (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x < 49) {
int index1 = ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
int index2 = k_outer_outer * 49 + ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 392; ++ax0_ax1_fused_outer_outer1) {
int2 _1;
int2 _2;
int2 _3;
int2 _4;
int2 _5 = make_int2(ax0_ax1_fused_outer_outer1 * 16 + (int)threadIdx.x * 2+1*0, ax0_ax1_fused_outer_outer1 * 16 + (int)threadIdx.x * 2+1*1);
int2 _6 = make_int2(49, 49);
_4.x = _5.x%_6.x;
_4.y = _5.y%_6.y;
int2 _7;
int2 _8 = make_int2(ax0_ax1_fused_outer_outer1 * 16 + (int)threadIdx.x * 2+1*0, ax0_ax1_fused_outer_outer1 * 16 + (int)threadIdx.x * 2+1*1);
int2 _9 = make_int2(49, 49);
_7.x = _8.x/_9.x;
_7.y = _8.y/_9.y;
int2 _10;
ushort2 _11;
ushort2 _12;
ushort2 _13;
int2 _14 = make_int2(49, 49);
int2 _15 = make_int2(0, 0);
_13.x = _14.x>=_15.x;
_13.y = _14.y>=_15.y;
ushort2 _16;
int2 _17 = make_int2(0, 0);
_16.x = _4.x>=_17.x;
_16.y = _4.y>=_17.y;
_12.x = _13.x&&_16.x;
_12.y = _13.y&&_16.y;
ushort2 _18;
ushort2 _19;
int2 _20 = make_int2(49, 49);
int2 _21 = make_int2(0, 0);
_19.x = _20.x<_21.x;
_19.y = _20.y<_21.y;
ushort2 _22;
int2 _23 = make_int2(0, 0);
_22.x = _4.x<=_23.x;
_22.y = _4.y<=_23.y;
_18.x = _19.x&&_22.x;
_18.y = _19.y&&_22.y;
_11.x = _12.x||_18.x;
_11.y = _12.y||_18.y;
int2 _24;
int2 _25 = make_int2(1, 1);
_24.x = _7.x-_25.x;
_24.y = _7.y-_25.y;
_10.x = bool(_11.x)?_7.x:_24.x;
_10.y = bool(_11.y)?_7.y:_24.y;
int2 _26 = make_int2(784, 784);
_3.x = _10.x*_26.x;
_3.y = _10.y*_26.y;
int2 _27 = make_int2(k_outer_outer * 49, k_outer_outer * 49);
_2.x = _3.x+_27.x;
_2.y = _3.y+_27.y;
int2 _28;
int2 _29 = make_int2(ax0_ax1_fused_outer_outer1 * 16 + (int)threadIdx.x * 2+1*0, ax0_ax1_fused_outer_outer1 * 16 + (int)threadIdx.x * 2+1*1);
int2 _30 = make_int2(49, 49);
_28.x = _29.x%_30.x;
_28.y = _29.y%_30.y;
int2 _31;
ushort2 _32;
ushort2 _33;
ushort2 _34;
int2 _35 = make_int2(49, 49);
int2 _36 = make_int2(0, 0);
_34.x = _35.x>=_36.x;
_34.y = _35.y>=_36.y;
ushort2 _37;
int2 _38 = make_int2(0, 0);
_37.x = _28.x>=_38.x;
_37.y = _28.y>=_38.y;
_33.x = _34.x&&_37.x;
_33.y = _34.y&&_37.y;
ushort2 _39;
ushort2 _40;
int2 _41 = make_int2(49, 49);
int2 _42 = make_int2(0, 0);
_40.x = _41.x<_42.x;
_40.y = _41.y<_42.y;
ushort2 _43;
int2 _44 = make_int2(0, 0);
_43.x = _28.x<=_44.x;
_43.y = _28.y<=_44.y;
_39.x = _40.x&&_43.x;
_39.y = _40.y&&_43.y;
_32.x = _33.x||_39.x;
_32.y = _33.y||_39.y;
int2 _45;
int2 _46 = make_int2(49, 49);
_45.x = _28.x+_46.x;
_45.y = _28.y+_46.y;
_31.x = bool(_32.x)?_28.x:_45.x;
_31.y = bool(_32.y)?_28.y:_45.y;
_1.x = _2.x+_31.x;
_1.y = _2.y+_31.y;
int index3 = _1.x;
int index4 = _1.y;
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer1 * 16 + (int)threadIdx.x * 2))[0] = make_float2(placeholder1[index3],placeholder1[index4]);
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 49; ++k_outer_inner) {
for (int j_inner = 0; j_inner < 16; ++j_inner) {
int index5 = j_inner;
int index6 = k_outer_inner;
int index7 = (int)threadIdx.x * 784 + j_inner * 49 + k_outer_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
for (int ax1_inner = 0; ax1_inner < 16; ++ax1_inner) {
int index8 = (int)threadIdx.x * 16 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpmv3rkhgw | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<158>;
.reg .b32 %r<121>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[112];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[1792];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(2+i*20), %r1, (2+i*2);
mov.u32 %r23, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r23, %r22;
mov.u32 %r4, %ctaid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(5+i*1), %r(4+i*-3), (12544+i*-12432);
add.s32 %r7, %r1, 14;
mov.u32 %r24, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r8, %r24, %r2;
mov.u32 %r116, 0;
mov.f32 %f142, 0f00000000;
for.size.1 i in range(0, 7, 1):
mov.f32 %f(143+i*1), %f142;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r10, %r116, 28;
add.s32 %r25, %r10, %r2;
mul.wide.s32 %rd10, %r25, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.3 i in range(0, 3, 1):
for.size.2 j in range(0, 4, 1):
ld.global.nc.f32 %f(33+i*4+j*1), [%rd5+(0+i*32+j*4)];
st.shared.f32 [%r3+(0+i*32+j*4)], %f(33+i*4+j*1);
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(45+i*1), [%rd5+(96+i*4)];
st.shared.f32 [%r3+(96+i*4)], %f(45+i*1);
$L__BB0_3:
add.s32 %r11, %r10, %r5;
mov.u32 %r119, 32;
mov.u32 %r117, %r1;
mov.u32 %r118, %r7;
$L__BB0_4:
mul.hi.s32 %r27, %r117, -1840700269;
add.s32 %r28, %r27, %r117;
shr.u32 %r29, %r28, 31;
shr.s32 %r30, %r28, 4;
add.s32 %r31, %r30, %r29;
mul.lo.s32 %r32, %r31, 28;
sub.s32 %r33, %r117, %r32;
add.s32 %r34, %r11, %r33;
mad.lo.s32 %r35, %r31, 784, %r34;
mul.wide.s32 %rd11, %r35, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f49, [%rd12];
add.s32 %r36, %r8, %r119;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r36+(-32+i*8)], %f(49+i*1);
add.s32 %r(37+i*10), %r117, (2+i*2);
mul.hi.s32 %r(38+i*10), %r(37+i*10), -1840700269;
add.s32 %r(39+i*10), %r(38+i*10), %r(37+i*10);
shr.u32 %r(40+i*10), %r(39+i*10), 31;
shr.s32 %r(41+i*10), %r(39+i*10), 4;
add.s32 %r(42+i*10), %r(41+i*10), %r(40+i*10);
mul.lo.s32 %r(43+i*10), %r(42+i*10), 28;
sub.s32 %r(44+i*10), %r(37+i*10), %r(43+i*10);
add.s32 %r(45+i*10), %r11, %r(44+i*10);
mad.lo.s32 %r(46+i*10), %r(42+i*10), 784, %r(45+i*10);
mul.wide.s32 %rd(13+i*2), %r(46+i*10), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(50+i*1), [%rd(14+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r36+(0+i*8)], %f(53+i*1);
add.s32 %r(77+i*10), %r118, (-4+i*2);
mul.hi.s32 %r(78+i*10), %r(77+i*10), -1840700269;
add.s32 %r(79+i*10), %r(78+i*10), %r(77+i*10);
shr.u32 %r(80+i*10), %r(79+i*10), 31;
shr.s32 %r(81+i*10), %r(79+i*10), 4;
add.s32 %r(82+i*10), %r(81+i*10), %r(80+i*10);
mul.lo.s32 %r(83+i*10), %r(82+i*10), 28;
sub.s32 %r(84+i*10), %r(77+i*10), %r(83+i*10);
add.s32 %r(85+i*10), %r11, %r(84+i*10);
mad.lo.s32 %r(86+i*10), %r(82+i*10), 784, %r(85+i*10);
mul.wide.s32 %rd(21+i*2), %r(86+i*10), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(54+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r36+16], %f55;
mul.hi.s32 %r97, %r118, -1840700269;
add.s32 %r98, %r97, %r118;
shr.u32 %r99, %r98, 31;
shr.s32 %r100, %r98, 4;
add.s32 %r101, %r100, %r99;
mul.lo.s32 %r102, %r101, 28;
sub.s32 %r103, %r118, %r102;
add.s32 %r104, %r11, %r103;
mad.lo.s32 %r105, %r101, 784, %r104;
mul.wide.s32 %rd25, %r105, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f56, [%rd26];
st.shared.f32 [%r36+24], %f56;
add.s32 %r119, %r119, 64;
setp.ne.s32 %p2, %r119, 1824;
for.size.1 i in range(0, 2, 1):
add.s32 %r(16+i*101), %r118, (16+i*-14);
mov.u32 %r118, %r16;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r120, 0;
$L__BB0_6:
add.s32 %r107, %r120, %r6;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(108+i*3), %r(107+i*13), 2;
add.s32 %r(110+i*3), %r(24+i*-1), %r(108+i*3);
ld.shared.f32 %f(57+i*1), [%r(110+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(59+i*2), %f58, %f(57+i*3), %f(149+i*-1);
ld.shared.f32 %f(60+i*2), [%r110+(896+i*-784)];
for.size.3 i in range(0, 2, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(63+i*4+j*2), %f58, %f(62+i*4+j*2), %f(147+i*-2+j*-1);
ld.shared.f32 %f(64+i*4+j*2), [%r110+(1008+i*112+j*-784)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(71+i*2), %f58, %f(70+i*2), %f(143+i*-1);
ld.shared.f32 %f(72+i*2), [%r110+(1232+i*-1228)];
for.size.17 i in range(0, 2, 1):
ld.shared.f32 %f(75+i*17), [%r113+(4+i*4)];
fma.rn.f32 %f(76+i*17), %f(75+i*17), %f(74+i*17), %f(59+i*17);
ld.shared.f32 %f(77+i*17), [%r110+(900+i*4)];
fma.rn.f32 %f(78+i*17), %f(75+i*17), %f(77+i*17), %f(61+i*17);
ld.shared.f32 %f(79+i*17), [%r110+(116+i*4)];
fma.rn.f32 %f(80+i*17), %f(75+i*17), %f(79+i*17), %f(63+i*17);
ld.shared.f32 %f(81+i*17), [%r110+(1012+i*4)];
fma.rn.f32 %f(82+i*17), %f(75+i*17), %f(81+i*17), %f(65+i*17);
ld.shared.f32 %f(83+i*17), [%r110+(228+i*4)];
fma.rn.f32 %f(84+i*17), %f(75+i*17), %f(83+i*17), %f(67+i*17);
ld.shared.f32 %f(85+i*17), [%r110+(1124+i*4)];
fma.rn.f32 %f(86+i*17), %f(75+i*17), %f(85+i*17), %f(69+i*17);
ld.shared.f32 %f(87+i*17), [%r110+(340+i*4)];
fma.rn.f32 %f(88+i*17), %f(75+i*17), %f(87+i*17), %f(71+i*17);
ld.shared.f32 %f(89+i*17), [%r110+(1236+i*4)];
fma.rn.f32 %f(90+i*17), %f(75+i*17), %f(89+i*17), %f(73+i*17);
ld.shared.f32 %f(91+i*17), [%r110+(8+i*4)];
ld.shared.f32 %f109, [%r113+12];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(149+i*-1), %f109, %f(108+i*2), %f(93+i*2);
ld.shared.f32 %f(110+i*1), [%r110+(908+i*-784)];
for.size.3 i in range(0, 2, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(147+i*-2+j*-1), %f109, %f(111+i*2+j*1), %f(97+i*4+j*2);
ld.shared.f32 %f(112+i*2+j*1), [%r110+(1020+i*112+j*-784)];
fma.rn.f32 %f143, %f109, %f115, %f105;
ld.shared.f32 %f116, [%r110+1244];
fma.rn.f32 %f142, %f109, %f116, %f107;
add.s32 %r120, %r120, 4;
setp.ne.s32 %p3, %r120, 28;
@%p3 bra $L__BB0_6;
add.s32 %r116, %r116, 1;
setp.lt.u32 %p4, %r116, 28;
@%p4 bra $L__BB0_1;
shl.b32 %r114, %r4, 4;
add.s32 %r115, %r114, %r2;
mul.wide.s32 %rd27, %r115, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f117, [%rd28];
add.f32 %f118, %f149, %f117;
mov.f32 %f119, 0f00000000;
max.f32 %f120, %f118, %f119;
add.s64 %rd29, %rd3, %rd27;
for.size.5 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(0+i*4+j*32)], %f(120+i*6+j*3);
ld.global.nc.f32 %f(121+i*6+j*3), [%rd28+(32+i*4+j*-28)];
add.f32 %f(122+i*6+j*3), %f(148+i*-2+j*-1), %f(121+i*6+j*3);
max.f32 %f(123+i*6+j*3), %f(122+i*6+j*3), %f119;
st.global.f32 [%rd29+12], %f138;
ld.global.nc.f32 %f139, [%rd28+44];
add.f32 %f140, %f142, %f139;
max.f32 %f141, %f140, %f119;
st.global.f32 [%rd29+44], %f141;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[8];
__shared__ float placeholder_d_shared[28];
__shared__ float placeholder_shared[448];
T_dense[0] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 28; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 28 + (int)threadIdx.x * 4;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
placeholder_d_shared[index0 + 3] = placeholder[index1 + 3];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 9] = placeholder[index1 + 9];
placeholder_d_shared[index0 + 10] = placeholder[index1 + 10];
placeholder_d_shared[index0 + 11] = placeholder[index1 + 11];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 17] = placeholder[index1 + 17];
placeholder_d_shared[index0 + 18] = placeholder[index1 + 18];
placeholder_d_shared[index0 + 19] = placeholder[index1 + 19];
if ((int)threadIdx.x < 1) {
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
}
if ((int)threadIdx.x < 1) {
placeholder_d_shared[index0 + 25] = placeholder[index1 + 25];
}
if ((int)threadIdx.x < 1) {
placeholder_d_shared[index0 + 26] = placeholder[index1 + 26];
}
if ((int)threadIdx.x < 1) {
placeholder_d_shared[index0 + 27] = placeholder[index1 + 27];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 224; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 12544 + (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x) / 28 * 784 + k_outer_outer * 28 + (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x) % 28;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 28; ++k_outer_inner) {
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 112 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 224]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 28]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 252]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 56]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 280]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 84]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 308]);
}
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index6 = (int)blockIdx.x * 16 + (int)threadIdx.x * 4 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 8] = max(T_dense[index7 + 4] + placeholder2[index6 + 8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpa0x1jr28 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<659>;
.reg .b32 %r<70>;
.reg .b64 %rd<23>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd11, [default_function_kernel0_param_1];
cvta.to.global.u64 %rd21, %rd10;
cvta.to.global.u64 %rd2, %rd11;
mov.u32 %r66, 0;
mov.u32 %r65, _ZZ24default_function_kernel0E20placeholder_d_shared;
$L__BB0_1:
for.size.2 i in range(0, 8, 1):
ld.global.nc.v2.u32 {%r(13+i*4), %r(14+i*4)}, [%rd21+(0+i*8)];
st.shared.v2.u32 [%r65+(0+i*8)], {%r(13+i*4), %r(14+i*4)};
add.s32 %r65, %r65, 64;
add.s64 %rd21, %rd21, 64;
add.s32 %r66, %r66, 8;
setp.ne.s32 %p1, %r66, 392;
@%p1 bra $L__BB0_1;
mov.u32 %r46, %ctaid.x;
mul.lo.s32 %r47, %r46, 50176;
mul.wide.s32 %rd12, %r47, 4;
add.s64 %rd13, %rd2, %rd12;
add.s64 %rd22, %rd13, 32;
mov.u32 %r67, 32;
$L__BB0_3:
ld.global.nc.f32 %f193, [%rd22+-32];
mov.u32 %r48, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r49, %r48, %r67;
for.size.2 i in range(0, 15, 1):
st.shared.f32 [%r49+(-32+i*4)], %f(193+i*1);
ld.global.nc.f32 %f(194+i*1), [%rd22+(-28+i*4)];
st.shared.f32 [%r49+28], %f208;
add.s64 %rd22, %rd22, 64;
add.s32 %r67, %r67, 64;
setp.ne.s32 %p2, %r67, 200736;
@%p2 bra $L__BB0_3;
bar.sync 0;
mov.u32 %r50, 0;
mov.f32 %f595, 0f00000000;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(596+i*1), %f595;
mov.u32 %r68, %r50;
$L__BB0_5:
mov.u32 %r69, %r50;
$L__BB0_6:
mov.u32 %r64, _ZZ24default_function_kernel0E18placeholder_shared;
shl.b32 %r63, %r68, 1;
add.s32 %r53, %r69, %r63;
shl.b32 %r54, %r53, 2;
mov.u32 %r55, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(56+i*2), %r(55+i*9), %r54;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(273+i*1), [%r(58+i*-2)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(658+i*-1), %f274, %f(273+i*2), %f(658+i*-1);
ld.shared.f32 %f(275+i*1), [%r58+(3136+i*3136)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(656+i*-1), %f274, %f(276+i*1), %f(656+i*-1);
ld.shared.f32 %f(277+i*1), [%r58+(9408+i*3136)];
fma.rn.f32 %f595, %f274, %f337, %f595;
add.s32 %r59, %r69, 1;
mov.u32 %r69, 1;
setp.lt.u32 %p3, %r59, 2;
@%p3 bra $L__BB0_6;
add.s32 %r68, %r68, 1;
setp.lt.u32 %p4, %r68, 392;
@%p4 bra $L__BB0_5;
ld.param.u64 %rd20, [default_function_kernel0_param_2];
ld.param.u64 %rd19, [default_function_kernel0_param_3];
mov.u32 %r62, %ctaid.x;
shl.b32 %r61, %r62, 6;
cvta.to.global.u64 %rd14, %rd19;
mul.wide.s32 %rd15, %r61, 4;
add.s64 %rd16, %rd14, %rd15;
ld.global.nc.f32 %f338, [%rd16];
add.f32 %f339, %f658, %f338;
mov.f32 %f340, 0f00000000;
max.f32 %f341, %f339, %f340;
cvta.to.global.u64 %rd17, %rd20;
add.s64 %rd18, %rd17, %rd15;
for.size.4 i in range(0, 63, 1):
st.global.f32 [%rd18+(0+i*4)], %f(341+i*3);
ld.global.nc.f32 %f(342+i*3), [%rd16+(4+i*4)];
add.f32 %f(343+i*3), %f(657+i*-1), %f(342+i*3);
max.f32 %f(344+i*3), %f(343+i*3), %f340;
st.global.f32 [%rd18+252], %f530;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[50176];
for (int j_inner_init = 0; j_inner_init < 64; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 392; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_d_shared + ax0_ax1_fused_outer_outer * 2))[0] = ((float2*)(placeholder + ax0_ax1_fused_outer_outer * 2))[0];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 50176; ++ax0_ax1_fused_outer_outer1) {
int index1 = ax0_ax1_fused_outer_outer1;
int index2 = (int)blockIdx.x * 50176 + ax0_ax1_fused_outer_outer1;
placeholder_shared[index1] = placeholder1[index2];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 392; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 2; ++k_inner) {
for (int j_inner = 0; j_inner < 64; ++j_inner) {
int index3 = j_inner;
int index4 = k_outer_inner * 2 + k_inner;
int index5 = j_inner * 784 + k_outer_inner * 2 + k_inner;
T_dense[index3] = T_dense[index3] + placeholder_d_shared[index4] * placeholder_shared[index5];
}
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index6 = (int)blockIdx.x * 64 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpmrzin8gz | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<389>;
.reg .b32 %r<22>;
.reg .b64 %rd<16>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[8];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[1024];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd7, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd7;
cvta.to.global.u64 %rd3, %rd5;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r1, %tid.x;
shl.b32 %r9, %r1, 2;
mov.u32 %r10, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r10, %r9;
shr.s32 %r11, %r1, 1;
mul.lo.s32 %r12, %r11, 784;
and.b32 %r13, %r1, 1;
or.b32 %r3, %r12, %r13;
mov.u32 %r14, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r4, %r14, %r9;
shl.b32 %r15, %r1, 8;
add.s32 %r5, %r14, %r15;
mov.u32 %r21, 0;
mov.f32 %f357, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(358+i*1), %f357;
$L__BB0_1:
bar.sync 0;
setp.lt.s32 %p1, %r1, 2;
@%p1 bra $L__BB0_2;
bra.uni $L__BB0_3;
$L__BB0_2:
shl.b32 %r16, %r21, 1;
add.s32 %r17, %r16, %r1;
mul.wide.s32 %rd9, %r17, 4;
add.s64 %rd10, %rd3, %rd9;
ld.global.nc.f32 %f97, [%rd10];
st.shared.f32 [%r2], %f97;
$L__BB0_3:
shl.b32 %r18, %r21, 1;
add.s32 %r19, %r3, %r18;
mul.wide.s32 %rd11, %r19, 4;
add.s64 %rd12, %rd1, %rd11;
for.size.2 i in range(0, 64, 1):
ld.global.nc.f32 %f(98+i*1), [%rd12+(0+i*6272)];
st.shared.f32 [%r4+(0+i*16)], %f(98+i*1);
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(162+i*5), [%r5+(0+i*4)];
ld.shared.f32 %f(163+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(164+i*224), %f(163+i*5), %f(162+i*5), %f(388+i*-224);
ld.shared.f32 %f(165+i*4), [%r5+(8+i*4)];
fma.rn.f32 %f(166+i*221), %f(163+i*5), %f(165+i*4), %f(387+i*-221);
for.size.6 i in range(0, 15, 1):
for.size.2 j in range(0, 2, 1):
ld.shared.f32 %f(170+i*6+j*2), [%r5+(16+i*16+j*8)];
fma.rn.f32 %f(171+i*6+j*2), %f163, %f(170+i*6+j*2), %f(386+i*-2+j*-1);
for.size.2 j in range(0, 2, 1):
ld.shared.f32 %f(174+i*6+j*1), [%r5+(20+i*16+j*8)];
fma.rn.f32 %f(386+i*-2+j*-1), %f168, %f(174+i*6+j*1), %f(171+i*6+j*2);
add.s32 %r21, %r21, 1;
setp.lt.u32 %p2, %r21, 392;
@%p2 bra $L__BB0_1;
shl.b32 %r20, %r1, 5;
mul.wide.s32 %rd13, %r20, 4;
add.s64 %rd14, %rd4, %rd13;
ld.global.nc.f32 %f260, [%rd14];
add.f32 %f261, %f388, %f260;
mov.f32 %f262, 0f00000000;
max.f32 %f263, %f261, %f262;
add.s64 %rd15, %rd2, %rd13;
for.size.4 i in range(0, 31, 1):
st.global.f32 [%rd15+(0+i*4)], %f(263+i*3);
ld.global.nc.f32 %f(264+i*3), [%rd14+(4+i*4)];
add.f32 %f(265+i*3), %f(387+i*-1), %f(264+i*3);
max.f32 %f(266+i*3), %f(265+i*3), %f262;
st.global.f32 [%rd15+124], %f356;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[2];
__shared__ float placeholder_shared[256];
for (int j_outer_inner_init = 0; j_outer_inner_init < 16; ++j_outer_inner_init) {
for (int j_inner_init = 0; j_inner_init < 2; ++j_inner_init) {
int index0 = j_outer_inner_init * 2 + j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
}
for (int k_outer_outer = 0; k_outer_outer < 392; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 2) {
int index1 = (int)threadIdx.x;
int index2 = k_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 64; ++ax0_ax1_fused_outer_outer) {
int index3 = ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x;
int index4 = ax0_ax1_fused_outer_outer * 1568 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 2 + ((int)threadIdx.x & 1);
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int j_outer_inner = 0; j_outer_inner < 16; ++j_outer_inner) {
for (int k_inner = 0; k_inner < 2; ++k_inner) {
for (int j_inner = 0; j_inner < 2; ++j_inner) {
int index5 = j_outer_inner * 2 + j_inner;
int index6 = k_inner;
int index7 = (int)threadIdx.x * 64 + j_outer_inner * 4 + j_inner * 2 + k_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
}
for (int ax1_inner = 0; ax1_inner < 32; ++ax1_inner) {
int index8 = (int)threadIdx.x * 32 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpqndedhm_ | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<302>;
.reg .b32 %r<15>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[4];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[512];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd11, [default_function_kernel0_param_1];
ld.param.u64 %rd12, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd11;
cvta.to.global.u64 %rd1, %rd12;
mov.u32 %r1, %tid.x;
shl.b32 %r4, %r1, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r9, %r4;
mov.u32 %r10, _ZZ24default_function_kernel0E18placeholder_shared;
mad.lo.s32 %r3, %r1, 56, %r10;
shl.b32 %r11, %r1, 4;
add.s32 %r5, %r10, %r11;
mul.lo.s32 %r12, %r1, 784;
cvta.to.global.u64 %rd14, %rd10;
mul.wide.s32 %rd15, %r12, 4;
add.s64 %rd23, %rd14, %rd15;
mul.lo.s32 %r13, %r1, 10976;
mul.wide.s32 %rd16, %r13, 4;
add.s64 %rd17, %rd13, %rd16;
add.s64 %rd22, %rd17, 197568;
mov.u32 %r14, 0;
mov.f32 %f270, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(271+i*1), %f270;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f97, [%rd23];
st.shared.f32 [%r2], %f97;
$L__BB0_3:
setp.gt.s32 %p2, %r1, 1;
add.s64 %rd6, %rd22, -197568;
for.size.3 i in range(0, 2, 1):
for.size.2 j in range(0, 14, 1):
ld.global.nc.f32 %f(98+i*14+j*1), [%rd22+(-197568+i*175616+j*3136)];
st.shared.f32 [%r3+(0+i*224+j*4)], %f(98+i*14+j*1);
@%p2 bra $L__BB0_5;
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(126+i*1), [%rd6+(351232+i*3136)];
st.shared.f32 [%r3+(448+i*4)], %f(126+i*1);
$L__BB0_5:
@%p1 bra $L__BB0_7;
for.size.2 i in range(0, 12, 1):
ld.global.nc.f32 %f(128+i*1), [%rd6+(357504+i*3136)];
st.shared.f32 [%r3+(456+i*4)], %f(128+i*1);
$L__BB0_7:
bar.sync 0;
ld.shared.f32 %f140, [%r5];
ld.shared.f32 %f141, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(301+i*-1), %f141, %f(140+i*2), %f(301+i*-1);
ld.shared.f32 %f(142+i*1), [%r5+(64+i*64)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(299+i*-1), %f141, %f(143+i*1), %f(299+i*-1);
ld.shared.f32 %f(144+i*1), [%r5+(192+i*64)];
for.size.3 i in range(0, 3, 1):
for.size.2 j in range(0, 8, 1):
fma.rn.f32 %f(294+i*-8+j*-1), %f141, %f(148+i*8+j*1), %f(294+i*-8+j*-1);
ld.shared.f32 %f(149+i*8+j*1), [%r5+(4+i*4+j*64)];
fma.rn.f32 %f270, %f141, %f172, %f270;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(23+i*-1), %rd(23+i*-1), 4;
add.s32 %r14, %r14, 1;
setp.ne.s32 %p4, %r14, 784;
@%p4 bra $L__BB0_1;
cvta.to.global.u64 %rd18, %rd9;
mul.wide.s32 %rd19, %r4, 4;
add.s64 %rd20, %rd18, %rd19;
ld.global.nc.f32 %f173, [%rd20];
add.f32 %f174, %f301, %f173;
mov.f32 %f175, 0f00000000;
max.f32 %f176, %f174, %f175;
add.s64 %rd21, %rd1, %rd19;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd21+(0+i*64)], %f(176+i*3);
ld.global.nc.f32 %f(177+i*3), [%rd20+(64+i*64)];
add.f32 %f(178+i*3), %f(300+i*-1), %f(177+i*3);
max.f32 %f(179+i*3), %f(178+i*3), %f175;
for.size.10 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd21+(448+i*4+j*-444)], %f(197+i*24+j*3);
ld.global.nc.f32 %f(198+i*24+j*3), [%rd20+(4+i*4+j*64)];
add.f32 %f(199+i*24+j*3), %f(293+i*-8+j*-1), %f(198+i*24+j*3);
max.f32 %f(200+i*24+j*3), %f(199+i*24+j*3), %f175;
for.size.4 j in range(0, 6, 1):
st.global.f32 [%rd21+(68+i*4+j*64)], %f(203+i*24+j*3);
ld.global.nc.f32 %f(204+i*24+j*3), [%rd20+(132+i*4+j*64)];
add.f32 %f(205+i*24+j*3), %f(291+i*-8+j*-1), %f(204+i*24+j*3);
max.f32 %f(206+i*24+j*3), %f(205+i*24+j*3), %f175;
st.global.f32 [%rd21+460], %f269;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[128];
T_dense[0] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index2 = (int)threadIdx.x * 10976 + k_outer_outer;
placeholder_shared[index0 * 14] = placeholder1[index2];
int index3 = (int)threadIdx.x * 14 + 1;
placeholder_shared[index3] = placeholder1[index2 + 784];
int index4 = (int)threadIdx.x * 14 + 2;
placeholder_shared[index4] = placeholder1[index2 + 1568];
int index5 = (int)threadIdx.x * 14 + 3;
placeholder_shared[index5] = placeholder1[index2 + 2352];
int index6 = (int)threadIdx.x * 14 + 4;
placeholder_shared[index6] = placeholder1[index2 + 3136];
int index7 = (int)threadIdx.x * 14 + 5;
placeholder_shared[index7] = placeholder1[index2 + 3920];
int index8 = (int)threadIdx.x * 14 + 6;
placeholder_shared[index8] = placeholder1[index2 + 4704];
int index9 = (int)threadIdx.x * 14 + 7;
placeholder_shared[index9] = placeholder1[index2 + 5488];
int index10 = (int)threadIdx.x * 14 + 8;
placeholder_shared[index10] = placeholder1[index2 + 6272];
int index11 = (int)threadIdx.x * 14 + 9;
placeholder_shared[index11] = placeholder1[index2 + 7056];
int index12 = (int)threadIdx.x * 14 + 10;
placeholder_shared[index12] = placeholder1[index2 + 7840];
int index13 = (int)threadIdx.x * 14 + 11;
placeholder_shared[index13] = placeholder1[index2 + 8624];
int index14 = (int)threadIdx.x * 14 + 12;
placeholder_shared[index14] = placeholder1[index2 + 9408];
int index15 = (int)threadIdx.x * 14 + 13;
placeholder_shared[index15] = placeholder1[index2 + 10192];
int index16 = (int)threadIdx.x * 14 + 56;
placeholder_shared[index16] = placeholder1[index2 + 43904];
int index17 = (int)threadIdx.x * 14 + 57;
placeholder_shared[index17] = placeholder1[index2 + 44688];
int index18 = (int)threadIdx.x * 14 + 58;
placeholder_shared[index18] = placeholder1[index2 + 45472];
int index19 = (int)threadIdx.x * 14 + 59;
placeholder_shared[index19] = placeholder1[index2 + 46256];
int index20 = (int)threadIdx.x * 14 + 60;
placeholder_shared[index20] = placeholder1[index2 + 47040];
int index21 = (int)threadIdx.x * 14 + 61;
placeholder_shared[index21] = placeholder1[index2 + 47824];
int index22 = (int)threadIdx.x * 14 + 62;
placeholder_shared[index22] = placeholder1[index2 + 48608];
int index23 = (int)threadIdx.x * 14 + 63;
placeholder_shared[index23] = placeholder1[index2 + 49392];
int index24 = (int)threadIdx.x * 14 + 64;
placeholder_shared[index24] = placeholder1[index2 + 50176];
int index25 = (int)threadIdx.x * 14 + 65;
placeholder_shared[index25] = placeholder1[index2 + 50960];
int index26 = (int)threadIdx.x * 14 + 66;
placeholder_shared[index26] = placeholder1[index2 + 51744];
int index27 = (int)threadIdx.x * 14 + 67;
placeholder_shared[index27] = placeholder1[index2 + 52528];
int index28 = (int)threadIdx.x * 14 + 68;
placeholder_shared[index28] = placeholder1[index2 + 53312];
int index29 = (int)threadIdx.x * 14 + 69;
placeholder_shared[index29] = placeholder1[index2 + 54096];
if ((int)threadIdx.x < 2) {
int index30 = (int)threadIdx.x * 14 + 112;
placeholder_shared[index30] = placeholder1[index2 + 87808];
}
if ((int)threadIdx.x < 2) {
int index31 = (int)threadIdx.x * 14 + 113;
placeholder_shared[index31] = placeholder1[index2 + 88592];
}
if ((int)threadIdx.x < 1) {
int index32 = (int)threadIdx.x * 14 + 114;
placeholder_shared[index32] = placeholder1[index2 + 89376];
}
if ((int)threadIdx.x < 1) {
int index33 = (int)threadIdx.x * 14 + 115;
placeholder_shared[index33] = placeholder1[index2 + 90160];
}
if ((int)threadIdx.x < 1) {
int index34 = (int)threadIdx.x * 14 + 116;
placeholder_shared[index34] = placeholder1[index2 + 90944];
}
if ((int)threadIdx.x < 1) {
int index35 = (int)threadIdx.x * 14 + 117;
placeholder_shared[index35] = placeholder1[index2 + 91728];
}
if ((int)threadIdx.x < 1) {
int index36 = (int)threadIdx.x * 14 + 118;
placeholder_shared[index36] = placeholder1[index2 + 92512];
}
if ((int)threadIdx.x < 1) {
int index37 = (int)threadIdx.x * 14 + 119;
placeholder_shared[index37] = placeholder1[index2 + 93296];
}
if ((int)threadIdx.x < 1) {
int index38 = (int)threadIdx.x * 14 + 120;
placeholder_shared[index38] = placeholder1[index2 + 94080];
}
if ((int)threadIdx.x < 1) {
int index39 = (int)threadIdx.x * 14 + 121;
placeholder_shared[index39] = placeholder1[index2 + 94864];
}
if ((int)threadIdx.x < 1) {
int index40 = (int)threadIdx.x * 14 + 122;
placeholder_shared[index40] = placeholder1[index2 + 95648];
}
if ((int)threadIdx.x < 1) {
int index41 = (int)threadIdx.x * 14 + 123;
placeholder_shared[index41] = placeholder1[index2 + 96432];
}
if ((int)threadIdx.x < 1) {
int index42 = (int)threadIdx.x * 14 + 124;
placeholder_shared[index42] = placeholder1[index2 + 97216];
}
if ((int)threadIdx.x < 1) {
int index43 = (int)threadIdx.x * 14 + 125;
placeholder_shared[index43] = placeholder1[index2 + 98000];
}
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 4];
int index44 = (int)threadIdx.x * 4 + 16;
T_dense[4] = T_dense[4] + placeholder_d_shared[0] * placeholder_shared[index44];
int index45 = (int)threadIdx.x * 4 + 32;
T_dense[8] = T_dense[8] + placeholder_d_shared[0] * placeholder_shared[index45];
int index46 = (int)threadIdx.x * 4 + 48;
T_dense[12] = T_dense[12] + placeholder_d_shared[0] * placeholder_shared[index46];
int index47 = (int)threadIdx.x * 4 + 64;
T_dense[16] = T_dense[16] + placeholder_d_shared[0] * placeholder_shared[index47];
int index48 = (int)threadIdx.x * 4 + 80;
T_dense[20] = T_dense[20] + placeholder_d_shared[0] * placeholder_shared[index48];
int index49 = (int)threadIdx.x * 4 + 96;
T_dense[24] = T_dense[24] + placeholder_d_shared[0] * placeholder_shared[index49];
int index50 = (int)threadIdx.x * 4 + 112;
T_dense[28] = T_dense[28] + placeholder_d_shared[0] * placeholder_shared[index50];
int index51 = (int)threadIdx.x * 4 + 1;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index51];
int index52 = (int)threadIdx.x * 4 + 17;
T_dense[5] = T_dense[5] + placeholder_d_shared[0] * placeholder_shared[index52];
int index53 = (int)threadIdx.x * 4 + 33;
T_dense[9] = T_dense[9] + placeholder_d_shared[0] * placeholder_shared[index53];
int index54 = (int)threadIdx.x * 4 + 49;
T_dense[13] = T_dense[13] + placeholder_d_shared[0] * placeholder_shared[index54];
int index55 = (int)threadIdx.x * 4 + 65;
T_dense[17] = T_dense[17] + placeholder_d_shared[0] * placeholder_shared[index55];
int index56 = (int)threadIdx.x * 4 + 81;
T_dense[21] = T_dense[21] + placeholder_d_shared[0] * placeholder_shared[index56];
int index57 = (int)threadIdx.x * 4 + 97;
T_dense[25] = T_dense[25] + placeholder_d_shared[0] * placeholder_shared[index57];
int index58 = (int)threadIdx.x * 4 + 113;
T_dense[29] = T_dense[29] + placeholder_d_shared[0] * placeholder_shared[index58];
int index59 = (int)threadIdx.x * 4 + 2;
T_dense[2] = T_dense[2] + placeholder_d_shared[0] * placeholder_shared[index59];
int index60 = (int)threadIdx.x * 4 + 18;
T_dense[6] = T_dense[6] + placeholder_d_shared[0] * placeholder_shared[index60];
int index61 = (int)threadIdx.x * 4 + 34;
T_dense[10] = T_dense[10] + placeholder_d_shared[0] * placeholder_shared[index61];
int index62 = (int)threadIdx.x * 4 + 50;
T_dense[14] = T_dense[14] + placeholder_d_shared[0] * placeholder_shared[index62];
int index63 = (int)threadIdx.x * 4 + 66;
T_dense[18] = T_dense[18] + placeholder_d_shared[0] * placeholder_shared[index63];
int index64 = (int)threadIdx.x * 4 + 82;
T_dense[22] = T_dense[22] + placeholder_d_shared[0] * placeholder_shared[index64];
int index65 = (int)threadIdx.x * 4 + 98;
T_dense[26] = T_dense[26] + placeholder_d_shared[0] * placeholder_shared[index65];
int index66 = (int)threadIdx.x * 4 + 114;
T_dense[30] = T_dense[30] + placeholder_d_shared[0] * placeholder_shared[index66];
int index67 = (int)threadIdx.x * 4 + 3;
T_dense[3] = T_dense[3] + placeholder_d_shared[0] * placeholder_shared[index67];
int index68 = (int)threadIdx.x * 4 + 19;
T_dense[7] = T_dense[7] + placeholder_d_shared[0] * placeholder_shared[index68];
int index69 = (int)threadIdx.x * 4 + 35;
T_dense[11] = T_dense[11] + placeholder_d_shared[0] * placeholder_shared[index69];
int index70 = (int)threadIdx.x * 4 + 51;
T_dense[15] = T_dense[15] + placeholder_d_shared[0] * placeholder_shared[index70];
int index71 = (int)threadIdx.x * 4 + 67;
T_dense[19] = T_dense[19] + placeholder_d_shared[0] * placeholder_shared[index71];
int index72 = (int)threadIdx.x * 4 + 83;
T_dense[23] = T_dense[23] + placeholder_d_shared[0] * placeholder_shared[index72];
int index73 = (int)threadIdx.x * 4 + 99;
T_dense[27] = T_dense[27] + placeholder_d_shared[0] * placeholder_shared[index73];
int index74 = (int)threadIdx.x * 4 + 115;
T_dense[31] = T_dense[31] + placeholder_d_shared[0] * placeholder_shared[index74];
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index75 = (int)threadIdx.x * 4 + ax1_inner;
int index76 = ax1_inner;
T_relu[index75] = max(T_dense[index76] + placeholder2[index75], 0.000000e+00f);
T_relu[index75 + 16] = max(T_dense[index76 + 4] + placeholder2[index75 + 16], 0.000000e+00f);
T_relu[index75 + 32] = max(T_dense[index76 + 8] + placeholder2[index75 + 32], 0.000000e+00f);
T_relu[index75 + 48] = max(T_dense[index76 + 12] + placeholder2[index75 + 48], 0.000000e+00f);
T_relu[index75 + 64] = max(T_dense[index76 + 16] + placeholder2[index75 + 64], 0.000000e+00f);
T_relu[index75 + 80] = max(T_dense[index76 + 20] + placeholder2[index75 + 80], 0.000000e+00f);
T_relu[index75 + 96] = max(T_dense[index76 + 24] + placeholder2[index75 + 96], 0.000000e+00f);
T_relu[index75 + 112] = max(T_dense[index76 + 28] + placeholder2[index75 + 112], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpl5meaozn | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<295>;
.reg .b32 %r<15>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[4];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[512];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd11, [default_function_kernel0_param_1];
ld.param.u64 %rd12, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd11;
cvta.to.global.u64 %rd1, %rd12;
mov.u32 %r1, %tid.x;
shl.b32 %r4, %r1, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r9, %r4;
mov.u32 %r10, _ZZ24default_function_kernel0E18placeholder_shared;
mad.lo.s32 %r3, %r1, 28, %r10;
shl.b32 %r11, %r1, 4;
add.s32 %r5, %r10, %r11;
mul.lo.s32 %r12, %r1, 784;
cvta.to.global.u64 %rd14, %rd10;
mul.wide.s32 %rd15, %r12, 4;
add.s64 %rd23, %rd14, %rd15;
mul.lo.s32 %r13, %r1, 5488;
mul.wide.s32 %rd16, %r13, 4;
add.s64 %rd17, %rd13, %rd16;
add.s64 %rd22, %rd17, 185024;
mov.u32 %r14, 0;
mov.f32 %f263, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(264+i*1), %f263;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f97, [%rd23];
st.shared.f32 [%r2], %f97;
$L__BB0_3:
setp.gt.s32 %p2, %r1, 2;
add.s64 %rd6, %rd22, -185024;
for.size.3 i in range(0, 4, 1):
for.size.2 j in range(0, 7, 1):
ld.global.nc.f32 %f(98+i*7+j*1), [%rd22+(-185024+i*87808+j*3136)];
st.shared.f32 [%r3+(0+i*112+j*4)], %f(98+i*7+j*1);
@%p2 bra $L__BB0_5;
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(126+i*1), [%rd6+(351232+i*3136)];
st.shared.f32 [%r3+(448+i*4)], %f(126+i*1);
$L__BB0_5:
setp.gt.s32 %p3, %r1, 1;
@%p3 bra $L__BB0_7;
for.size.2 i in range(0, 5, 1):
ld.global.nc.f32 %f(128+i*1), [%rd6+(357504+i*3136)];
st.shared.f32 [%r3+(456+i*4)], %f(128+i*1);
$L__BB0_7:
bar.sync 0;
ld.shared.f32 %f133, [%r5];
ld.shared.f32 %f134, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(294+i*-1), %f134, %f(133+i*2), %f(294+i*-1);
ld.shared.f32 %f(135+i*1), [%r5+(64+i*64)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(292+i*-1), %f134, %f(136+i*1), %f(292+i*-1);
ld.shared.f32 %f(137+i*1), [%r5+(192+i*64)];
for.size.3 i in range(0, 3, 1):
for.size.2 j in range(0, 8, 1):
fma.rn.f32 %f(287+i*-8+j*-1), %f134, %f(141+i*8+j*1), %f(287+i*-8+j*-1);
ld.shared.f32 %f(142+i*8+j*1), [%r5+(4+i*4+j*64)];
fma.rn.f32 %f263, %f134, %f165, %f263;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(23+i*-1), %rd(23+i*-1), 4;
add.s32 %r14, %r14, 1;
setp.ne.s32 %p4, %r14, 784;
@%p4 bra $L__BB0_1;
cvta.to.global.u64 %rd18, %rd9;
mul.wide.s32 %rd19, %r4, 4;
add.s64 %rd20, %rd18, %rd19;
ld.global.nc.f32 %f166, [%rd20];
add.f32 %f167, %f294, %f166;
mov.f32 %f168, 0f00000000;
max.f32 %f169, %f167, %f168;
add.s64 %rd21, %rd1, %rd19;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd21+(0+i*64)], %f(169+i*3);
ld.global.nc.f32 %f(170+i*3), [%rd20+(64+i*64)];
add.f32 %f(171+i*3), %f(293+i*-1), %f(170+i*3);
max.f32 %f(172+i*3), %f(171+i*3), %f168;
for.size.10 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd21+(448+i*4+j*-444)], %f(190+i*24+j*3);
ld.global.nc.f32 %f(191+i*24+j*3), [%rd20+(4+i*4+j*64)];
add.f32 %f(192+i*24+j*3), %f(286+i*-8+j*-1), %f(191+i*24+j*3);
max.f32 %f(193+i*24+j*3), %f(192+i*24+j*3), %f168;
for.size.4 j in range(0, 6, 1):
st.global.f32 [%rd21+(68+i*4+j*64)], %f(196+i*24+j*3);
ld.global.nc.f32 %f(197+i*24+j*3), [%rd20+(132+i*4+j*64)];
add.f32 %f(198+i*24+j*3), %f(284+i*-8+j*-1), %f(197+i*24+j*3);
max.f32 %f(199+i*24+j*3), %f(198+i*24+j*3), %f168;
st.global.f32 [%rd21+460], %f262;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[128];
T_dense[0] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index2 = (int)threadIdx.x * 5488 + k_outer_outer;
placeholder_shared[index0 * 7] = placeholder1[index2];
int index3 = (int)threadIdx.x * 7 + 1;
placeholder_shared[index3] = placeholder1[index2 + 784];
int index4 = (int)threadIdx.x * 7 + 2;
placeholder_shared[index4] = placeholder1[index2 + 1568];
int index5 = (int)threadIdx.x * 7 + 3;
placeholder_shared[index5] = placeholder1[index2 + 2352];
int index6 = (int)threadIdx.x * 7 + 4;
placeholder_shared[index6] = placeholder1[index2 + 3136];
int index7 = (int)threadIdx.x * 7 + 5;
placeholder_shared[index7] = placeholder1[index2 + 3920];
int index8 = (int)threadIdx.x * 7 + 6;
placeholder_shared[index8] = placeholder1[index2 + 4704];
int index9 = (int)threadIdx.x * 7 + 28;
placeholder_shared[index9] = placeholder1[index2 + 21952];
int index10 = (int)threadIdx.x * 7 + 29;
placeholder_shared[index10] = placeholder1[index2 + 22736];
int index11 = (int)threadIdx.x * 7 + 30;
placeholder_shared[index11] = placeholder1[index2 + 23520];
int index12 = (int)threadIdx.x * 7 + 31;
placeholder_shared[index12] = placeholder1[index2 + 24304];
int index13 = (int)threadIdx.x * 7 + 32;
placeholder_shared[index13] = placeholder1[index2 + 25088];
int index14 = (int)threadIdx.x * 7 + 33;
placeholder_shared[index14] = placeholder1[index2 + 25872];
int index15 = (int)threadIdx.x * 7 + 34;
placeholder_shared[index15] = placeholder1[index2 + 26656];
int index16 = (int)threadIdx.x * 7 + 56;
placeholder_shared[index16] = placeholder1[index2 + 43904];
int index17 = (int)threadIdx.x * 7 + 57;
placeholder_shared[index17] = placeholder1[index2 + 44688];
int index18 = (int)threadIdx.x * 7 + 58;
placeholder_shared[index18] = placeholder1[index2 + 45472];
int index19 = (int)threadIdx.x * 7 + 59;
placeholder_shared[index19] = placeholder1[index2 + 46256];
int index20 = (int)threadIdx.x * 7 + 60;
placeholder_shared[index20] = placeholder1[index2 + 47040];
int index21 = (int)threadIdx.x * 7 + 61;
placeholder_shared[index21] = placeholder1[index2 + 47824];
int index22 = (int)threadIdx.x * 7 + 62;
placeholder_shared[index22] = placeholder1[index2 + 48608];
int index23 = (int)threadIdx.x * 7 + 84;
placeholder_shared[index23] = placeholder1[index2 + 65856];
int index24 = (int)threadIdx.x * 7 + 85;
placeholder_shared[index24] = placeholder1[index2 + 66640];
int index25 = (int)threadIdx.x * 7 + 86;
placeholder_shared[index25] = placeholder1[index2 + 67424];
int index26 = (int)threadIdx.x * 7 + 87;
placeholder_shared[index26] = placeholder1[index2 + 68208];
int index27 = (int)threadIdx.x * 7 + 88;
placeholder_shared[index27] = placeholder1[index2 + 68992];
int index28 = (int)threadIdx.x * 7 + 89;
placeholder_shared[index28] = placeholder1[index2 + 69776];
int index29 = (int)threadIdx.x * 7 + 90;
placeholder_shared[index29] = placeholder1[index2 + 70560];
if ((int)threadIdx.x < 3) {
int index30 = (int)threadIdx.x * 7 + 112;
placeholder_shared[index30] = placeholder1[index2 + 87808];
}
if ((int)threadIdx.x < 3) {
int index31 = (int)threadIdx.x * 7 + 113;
placeholder_shared[index31] = placeholder1[index2 + 88592];
}
if ((int)threadIdx.x < 2) {
int index32 = (int)threadIdx.x * 7 + 114;
placeholder_shared[index32] = placeholder1[index2 + 89376];
}
if ((int)threadIdx.x < 2) {
int index33 = (int)threadIdx.x * 7 + 115;
placeholder_shared[index33] = placeholder1[index2 + 90160];
}
if ((int)threadIdx.x < 2) {
int index34 = (int)threadIdx.x * 7 + 116;
placeholder_shared[index34] = placeholder1[index2 + 90944];
}
if ((int)threadIdx.x < 2) {
int index35 = (int)threadIdx.x * 7 + 117;
placeholder_shared[index35] = placeholder1[index2 + 91728];
}
if ((int)threadIdx.x < 2) {
int index36 = (int)threadIdx.x * 7 + 118;
placeholder_shared[index36] = placeholder1[index2 + 92512];
}
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 4];
int index37 = (int)threadIdx.x * 4 + 16;
T_dense[4] = T_dense[4] + placeholder_d_shared[0] * placeholder_shared[index37];
int index38 = (int)threadIdx.x * 4 + 32;
T_dense[8] = T_dense[8] + placeholder_d_shared[0] * placeholder_shared[index38];
int index39 = (int)threadIdx.x * 4 + 48;
T_dense[12] = T_dense[12] + placeholder_d_shared[0] * placeholder_shared[index39];
int index40 = (int)threadIdx.x * 4 + 64;
T_dense[16] = T_dense[16] + placeholder_d_shared[0] * placeholder_shared[index40];
int index41 = (int)threadIdx.x * 4 + 80;
T_dense[20] = T_dense[20] + placeholder_d_shared[0] * placeholder_shared[index41];
int index42 = (int)threadIdx.x * 4 + 96;
T_dense[24] = T_dense[24] + placeholder_d_shared[0] * placeholder_shared[index42];
int index43 = (int)threadIdx.x * 4 + 112;
T_dense[28] = T_dense[28] + placeholder_d_shared[0] * placeholder_shared[index43];
int index44 = (int)threadIdx.x * 4 + 1;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index44];
int index45 = (int)threadIdx.x * 4 + 17;
T_dense[5] = T_dense[5] + placeholder_d_shared[0] * placeholder_shared[index45];
int index46 = (int)threadIdx.x * 4 + 33;
T_dense[9] = T_dense[9] + placeholder_d_shared[0] * placeholder_shared[index46];
int index47 = (int)threadIdx.x * 4 + 49;
T_dense[13] = T_dense[13] + placeholder_d_shared[0] * placeholder_shared[index47];
int index48 = (int)threadIdx.x * 4 + 65;
T_dense[17] = T_dense[17] + placeholder_d_shared[0] * placeholder_shared[index48];
int index49 = (int)threadIdx.x * 4 + 81;
T_dense[21] = T_dense[21] + placeholder_d_shared[0] * placeholder_shared[index49];
int index50 = (int)threadIdx.x * 4 + 97;
T_dense[25] = T_dense[25] + placeholder_d_shared[0] * placeholder_shared[index50];
int index51 = (int)threadIdx.x * 4 + 113;
T_dense[29] = T_dense[29] + placeholder_d_shared[0] * placeholder_shared[index51];
int index52 = (int)threadIdx.x * 4 + 2;
T_dense[2] = T_dense[2] + placeholder_d_shared[0] * placeholder_shared[index52];
int index53 = (int)threadIdx.x * 4 + 18;
T_dense[6] = T_dense[6] + placeholder_d_shared[0] * placeholder_shared[index53];
int index54 = (int)threadIdx.x * 4 + 34;
T_dense[10] = T_dense[10] + placeholder_d_shared[0] * placeholder_shared[index54];
int index55 = (int)threadIdx.x * 4 + 50;
T_dense[14] = T_dense[14] + placeholder_d_shared[0] * placeholder_shared[index55];
int index56 = (int)threadIdx.x * 4 + 66;
T_dense[18] = T_dense[18] + placeholder_d_shared[0] * placeholder_shared[index56];
int index57 = (int)threadIdx.x * 4 + 82;
T_dense[22] = T_dense[22] + placeholder_d_shared[0] * placeholder_shared[index57];
int index58 = (int)threadIdx.x * 4 + 98;
T_dense[26] = T_dense[26] + placeholder_d_shared[0] * placeholder_shared[index58];
int index59 = (int)threadIdx.x * 4 + 114;
T_dense[30] = T_dense[30] + placeholder_d_shared[0] * placeholder_shared[index59];
int index60 = (int)threadIdx.x * 4 + 3;
T_dense[3] = T_dense[3] + placeholder_d_shared[0] * placeholder_shared[index60];
int index61 = (int)threadIdx.x * 4 + 19;
T_dense[7] = T_dense[7] + placeholder_d_shared[0] * placeholder_shared[index61];
int index62 = (int)threadIdx.x * 4 + 35;
T_dense[11] = T_dense[11] + placeholder_d_shared[0] * placeholder_shared[index62];
int index63 = (int)threadIdx.x * 4 + 51;
T_dense[15] = T_dense[15] + placeholder_d_shared[0] * placeholder_shared[index63];
int index64 = (int)threadIdx.x * 4 + 67;
T_dense[19] = T_dense[19] + placeholder_d_shared[0] * placeholder_shared[index64];
int index65 = (int)threadIdx.x * 4 + 83;
T_dense[23] = T_dense[23] + placeholder_d_shared[0] * placeholder_shared[index65];
int index66 = (int)threadIdx.x * 4 + 99;
T_dense[27] = T_dense[27] + placeholder_d_shared[0] * placeholder_shared[index66];
int index67 = (int)threadIdx.x * 4 + 115;
T_dense[31] = T_dense[31] + placeholder_d_shared[0] * placeholder_shared[index67];
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index68 = (int)threadIdx.x * 4 + ax1_inner;
int index69 = ax1_inner;
T_relu[index68] = max(T_dense[index69] + placeholder2[index68], 0.000000e+00f);
T_relu[index68 + 16] = max(T_dense[index69 + 4] + placeholder2[index68 + 16], 0.000000e+00f);
T_relu[index68 + 32] = max(T_dense[index69 + 8] + placeholder2[index68 + 32], 0.000000e+00f);
T_relu[index68 + 48] = max(T_dense[index69 + 12] + placeholder2[index68 + 48], 0.000000e+00f);
T_relu[index68 + 64] = max(T_dense[index69 + 16] + placeholder2[index68 + 64], 0.000000e+00f);
T_relu[index68 + 80] = max(T_dense[index69 + 20] + placeholder2[index68 + 80], 0.000000e+00f);
T_relu[index68 + 96] = max(T_dense[index69 + 24] + placeholder2[index68 + 96], 0.000000e+00f);
T_relu[index68 + 112] = max(T_dense[index69 + 28] + placeholder2[index68 + 112], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmppogih2g9 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<84>;
.reg .b32 %r<48>;
.reg .b64 %rd<22>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[8];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[1024];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd1, %rd11;
mov.u32 %r1, %tid.x;
shl.b32 %r7, %r1, 3;
mov.u32 %r8, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r8, %r7;
mov.u32 %r9, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r9, %r7;
mul.lo.s32 %r10, %r1, 784;
cvta.to.global.u64 %rd13, %rd9;
mul.wide.s32 %rd14, %r10, 4;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(21+i*-6), %rd(13+i*-1), %rd14;
add.s64 %rd20, %rd15, 200704;
mov.f32 %f76, 0f00000000;
mov.u32 %r47, 0;
for.size.1 i in range(0, 7, 1):
mov.f32 %f(77+i*1), %f76;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.v2.u32 {%r11, %r12}, [%rd21];
st.shared.v2.u32 [%r2], {%r11, %r12};
$L__BB0_3:
for.size.2 i in range(0, 8, 1):
ld.global.nc.v2.u32 {%r(15+i*4), %r(16+i*4)}, [%rd20+(-200704+i*50176)];
st.shared.v2.u32 [%r3+(0+i*128)], {%r(15+i*4), %r(16+i*4)};
bar.sync 0;
for.size.17 i in range(0, 2, 1):
ld.shared.f32 %f(25+i*17), [%r3+(0+i*4)];
ld.shared.f32 %f(26+i*17), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(27+i*49), %f(26+i*17), %f(25+i*17), %f(76+i*-49);
ld.shared.f32 %f(28+i*16), [%r3+(128+i*4)];
fma.rn.f32 %f(29+i*48), %f(26+i*17), %f(28+i*16), %f(77+i*-48);
ld.shared.f32 %f(30+i*15), [%r3+(256+i*4)];
fma.rn.f32 %f(31+i*47), %f(26+i*17), %f(30+i*15), %f(78+i*-47);
ld.shared.f32 %f(32+i*14), [%r3+(384+i*4)];
fma.rn.f32 %f(33+i*46), %f(26+i*17), %f(32+i*14), %f(79+i*-46);
ld.shared.f32 %f(34+i*13), [%r3+(512+i*4)];
fma.rn.f32 %f(35+i*45), %f(26+i*17), %f(34+i*13), %f(80+i*-45);
ld.shared.f32 %f(36+i*12), [%r3+(640+i*4)];
fma.rn.f32 %f(37+i*44), %f(26+i*17), %f(36+i*12), %f(81+i*-44);
ld.shared.f32 %f(38+i*11), [%r3+(768+i*4)];
fma.rn.f32 %f(39+i*43), %f(26+i*17), %f(38+i*11), %f(82+i*-43);
ld.shared.f32 %f(40+i*10), [%r3+(896+i*4)];
fma.rn.f32 %f(41+i*42), %f(26+i*17), %f(40+i*10), %f(83+i*-42);
for.size.1 i in range(0, 2, 1):
add.s64 %rd(21+i*-1), %rd(21+i*-1), 8;
add.s32 %r47, %r47, 1;
setp.ne.s32 %p2, %r47, 392;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd16, %rd8;
mul.wide.s32 %rd17, %r1, 4;
add.s64 %rd18, %rd16, %rd17;
ld.global.nc.f32 %f51, [%rd18];
add.f32 %f52, %f76, %f51;
mov.f32 %f53, 0f00000000;
max.f32 %f54, %f52, %f53;
add.s64 %rd19, %rd1, %rd17;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd19+(0+i*64)], %f(54+i*3);
ld.global.nc.f32 %f(55+i*3), [%rd18+(64+i*64)];
add.f32 %f(56+i*3), %f(77+i*1), %f(55+i*3);
max.f32 %f(57+i*3), %f(56+i*3), %f53;
st.global.f32 [%rd19+448], %f75;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[8];
__shared__ float placeholder_d_shared[2];
__shared__ float placeholder_shared[256];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 392; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
((float2*)(placeholder_d_shared + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder + (int)threadIdx.x * 784 + k_outer_outer * 2))[0];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 8; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder1 + ax0_ax1_fused_outer_outer * 12544 + (int)threadIdx.x * 784 + k_outer_outer * 2))[0];
}
__syncthreads();
for (int k_inner = 0; k_inner < 2; ++k_inner) {
int index0 = k_inner;
int index1 = (int)threadIdx.x * 2 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index0] * placeholder_shared[index1];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index0] * placeholder_shared[index1 + 32]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index0] * placeholder_shared[index1 + 64]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index0] * placeholder_shared[index1 + 96]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index0] * placeholder_shared[index1 + 128]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index0] * placeholder_shared[index1 + 160]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index0] * placeholder_shared[index1 + 192]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index0] * placeholder_shared[index1 + 224]);
}
}
int index2 = (int)threadIdx.x;
T_relu[index2] = max(T_dense[0] + placeholder2[index2], 0.000000e+00f);
T_relu[index2 + 16] = max(T_dense[1] + placeholder2[index2 + 16], 0.000000e+00f);
T_relu[index2 + 32] = max(T_dense[2] + placeholder2[index2 + 32], 0.000000e+00f);
T_relu[index2 + 48] = max(T_dense[3] + placeholder2[index2 + 48], 0.000000e+00f);
T_relu[index2 + 64] = max(T_dense[4] + placeholder2[index2 + 64], 0.000000e+00f);
T_relu[index2 + 80] = max(T_dense[5] + placeholder2[index2 + 80], 0.000000e+00f);
T_relu[index2 + 96] = max(T_dense[6] + placeholder2[index2 + 96], 0.000000e+00f);
T_relu[index2 + 112] = max(T_dense[7] + placeholder2[index2 + 112], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpllbntl1u | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<102>;
.reg .b32 %r<21>;
.reg .b64 %rd<26>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[4];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[64];
ld.param.u64 %rd11, [default_function_kernel0_param_0];
ld.param.u64 %rd12, [default_function_kernel0_param_1];
ld.param.u64 %rd13, [default_function_kernel0_param_2];
ld.param.u64 %rd10, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd14, %rd12;
cvta.to.global.u64 %rd1, %rd13;
mov.u32 %r1, %tid.x;
shl.b32 %r11, %r1, 2;
mov.u32 %r12, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r12, %r11;
shl.b32 %r4, %r1, 3;
mov.u32 %r13, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r13, %r4;
shl.b32 %r14, %r1, 5;
add.s32 %r5, %r13, %r14;
mul.lo.s32 %r15, %r1, 784;
cvta.to.global.u64 %rd15, %rd11;
mul.wide.s32 %rd16, %r15, 4;
add.s64 %rd17, %rd15, %rd16;
add.s64 %rd25, %rd17, 4;
mov.u32 %r6, %ctaid.x;
mul.lo.s32 %r16, %r1, 1568;
mad.lo.s32 %r17, %r6, 12544, %r16;
mul.wide.s32 %rd18, %r17, 4;
add.s64 %rd19, %rd14, %rd18;
add.s64 %rd24, %rd19, 25088;
mov.u32 %r20, 0;
mov.f32 %f94, 0f00000000;
for.size.1 i in range(0, 7, 1):
mov.f32 %f(95+i*1), %f94;
$L__BB0_1:
bar.sync 0;
add.s64 %rd6, %rd25, -4;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f33, [%rd6];
st.shared.f32 [%r2], %f33;
$L__BB0_3:
add.s64 %rd7, %rd24, -25088;
for.size.1 i in range(0, 2, 1):
ld.global.nc.f32 %f(34+i*1), [%rd24+(-21952+i*-3136)];
for.size.3 i in range(0, 3, 1):
st.shared.v2.f32 [%r3+(0+i*16)], {%f(35+i*2), %f(34+i*2)};
ld.global.nc.f32 %f(36+i*2), [%rd24+(-9408+i*12544)];
ld.global.nc.f32 %f(37+i*2), [%rd24+(-12544+i*12544)];
st.shared.v2.f32 [%r3+48], {%f41, %f40};
bar.sync 0;
ld.shared.f32 %f42, [%r5];
ld.shared.f32 %f43, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(9+i*1), %f43, %f(42+i*2), %f(101+i*-1);
ld.shared.f32 %f(44+i*1), [%r5+(4+i*4)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(11+i*1), %f43, %f(45+i*1), %f(99+i*-1);
ld.shared.f32 %f(46+i*1), [%r5+(12+i*4)];
fma.rn.f32 %f16, %f43, %f50, %f94;
bar.sync 0;
@%p1 bra $L__BB0_5;
ld.global.nc.f32 %f51, [%rd6+4];
st.shared.f32 [%r2], %f51;
$L__BB0_5:
for.size.1 i in range(0, 2, 1):
ld.global.nc.f32 %f(52+i*1), [%rd7+(3140+i*-3136)];
for.size.3 i in range(0, 3, 1):
st.shared.v2.f32 [%r3+(0+i*16)], {%f(53+i*2), %f(52+i*2)};
ld.global.nc.f32 %f(54+i*2), [%rd7+(15684+i*12544)];
ld.global.nc.f32 %f(55+i*2), [%rd7+(12548+i*12544)];
st.shared.v2.f32 [%r3+48], {%f59, %f58};
bar.sync 0;
ld.shared.f32 %f60, [%r5];
ld.shared.f32 %f61, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(101+i*-1), %f61, %f(60+i*2), %f(9+i*1);
ld.shared.f32 %f(62+i*1), [%r5+(4+i*4)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(99+i*-1), %f61, %f(63+i*1), %f(11+i*1);
ld.shared.f32 %f(64+i*1), [%r5+(12+i*4)];
fma.rn.f32 %f94, %f61, %f68, %f16;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(25+i*-1), %rd(25+i*-1), 8;
add.s32 %r20, %r20, 2;
setp.ne.s32 %p3, %r20, 784;
@%p3 bra $L__BB0_1;
cvta.to.global.u64 %rd20, %rd10;
shl.b32 %r18, %r6, 4;
add.s32 %r19, %r18, %r4;
mul.wide.s32 %rd21, %r19, 4;
add.s64 %rd22, %rd20, %rd21;
ld.global.nc.f32 %f69, [%rd22];
add.f32 %f70, %f101, %f69;
mov.f32 %f71, 0f00000000;
max.f32 %f72, %f70, %f71;
add.s64 %rd23, %rd1, %rd21;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd23+(0+i*4)], %f(72+i*3);
ld.global.nc.f32 %f(73+i*3), [%rd22+(4+i*4)];
add.f32 %f(74+i*3), %f(100+i*-1), %f(73+i*3);
max.f32 %f(75+i*3), %f(74+i*3), %f71;
st.global.f32 [%rd23+28], %f93;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[8];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[16];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_d_shared[index0] = placeholder[index1];
}
int2 _1 = make_int2((int)blockIdx.x * 12544 + (int)threadIdx.x * 1568 + k_outer_outer+784*0, (int)blockIdx.x * 12544 + (int)threadIdx.x * 1568 + k_outer_outer+784*1);
int index2 = _1.x;
int index3 = _1.y;
((float2*)(placeholder_shared + (int)threadIdx.x * 2))[0] = make_float2(placeholder1[index2],placeholder1[index3]);
int2 _2 = make_int2((int)blockIdx.x * 12544 + (int)threadIdx.x * 1568 + k_outer_outer + 3136+784*0, (int)blockIdx.x * 12544 + (int)threadIdx.x * 1568 + k_outer_outer + 3136+784*1);
int index4 = _2.x;
int index5 = _2.y;
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 4))[0] = make_float2(placeholder1[index4],placeholder1[index5]);
int2 _3 = make_int2((int)blockIdx.x * 12544 + (int)threadIdx.x * 1568 + k_outer_outer + 6272+784*0, (int)blockIdx.x * 12544 + (int)threadIdx.x * 1568 + k_outer_outer + 6272+784*1);
int index6 = _3.x;
int index7 = _3.y;
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 8))[0] = make_float2(placeholder1[index6],placeholder1[index7]);
int2 _4 = make_int2((int)blockIdx.x * 12544 + (int)threadIdx.x * 1568 + k_outer_outer + 9408+784*0, (int)blockIdx.x * 12544 + (int)threadIdx.x * 1568 + k_outer_outer + 9408+784*1);
int index8 = _4.x;
int index9 = _4.y;
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 12))[0] = make_float2(placeholder1[index8],placeholder1[index9]);
__syncthreads();
int index0 = (int)threadIdx.x;
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 8];
int index10 = (int)threadIdx.x * 8 + 1;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index10];
int index11 = (int)threadIdx.x * 8 + 2;
T_dense[2] = T_dense[2] + placeholder_d_shared[0] * placeholder_shared[index11];
int index12 = (int)threadIdx.x * 8 + 3;
T_dense[3] = T_dense[3] + placeholder_d_shared[0] * placeholder_shared[index12];
int index13 = (int)threadIdx.x * 8 + 4;
T_dense[4] = T_dense[4] + placeholder_d_shared[0] * placeholder_shared[index13];
int index14 = (int)threadIdx.x * 8 + 5;
T_dense[5] = T_dense[5] + placeholder_d_shared[0] * placeholder_shared[index14];
int index15 = (int)threadIdx.x * 8 + 6;
T_dense[6] = T_dense[6] + placeholder_d_shared[0] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 8 + 7;
T_dense[7] = T_dense[7] + placeholder_d_shared[0] * placeholder_shared[index16];
}
for (int ax1_inner = 0; ax1_inner < 8; ++ax1_inner) {
int index17 = (int)blockIdx.x * 16 + (int)threadIdx.x * 8 + ax1_inner;
int index18 = ax1_inner;
T_relu[index17] = max(T_dense[index18] + placeholder2[index17], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpn1e9vict | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<775>;
.reg .b32 %r<184>;
.reg .b64 %rd<18>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[4096];
ld.param.u64 %rd1, [default_function_kernel0_param_0];
ld.param.u64 %rd2, [default_function_kernel0_param_1];
mov.u32 %r12, %tid.x;
mov.u32 %r180, 0;
mov.f32 %f711, 0f00000000;
cvta.to.global.u64 %rd5, %rd1;
shl.b32 %r17, %r12, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r19, %r18, %r17;
cvta.to.global.u64 %rd8, %rd2;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(712+i*1), %f711;
$L__BB0_1:
mov.u32 %r181, 0;
bar.sync 0;
shl.b32 %r15, %r180, 3;
add.s32 %r16, %r15, %r12;
mul.wide.s32 %rd6, %r16, 4;
add.s64 %rd7, %rd5, %rd6;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(321+i*1), [%rd7+(0+i*8)];
st.shared.f32 [%r19+(0+i*8)], %f(321+i*1);
$L__BB0_2:
shl.b32 %r173, %r180, 3;
add.s32 %r172, %r173, %r12;
mad.lo.s32 %r171, %r12, 3, %r172;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(170+i*-148), %r(12+i*169), (2+i*1);
add.s32 %r23, %r22, %r170;
shl.b32 %r24, %r23, 2;
mov.u32 %r25, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r26, %r25, %r24;
mad.lo.s32 %r27, %r181, 784, %r171;
mul.wide.s32 %rd9, %r27, 4;
add.s64 %rd10, %rd8, %rd9;
for.size.2 i in range(0, 16, 1):
ld.global.nc.v4.u32 {%r(28+i*8), %r(29+i*8), %r(30+i*8), %r(31+i*8)}, [%rd10+(0+i*3136)];
st.shared.v4.u32 [%r26+(0+i*32)], {%r(28+i*8), %r(29+i*8), %r(30+i*8), %r(31+i*8)};
add.s32 %r181, %r181, 16;
setp.ne.s32 %p1, %r181, 128;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r182, 0;
$L__BB0_4:
mov.u32 %r183, 0;
$L__BB0_5:
shl.b32 %r179, %r182, 1;
mov.u32 %r176, _ZZ24default_function_kernel0E18placeholder_shared;
shl.b32 %r175, %r12, 9;
mov.u32 %r174, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.3 i in range(0, 2, 1):
add.s32 %r(159+i*4), %r(183+i*-24), %r(179+i*-4);
shl.b32 %r(160+i*4), %r(159+i*4), 2;
add.s32 %r(162+i*4), %r(174+i*2), %r(160+i*4);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(325+i*1), [%r(166+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(774+i*-1), %f326, %f(325+i*2), %f(774+i*-1);
ld.shared.f32 %f(327+i*1), [%r166+(32+i*32)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(772+i*-1), %f326, %f(328+i*1), %f(772+i*-1);
ld.shared.f32 %f(329+i*1), [%r166+(96+i*32)];
fma.rn.f32 %f711, %f326, %f389, %f711;
add.s32 %r167, %r183, 1;
mov.u32 %r183, 1;
setp.lt.u32 %p2, %r167, 2;
@%p2 bra $L__BB0_5;
add.s32 %r182, %r182, 1;
setp.lt.u32 %p3, %r182, 4;
@%p3 bra $L__BB0_4;
add.s32 %r180, %r180, 1;
setp.lt.u32 %p4, %r180, 98;
@%p4 bra $L__BB0_1;
ld.param.u64 %rd17, [default_function_kernel0_param_2];
ld.param.u64 %rd16, [default_function_kernel0_param_3];
shl.b32 %r169, %r12, 6;
cvta.to.global.u64 %rd11, %rd16;
mul.wide.s32 %rd12, %r169, 4;
add.s64 %rd13, %rd11, %rd12;
ld.global.nc.f32 %f390, [%rd13];
add.f32 %f391, %f774, %f390;
mov.f32 %f392, 0f00000000;
max.f32 %f393, %f391, %f392;
cvta.to.global.u64 %rd14, %rd17;
add.s64 %rd15, %rd14, %rd12;
for.size.4 i in range(0, 63, 1):
st.global.f32 [%rd15+(0+i*4)], %f(393+i*3);
ld.global.nc.f32 %f(394+i*3), [%rd13+(4+i*4)];
add.f32 %f(395+i*3), %f(773+i*-1), %f(394+i*3);
max.f32 %f(396+i*3), %f(395+i*3), %f392;
st.global.f32 [%rd15+252], %f582;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[1024];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
T_dense[32] = 0.000000e+00f;
T_dense[33] = 0.000000e+00f;
T_dense[34] = 0.000000e+00f;
T_dense[35] = 0.000000e+00f;
T_dense[36] = 0.000000e+00f;
T_dense[37] = 0.000000e+00f;
T_dense[38] = 0.000000e+00f;
T_dense[39] = 0.000000e+00f;
T_dense[40] = 0.000000e+00f;
T_dense[41] = 0.000000e+00f;
T_dense[42] = 0.000000e+00f;
T_dense[43] = 0.000000e+00f;
T_dense[44] = 0.000000e+00f;
T_dense[45] = 0.000000e+00f;
T_dense[46] = 0.000000e+00f;
T_dense[47] = 0.000000e+00f;
T_dense[48] = 0.000000e+00f;
T_dense[49] = 0.000000e+00f;
T_dense[50] = 0.000000e+00f;
T_dense[51] = 0.000000e+00f;
T_dense[52] = 0.000000e+00f;
T_dense[53] = 0.000000e+00f;
T_dense[54] = 0.000000e+00f;
T_dense[55] = 0.000000e+00f;
T_dense[56] = 0.000000e+00f;
T_dense[57] = 0.000000e+00f;
T_dense[58] = 0.000000e+00f;
T_dense[59] = 0.000000e+00f;
T_dense[60] = 0.000000e+00f;
T_dense[61] = 0.000000e+00f;
T_dense[62] = 0.000000e+00f;
T_dense[63] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 8 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
placeholder_d_shared[index0 + 6] = placeholder[index1 + 6];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 128; ++ax0_ax1_fused_outer_outer) {
((float4*)(placeholder_shared + ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x * 4))[0] = ((float4*)(placeholder1 + ax0_ax1_fused_outer_outer * 784 + k_outer_outer * 8 + (int)threadIdx.x * 4))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 4; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 2; ++k_inner) {
int index2 = k_outer_inner * 2 + k_inner;
int index3 = (int)threadIdx.x * 512 + k_outer_inner * 2 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 8]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 16]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 24]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 32]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 40]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 48]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 56]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 64]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 72]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 80]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 88]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 96]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 104]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 112]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 120]);
T_dense[16] = T_dense[16] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 128]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 136]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 144]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 152]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 160]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 168]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 176]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 184]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 192]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 200]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 208]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 216]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 224]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 232]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 240]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 248]);
T_dense[32] = T_dense[32] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 256]);
T_dense[33] = T_dense[33] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 264]);
T_dense[34] = T_dense[34] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 272]);
T_dense[35] = T_dense[35] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 280]);
T_dense[36] = T_dense[36] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 288]);
T_dense[37] = T_dense[37] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 296]);
T_dense[38] = T_dense[38] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 304]);
T_dense[39] = T_dense[39] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 312]);
T_dense[40] = T_dense[40] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 320]);
T_dense[41] = T_dense[41] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 328]);
T_dense[42] = T_dense[42] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 336]);
T_dense[43] = T_dense[43] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 344]);
T_dense[44] = T_dense[44] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 352]);
T_dense[45] = T_dense[45] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 360]);
T_dense[46] = T_dense[46] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 368]);
T_dense[47] = T_dense[47] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 376]);
T_dense[48] = T_dense[48] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 384]);
T_dense[49] = T_dense[49] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 392]);
T_dense[50] = T_dense[50] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 400]);
T_dense[51] = T_dense[51] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 408]);
T_dense[52] = T_dense[52] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 416]);
T_dense[53] = T_dense[53] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 424]);
T_dense[54] = T_dense[54] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 432]);
T_dense[55] = T_dense[55] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 440]);
T_dense[56] = T_dense[56] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 448]);
T_dense[57] = T_dense[57] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 456]);
T_dense[58] = T_dense[58] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 464]);
T_dense[59] = T_dense[59] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 472]);
T_dense[60] = T_dense[60] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 480]);
T_dense[61] = T_dense[61] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 488]);
T_dense[62] = T_dense[62] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 496]);
T_dense[63] = T_dense[63] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 504]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index4 = (int)threadIdx.x * 64 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp6_8k0idv | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<67>;
.reg .b32 %r<116>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[224];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[28672];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd7, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd5;
cvta.to.global.u64 %rd3, %rd7;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r1, %tid.x;
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r19, %r18;
mul.lo.s32 %r3, %r1, 112;
add.s32 %r4, %r1, 448;
mov.u32 %r20, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r20, %r18;
mov.u32 %r111, 0;
mov.f32 %f63, 0f00000000;
mov.f32 %f64, %f63;
$L__BB0_1:
bar.sync 0;
setp.lt.s32 %p1, %r1, 56;
@%p1 bra $L__BB0_2;
bra.uni $L__BB0_3;
$L__BB0_2:
mad.lo.s32 %r21, %r111, 56, %r1;
mul.wide.s32 %rd9, %r21, 4;
add.s64 %rd10, %rd2, %rd9;
ld.global.nc.f32 %f9, [%rd10];
st.shared.f32 [%r2], %f9;
$L__BB0_3:
mul.lo.s32 %r7, %r111, 56;
mov.u32 %r114, 1024;
mov.u32 %r112, %r1;
mov.u32 %r113, %r4;
$L__BB0_4:
mul.hi.s32 %r23, %r112, -1840700269;
add.s32 %r24, %r23, %r112;
shr.u32 %r25, %r24, 31;
shr.s32 %r26, %r24, 5;
add.s32 %r27, %r26, %r25;
mul.lo.s32 %r28, %r27, 56;
sub.s32 %r29, %r112, %r28;
add.s32 %r30, %r29, %r7;
mad.lo.s32 %r31, %r27, 784, %r30;
mul.wide.s32 %rd11, %r31, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f10, [%rd12];
add.s32 %r32, %r5, %r114;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r32+(-1024+i*256)], %f(10+i*1);
add.s32 %r(33+i*10), %r112, (64+i*64);
mul.hi.s32 %r(34+i*10), %r(33+i*10), -1840700269;
add.s32 %r(35+i*10), %r(34+i*10), %r(33+i*10);
shr.u32 %r(36+i*10), %r(35+i*10), 31;
shr.s32 %r(37+i*10), %r(35+i*10), 5;
add.s32 %r(38+i*10), %r(37+i*10), %r(36+i*10);
mul.lo.s32 %r(39+i*10), %r(38+i*10), 56;
sub.s32 %r(40+i*10), %r(33+i*10), %r(39+i*10);
add.s32 %r(41+i*10), %r(40+i*10), %r7;
mad.lo.s32 %r(42+i*10), %r(38+i*10), 784, %r(41+i*10);
mul.wide.s32 %rd(13+i*2), %r(42+i*10), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(11+i*1), [%rd(14+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r32+(0+i*256)], %f(14+i*1);
add.s32 %r(73+i*10), %r113, (-128+i*64);
mul.hi.s32 %r(74+i*10), %r(73+i*10), -1840700269;
add.s32 %r(75+i*10), %r(74+i*10), %r(73+i*10);
shr.u32 %r(76+i*10), %r(75+i*10), 31;
shr.s32 %r(77+i*10), %r(75+i*10), 5;
add.s32 %r(78+i*10), %r(77+i*10), %r(76+i*10);
mul.lo.s32 %r(79+i*10), %r(78+i*10), 56;
sub.s32 %r(80+i*10), %r(73+i*10), %r(79+i*10);
add.s32 %r(81+i*10), %r(80+i*10), %r7;
mad.lo.s32 %r(82+i*10), %r(78+i*10), 784, %r(81+i*10);
mul.wide.s32 %rd(21+i*2), %r(82+i*10), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(15+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r32+512], %f16;
mul.hi.s32 %r93, %r113, -1840700269;
add.s32 %r94, %r93, %r113;
shr.u32 %r95, %r94, 31;
shr.s32 %r96, %r94, 5;
add.s32 %r97, %r96, %r95;
mul.lo.s32 %r98, %r97, 56;
sub.s32 %r99, %r113, %r98;
add.s32 %r100, %r99, %r7;
mad.lo.s32 %r101, %r97, 784, %r100;
mul.wide.s32 %rd25, %r101, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f17, [%rd26];
st.shared.f32 [%r32+768], %f17;
for.size.1 i in range(0, 2, 1):
add.s32 %r(113+i*-1), %r(113+i*-1), 512;
add.s32 %r114, %r114, 2048;
setp.ne.s32 %p2, %r114, 29696;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r115, 0;
$L__BB0_6:
add.s32 %r103, %r115, %r3;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(104+i*3), %r(103+i*12), 2;
add.s32 %r(106+i*3), %r(20+i*-1), %r(104+i*3);
ld.shared.f32 %f(18+i*1), [%r(106+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(20+i*2), %f19, %f(18+i*3), %f(64+i*-1);
ld.shared.f32 %f(21+i*2), [%r106+(224+i*-220)];
for.size.5 i in range(0, 6, 1):
ld.shared.f32 %f(24+i*5), [%r109+(4+i*4)];
fma.rn.f32 %f(25+i*5), %f(24+i*5), %f(23+i*5), %f(20+i*5);
ld.shared.f32 %f(26+i*5), [%r106+(228+i*4)];
fma.rn.f32 %f(27+i*5), %f(24+i*5), %f(26+i*5), %f(22+i*5);
ld.shared.f32 %f(28+i*5), [%r106+(8+i*4)];
ld.shared.f32 %f54, [%r109+28];
fma.rn.f32 %f64, %f54, %f53, %f50;
ld.shared.f32 %f55, [%r106+252];
fma.rn.f32 %f63, %f54, %f55, %f52;
add.s32 %r115, %r115, 8;
setp.ne.s32 %p3, %r115, 56;
@%p3 bra $L__BB0_6;
add.s32 %r111, %r111, 1;
setp.lt.u32 %p4, %r111, 14;
@%p4 bra $L__BB0_1;
shl.b32 %r110, %r1, 1;
mul.wide.s32 %rd27, %r110, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f56, [%rd28];
add.f32 %f57, %f64, %f56;
mov.f32 %f58, 0f00000000;
max.f32 %f59, %f57, %f58;
add.s64 %rd29, %rd3, %rd27;
st.global.f32 [%rd29], %f59;
ld.global.nc.f32 %f60, [%rd28+4];
add.f32 %f61, %f63, %f60;
max.f32 %f62, %f61, %f58;
st.global.f32 [%rd29+4], %f62;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[56];
__shared__ float placeholder_shared[7168];
for (int j_inner_init = 0; j_inner_init < 2; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 14; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 56) {
int index1 = (int)threadIdx.x;
int index2 = k_outer_outer * 56 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 112; ++ax0_ax1_fused_outer_outer) {
int index3 = ax0_ax1_fused_outer_outer * 64 + (int)threadIdx.x;
int index4 = (ax0_ax1_fused_outer_outer * 64 + (int)threadIdx.x) / 56 * 784 + k_outer_outer * 56 + (ax0_ax1_fused_outer_outer * 64 + (int)threadIdx.x) % 56;
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 56; ++k_outer_inner) {
for (int j_inner = 0; j_inner < 2; ++j_inner) {
int index5 = j_inner;
int index6 = k_outer_inner;
int index7 = (int)threadIdx.x * 112 + j_inner * 56 + k_outer_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index8 = (int)threadIdx.x * 2 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp57j9aux5 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<48>;
.reg .b32 %r<118>;
.reg .b64 %rd<34>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[3136];
ld.param.u64 %rd8, [default_function_kernel0_param_0];
ld.param.u64 %rd9, [default_function_kernel0_param_1];
ld.param.u64 %rd10, [default_function_kernel0_param_2];
ld.param.u64 %rd11, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd8;
cvta.to.global.u64 %rd1, %rd9;
cvta.to.global.u64 %rd2, %rd10;
cvta.to.global.u64 %rd3, %rd11;
mov.u32 %r1, %ctaid.x;
mul.lo.s32 %r2, %r1, 3136;
mov.u32 %r3, %tid.x;
mul.lo.s32 %r4, %r3, 196;
add.s64 %rd4, %rd12, 8;
shl.b32 %r20, %r3, 2;
mov.u32 %r21, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r22, %r21, %r20;
add.s32 %r5, %r22, 8;
mov.f32 %f44, 0f00000000;
mov.u32 %r19, 0;
mov.u32 %r113, %r19;
mov.f32 %f45, %f44;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r113, 196;
add.s32 %r24, %r3, %r8;
mul.wide.s32 %rd13, %r24, 4;
add.s64 %rd33, %rd4, %rd13;
mov.u32 %r114, %r5;
mov.u32 %r115, %r19;
$L__BB0_2:
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(9+i*1), [%rd33+(-8+i*8)];
st.shared.f32 [%r114+(-8+i*8)], %f(9+i*1);
add.s64 %rd33, %rd33, 16;
for.size.1 i in range(0, 2, 1):
add.s32 %r(114+i*1), %r(114+i*1), (16+i*-14);
setp.ne.s32 %p1, %r115, 98;
@%p1 bra $L__BB0_2;
add.s32 %r13, %r8, %r2;
mov.u32 %r116, 0;
$L__BB0_4:
shl.b32 %r26, %r116, 1;
add.s32 %r27, %r26, %r3;
mul.hi.s32 %r28, %r27, 1402438301;
shr.u32 %r29, %r28, 31;
shr.s32 %r30, %r28, 6;
add.s32 %r31, %r30, %r29;
mul.lo.s32 %r32, %r31, 196;
sub.s32 %r33, %r27, %r32;
add.s32 %r34, %r13, %r33;
mad.lo.s32 %r35, %r31, 784, %r34;
mul.wide.s32 %rd14, %r35, 4;
add.s64 %rd15, %rd1, %rd14;
ld.global.nc.f32 %f11, [%rd15];
shl.b32 %r36, %r27, 2;
mov.u32 %r37, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r38, %r37, %r36;
for.size.13 i in range(0, 7, 1):
st.shared.f32 [%r38+(0+i*8)], %f(11+i*1);
add.s32 %r(39+i*9), %r27, (2+i*2);
mul.hi.s32 %r(40+i*9), %r(39+i*9), 1402438301;
shr.u32 %r(41+i*9), %r(40+i*9), 31;
shr.s32 %r(42+i*9), %r(40+i*9), 6;
add.s32 %r(43+i*9), %r(42+i*9), %r(41+i*9);
mul.lo.s32 %r(44+i*9), %r(43+i*9), 196;
sub.s32 %r(45+i*9), %r(39+i*9), %r(44+i*9);
add.s32 %r(46+i*9), %r13, %r(45+i*9);
mad.lo.s32 %r(47+i*9), %r(43+i*9), 784, %r(46+i*9);
mul.wide.s32 %rd(16+i*2), %r(47+i*9), 4;
add.s64 %rd(17+i*2), %rd1, %rd(16+i*2);
ld.global.nc.f32 %f(12+i*1), [%rd(17+i*2)+0];
st.shared.f32 [%r38+56], %f18;
add.s32 %r116, %r116, 8;
setp.ne.s32 %p2, %r116, 392;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r117, 0;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(103+i*1), %r117, (1+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(106+i*1), %r(21+i*82), %r(104+i*-100);
shl.b32 %r108, %r107, 2;
add.s32 %r110, %r37, %r108;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(19+i*1), [%r(110+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(21+i*2), %f20, %f(19+i*3), %f(44+i*1);
ld.shared.f32 %f(22+i*2), [%r110+(1568+i*-1564)];
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(25+i*5), [%r106+(4+i*4)];
fma.rn.f32 %f(26+i*5), %f(25+i*5), %f(24+i*5), %f(21+i*5);
ld.shared.f32 %f(27+i*5), [%r110+(1572+i*4)];
fma.rn.f32 %f(28+i*5), %f(25+i*5), %f(27+i*5), %f(23+i*5);
ld.shared.f32 %f(29+i*5), [%r110+(8+i*4)];
ld.shared.f32 %f35, [%r106+12];
fma.rn.f32 %f44, %f35, %f34, %f31;
ld.shared.f32 %f36, [%r110+1580];
fma.rn.f32 %f45, %f35, %f36, %f33;
add.s32 %r117, %r117, 2;
setp.ne.s32 %p3, %r117, 98;
@%p3 bra $L__BB0_6;
add.s32 %r113, %r113, 1;
setp.lt.u32 %p4, %r113, 4;
@%p4 bra $L__BB0_1;
shl.b32 %r111, %r1, 2;
add.s32 %r112, %r111, %r3;
mul.wide.s32 %rd30, %r112, 4;
add.s64 %rd31, %rd3, %rd30;
ld.global.nc.f32 %f37, [%rd31];
add.f32 %f38, %f44, %f37;
mov.f32 %f39, 0f00000000;
max.f32 %f40, %f38, %f39;
add.s64 %rd32, %rd2, %rd30;
st.global.f32 [%rd32], %f40;
ld.global.nc.f32 %f41, [%rd31+8];
add.f32 %f42, %f45, %f41;
max.f32 %f43, %f42, %f39;
st.global.f32 [%rd32+8], %f43;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[784];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
int index0 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index1 = k_outer_outer * 196 + ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 392; ++ax0_ax1_fused_outer_outer1) {
int index2 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 3136 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) % 196;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 98; ++k_outer_inner) {
int index4 = k_outer_inner * 2;
int index5 = (int)threadIdx.x * 196 + k_outer_inner * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 392]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 393]);
}
}
int index6 = (int)blockIdx.x * 4 + (int)threadIdx.x;
T_relu[index6] = max(T_dense[0] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 2] = max(T_dense[1] + placeholder2[index6 + 2], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp_011dv37 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<64>;
.reg .b32 %r<153>;
.reg .b64 %rd<46>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[28];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[3584];
ld.param.u64 %rd22, [default_function_kernel0_param_0];
ld.param.u64 %rd23, [default_function_kernel0_param_1];
ld.param.u64 %rd24, [default_function_kernel0_param_2];
ld.param.u64 %rd21, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd45, %rd23;
cvta.to.global.u64 %rd2, %rd24;
mov.u32 %r1, %tid.x;
shl.b32 %r8, %r1, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r9, %r8;
mul.hi.s32 %r10, %r1, -1840700269;
add.s32 %r11, %r10, %r1;
shr.u32 %r12, %r11, 31;
shr.s32 %r13, %r11, 2;
add.s32 %r14, %r13, %r12;
mul.lo.s32 %r15, %r14, 7;
sub.s32 %r16, %r1, %r15;
mov.u32 %r17, _ZZ24default_function_kernel0E18placeholder_shared;
for.size.6 i in range(0, 2, 1):
add.s32 %r(3+i*20), %r(17+i*5), %r(8+i*13);
add.s32 %r(18+i*6), %r1, (64+i*-63);
mul.hi.s32 %r(19+i*6), %r(18+i*6), -1840700269;
add.s32 %r(20+i*6), %r(19+i*6), %r(18+i*6);
shr.u32 %r(21+i*6), %r(20+i*6), 31;
shr.s32 %r(22+i*6), %r(20+i*6), 2;
for.size.14 i in range(0, 5, 1):
add.s32 %r(29+i*14), %r(28+i*14), %r(27+i*14);
mul.lo.s32 %r(30+i*14), %r(29+i*14), 7;
sub.s32 %r(31+i*14), %r(24+i*14), %r(30+i*14);
add.s32 %r(32+i*14), %r1, (128+i*64);
mul.hi.s32 %r(33+i*14), %r(32+i*14), -1840700269;
add.s32 %r(34+i*14), %r(33+i*14), %r(32+i*14);
shr.u32 %r(35+i*14), %r(34+i*14), 31;
shr.s32 %r(36+i*14), %r(34+i*14), 2;
add.s32 %r(37+i*14), %r(36+i*14), %r(35+i*14);
add.s32 %r(38+i*14), %r1, (2+i*1);
mul.hi.s32 %r(39+i*14), %r(38+i*14), -1840700269;
add.s32 %r(40+i*14), %r(39+i*14), %r(38+i*14);
shr.u32 %r(41+i*14), %r(40+i*14), 31;
shr.s32 %r(42+i*14), %r(40+i*14), 2;
add.s32 %r99, %r98, %r97;
mul.lo.s32 %r100, %r99, 7;
sub.s32 %r101, %r94, %r100;
for.size.6 i in range(0, 6, 1):
add.s32 %r(102+i*6), %r1, (512+i*64);
mul.hi.s32 %r(103+i*6), %r(102+i*6), -1840700269;
add.s32 %r(104+i*6), %r(103+i*6), %r(102+i*6);
shr.u32 %r(105+i*6), %r(104+i*6), 31;
shr.s32 %r(106+i*6), %r(104+i*6), 2;
add.s32 %r(107+i*6), %r(106+i*6), %r(105+i*6);
for.size.1 i in range(0, 2, 1):
mad.lo.s32 %r(4+i*134), %r(1+i*136), (56+i*728), %r(17+i*84);
for.size.2 i in range(0, 5, 1):
mul.wide.s32 %rd(3+i*1), %r(138+i*1), 4;
mad.lo.s32 %r(139+i*1), %r(131+i*-6), 784, %r(87+i*-14);
for.size.2 i in range(0, 6, 1):
mul.wide.s32 %rd(8+i*1), %r(143+i*1), 4;
mad.lo.s32 %r(144+i*1), %r(93+i*-14), 784, %r(101+i*-14);
mul.wide.s32 %rd14, %r149, 4;
mad.lo.s32 %r150, %r14, 784, %r16;
mul.wide.s32 %rd15, %r150, 4;
cvta.to.global.u64 %rd25, %rd22;
mul.wide.s32 %rd26, %r1, 4;
add.s64 %rd44, %rd25, %rd26;
mov.u32 %r152, 0;
mov.f32 %f62, 0f00000000;
mov.f32 %f63, %f62;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 6;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f7, [%rd44];
st.shared.f32 [%r2], %f7;
$L__BB0_3:
for.size.3 i in range(0, 7, 1):
add.s64 %rd(27+i*1), %rd45, %rd(15+i*-1);
ld.global.nc.f32 %f(8+i*1), [%rd(27+i*1)+0];
st.shared.f32 [%r3+(0+i*256)], %f(8+i*1);
ld.global.nc.f32 %f15, [%rd27+200704];
for.size.3 i in range(0, 6, 1):
st.shared.f32 [%r3+(1792+i*256)], %f(15+i*1);
add.s64 %rd(34+i*1), %rd45, %rd(8+i*-1);
ld.global.nc.f32 %f(16+i*1), [%rd(34+i*1)+0];
st.shared.f32 [%r3+3328], %f21;
bar.sync 0;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(22+i*3), [%r4+(0+i*4)];
ld.shared.f32 %f(23+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(24+i*3), %f(23+i*3), %f(22+i*3), %f(63+i*-39);
for.size.3 i in range(0, 4, 1):
ld.shared.f32 %f(28+i*3), [%r4+(8+i*4)];
ld.shared.f32 %f(29+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(30+i*3), %f(29+i*3), %f(28+i*3), %f(27+i*3);
ld.shared.f32 %f40, [%r4+24];
ld.shared.f32 %f41, [_ZZ24default_function_kernel0E20placeholder_d_shared+24];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(63+i*-20), %f(41+i*-18), %f(40+i*2), %f(39+i*23);
ld.shared.f32 %f(42+i*2), [%r4+(28+i*4)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(45+i*2), %f(26+i*3), %f(44+i*2), %f(43+i*2);
ld.shared.f32 %f(46+i*2), [%r4+(36+i*4)];
fma.rn.f32 %f62, %f41, %f54, %f53;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(45+i*-1), %rd(45+i*-1), 28;
add.s32 %r152, %r152, 1;
setp.ne.s32 %p2, %r152, 112;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd40, %rd21;
shl.b32 %r151, %r1, 1;
mul.wide.s32 %rd41, %r151, 4;
add.s64 %rd42, %rd40, %rd41;
ld.global.nc.f32 %f55, [%rd42];
add.f32 %f56, %f63, %f55;
mov.f32 %f57, 0f00000000;
max.f32 %f58, %f56, %f57;
add.s64 %rd43, %rd2, %rd41;
st.global.f32 [%rd43], %f58;
ld.global.nc.f32 %f59, [%rd42+4];
add.f32 %f60, %f62, %f59;
max.f32 %f61, %f60, %f57;
st.global.f32 [%rd43+4], %f61;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[7];
__shared__ float placeholder_shared[896];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 112; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 7) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 7 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index2 = (int)threadIdx.x / 7 * 784 + k_outer_outer * 7 + (int)threadIdx.x % 7;
placeholder_shared[index0] = placeholder1[index2];
int index3 = ((int)threadIdx.x + 64) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 1) % 7;
placeholder_shared[index0 + 64] = placeholder1[index3];
int index4 = ((int)threadIdx.x + 128) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 2) % 7;
placeholder_shared[index0 + 128] = placeholder1[index4];
int index5 = ((int)threadIdx.x + 192) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 3) % 7;
placeholder_shared[index0 + 192] = placeholder1[index5];
int index6 = ((int)threadIdx.x + 256) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 4) % 7;
placeholder_shared[index0 + 256] = placeholder1[index6];
int index7 = ((int)threadIdx.x + 320) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 5) % 7;
placeholder_shared[index0 + 320] = placeholder1[index7];
int index8 = ((int)threadIdx.x + 384) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 6) % 7;
placeholder_shared[index0 + 384] = placeholder1[index8];
placeholder_shared[index0 + 448] = placeholder1[index2 + 50176];
int index9 = ((int)threadIdx.x + 512) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 1) % 7;
placeholder_shared[index0 + 512] = placeholder1[index9];
int index10 = ((int)threadIdx.x + 576) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 2) % 7;
placeholder_shared[index0 + 576] = placeholder1[index10];
int index11 = ((int)threadIdx.x + 640) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 3) % 7;
placeholder_shared[index0 + 640] = placeholder1[index11];
int index12 = ((int)threadIdx.x + 704) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 4) % 7;
placeholder_shared[index0 + 704] = placeholder1[index12];
int index13 = ((int)threadIdx.x + 768) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 5) % 7;
placeholder_shared[index0 + 768] = placeholder1[index13];
int index14 = ((int)threadIdx.x + 832) / 7 * 784 + k_outer_outer * 7 + ((int)threadIdx.x + 6) % 7;
placeholder_shared[index0 + 832] = placeholder1[index14];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 14];
int index15 = (int)threadIdx.x * 14 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 14 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index16];
int index17 = (int)threadIdx.x * 14 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index17];
int index18 = (int)threadIdx.x * 14 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index18];
int index19 = (int)threadIdx.x * 14 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index19];
int index20 = (int)threadIdx.x * 14 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index20];
int index21 = (int)threadIdx.x * 14 + 7;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index21];
int index22 = (int)threadIdx.x * 14 + 8;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index22];
int index23 = (int)threadIdx.x * 14 + 9;
T_dense[1] = T_dense[1] + placeholder_d_shared[2] * placeholder_shared[index23];
int index24 = (int)threadIdx.x * 14 + 10;
T_dense[1] = T_dense[1] + placeholder_d_shared[3] * placeholder_shared[index24];
int index25 = (int)threadIdx.x * 14 + 11;
T_dense[1] = T_dense[1] + placeholder_d_shared[4] * placeholder_shared[index25];
int index26 = (int)threadIdx.x * 14 + 12;
T_dense[1] = T_dense[1] + placeholder_d_shared[5] * placeholder_shared[index26];
int index27 = (int)threadIdx.x * 14 + 13;
T_dense[1] = T_dense[1] + placeholder_d_shared[6] * placeholder_shared[index27];
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index28 = (int)threadIdx.x * 2 + ax1_inner;
int index29 = ax1_inner;
T_relu[index28] = max(T_dense[index29] + placeholder2[index28], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp7v6dpfxn | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<118>;
.reg .b32 %r<116>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[1568];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd3, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd5, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 784;
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r19, %r18;
mov.u32 %r111, 0;
mov.f32 %f110, 0f00000000;
cvta.to.global.u64 %rd7, %rd3;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(111+i*1), %f110;
$L__BB0_1:
setp.gt.s32 %p1, %r1, 7;
bar.sync 0;
mul.lo.s32 %r5, %r111, 392;
add.s32 %r21, %r5, %r1;
mul.wide.s32 %rd8, %r21, 4;
add.s64 %rd2, %rd7, %rd8;
ld.global.nc.f32 %f17, [%rd2];
mov.u32 %r23, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r24, %r23, %r18;
for.size.2 i in range(0, 11, 1):
st.shared.f32 [%r24+(0+i*128)], %f(17+i*1);
ld.global.nc.f32 %f(18+i*1), [%rd2+(128+i*128)];
st.shared.f32 [%r24+1408], %f28;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f29, [%rd2+1536];
mov.u32 %r25, %tid.x;
shl.b32 %r26, %r25, 2;
add.s32 %r28, %r23, %r26;
st.shared.f32 [%r28+1536], %f29;
$L__BB0_3:
add.s32 %r113, %r1, 224;
mov.u32 %r114, 512;
mov.u32 %r112, %r1;
$L__BB0_4:
mul.hi.s32 %r30, %r112, 1402438301;
shr.u32 %r31, %r30, 31;
shr.s32 %r32, %r30, 7;
add.s32 %r33, %r32, %r31;
mul.lo.s32 %r34, %r33, 392;
sub.s32 %r35, %r112, %r34;
add.s32 %r36, %r35, %r5;
mad.lo.s32 %r37, %r33, 784, %r36;
mul.wide.s32 %rd9, %r37, 4;
add.s64 %rd10, %rd1, %rd9;
ld.global.nc.f32 %f30, [%rd10];
add.s32 %r38, %r3, %r114;
for.size.13 i in range(0, 4, 1):
st.shared.f32 [%r38+(-512+i*128)], %f(30+i*1);
add.s32 %r(39+i*9), %r112, (32+i*32);
mul.hi.s32 %r(40+i*9), %r(39+i*9), 1402438301;
shr.u32 %r(41+i*9), %r(40+i*9), 31;
shr.s32 %r(42+i*9), %r(40+i*9), 7;
add.s32 %r(43+i*9), %r(42+i*9), %r(41+i*9);
mul.lo.s32 %r(44+i*9), %r(43+i*9), 392;
sub.s32 %r(45+i*9), %r(39+i*9), %r(44+i*9);
add.s32 %r(46+i*9), %r(45+i*9), %r5;
mad.lo.s32 %r(47+i*9), %r(43+i*9), 784, %r(46+i*9);
mul.wide.s32 %rd(11+i*2), %r(47+i*9), 4;
add.s64 %rd(12+i*2), %rd1, %rd(11+i*2);
ld.global.nc.f32 %f(31+i*1), [%rd(12+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.f32 [%r38+(0+i*128)], %f(34+i*1);
add.s32 %r(75+i*9), %r113, (-64+i*32);
mul.hi.s32 %r(76+i*9), %r(75+i*9), 1402438301;
shr.u32 %r(77+i*9), %r(76+i*9), 31;
shr.s32 %r(78+i*9), %r(76+i*9), 7;
add.s32 %r(79+i*9), %r(78+i*9), %r(77+i*9);
mul.lo.s32 %r(80+i*9), %r(79+i*9), 392;
sub.s32 %r(81+i*9), %r(75+i*9), %r(80+i*9);
add.s32 %r(82+i*9), %r(81+i*9), %r5;
mad.lo.s32 %r(83+i*9), %r(79+i*9), 784, %r(82+i*9);
mul.wide.s32 %rd(19+i*2), %r(83+i*9), 4;
add.s64 %rd(20+i*2), %rd1, %rd(19+i*2);
ld.global.nc.f32 %f(35+i*1), [%rd(20+i*2)+0];
st.shared.f32 [%r38+256], %f36;
mul.hi.s32 %r93, %r113, 1402438301;
shr.u32 %r94, %r93, 31;
shr.s32 %r95, %r93, 7;
add.s32 %r96, %r95, %r94;
mul.lo.s32 %r97, %r96, 392;
sub.s32 %r98, %r113, %r97;
add.s32 %r99, %r98, %r5;
mad.lo.s32 %r100, %r96, 784, %r99;
mul.wide.s32 %rd23, %r100, 4;
add.s64 %rd24, %rd1, %rd23;
ld.global.nc.f32 %f37, [%rd24];
st.shared.f32 [%r38+384], %f37;
for.size.1 i in range(0, 2, 1):
add.s32 %r(113+i*-1), %r(113+i*-1), 256;
add.s32 %r114, %r114, 1024;
setp.ne.s32 %p2, %r114, 201216;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r115, 0;
$L__BB0_6:
mul.lo.s32 %r102, %r115, 7;
shl.b32 %r103, %r102, 2;
for.size.1 i in range(0, 2, 1):
add.s32 %r(105+i*1), %r(23+i*79), %r(103+i*-101);
shl.b32 %r107, %r106, 2;
add.s32 %r109, %r19, %r107;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(38+i*1), [%r(109+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(40+i*2), %f39, %f(38+i*3), %f(113+i*-2);
ld.shared.f32 %f(41+i*2), [%r109+(100352+i*-100348)];
for.size.5 i in range(0, 5, 1):
ld.shared.f32 %f(44+i*5), [%r105+(4+i*4)];
fma.rn.f32 %f(45+i*5), %f(44+i*5), %f(43+i*5), %f(40+i*5);
ld.shared.f32 %f(46+i*5), [%r109+(100356+i*4)];
fma.rn.f32 %f(47+i*5), %f(44+i*5), %f(46+i*5), %f(42+i*5);
ld.shared.f32 %f(48+i*5), [%r109+(8+i*4)];
ld.shared.f32 %f69, [%r105+24];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(113+i*-2), %f69, %f(68+i*2), %f(65+i*2);
ld.shared.f32 %f(70+i*1), [%r109+(100376+i*-98808)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(72+i*2), %f39, %f(71+i*2), %f(112+i*-2);
ld.shared.f32 %f(73+i*2), [%r109+(101920+i*-100348)];
for.size.3 i in range(0, 5, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(76+i*4+j*2), %f(44+j*5), %f(75+i*4+j*2), %f(72+i*4+j*2);
ld.shared.f32 %f(77+i*4+j*2), [%r109+(101924+i*4+j*-100348)];
fma.rn.f32 %f112, %f69, %f95, %f92;
ld.shared.f32 %f96, [%r109+101944];
fma.rn.f32 %f110, %f69, %f96, %f94;
add.s32 %r115, %r115, 1;
setp.ne.s32 %p3, %r115, 56;
@%p3 bra $L__BB0_6;
add.s32 %r111, %r111, 1;
setp.lt.u32 %p4, %r111, 2;
@%p4 bra $L__BB0_1;
shl.b32 %r110, %r1, 1;
cvta.to.global.u64 %rd25, %rd5;
mul.wide.s32 %rd26, %r110, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f97, [%rd27];
add.f32 %f98, %f113, %f97;
mov.f32 %f99, 0f00000000;
max.f32 %f100, %f98, %f99;
cvta.to.global.u64 %rd28, %rd4;
add.s64 %rd29, %rd28, %rd26;
for.size.4 i in range(0, 2, 1):
st.global.f32 [%rd29+(0+i*256)], %f(100+i*3);
ld.global.nc.f32 %f(101+i*3), [%rd27+(256+i*-252)];
add.f32 %f(102+i*3), %f(111+i*1), %f(101+i*3);
max.f32 %f(103+i*3), %f(102+i*3), %f99;
st.global.f32 [%rd29+4], %f106;
ld.global.nc.f32 %f107, [%rd27+260];
add.f32 %f108, %f110, %f107;
max.f32 %f109, %f108, %f99;
st.global.f32 [%rd29+260], %f109;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[392];
__shared__ float placeholder_shared[50176];
T_dense[0] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 2; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 392 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
placeholder_d_shared[index0 + 128] = placeholder[index1 + 128];
placeholder_d_shared[index0 + 160] = placeholder[index1 + 160];
placeholder_d_shared[index0 + 192] = placeholder[index1 + 192];
placeholder_d_shared[index0 + 224] = placeholder[index1 + 224];
placeholder_d_shared[index0 + 256] = placeholder[index1 + 256];
placeholder_d_shared[index0 + 288] = placeholder[index1 + 288];
placeholder_d_shared[index0 + 320] = placeholder[index1 + 320];
placeholder_d_shared[index0 + 352] = placeholder[index1 + 352];
if ((int)threadIdx.x < 8) {
placeholder_d_shared[index0 + 384] = placeholder[index1 + 384];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1568; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) / 392 * 784 + k_outer_outer * 392 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) % 392;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 56; ++k_outer_inner) {
for (int j_outer_inner = 0; j_outer_inner < 2; ++j_outer_inner) {
int index4 = j_outer_inner;
int index5 = k_outer_inner * 7;
int index6 = (int)threadIdx.x * 784 + j_outer_inner * 392 + k_outer_inner * 7;
T_dense[index4] = T_dense[index4] + placeholder_d_shared[index5] * placeholder_shared[index6];
T_dense[index4 + 2] = T_dense[index4 + 2] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 25088]);
T_dense[index4] = T_dense[index4] + (placeholder_d_shared[index5 + 1] * placeholder_shared[index6 + 1]);
T_dense[index4 + 2] = T_dense[index4 + 2] + (placeholder_d_shared[index5 + 1] * placeholder_shared[index6 + 25089]);
T_dense[index4] = T_dense[index4] + (placeholder_d_shared[index5 + 2] * placeholder_shared[index6 + 2]);
T_dense[index4 + 2] = T_dense[index4 + 2] + (placeholder_d_shared[index5 + 2] * placeholder_shared[index6 + 25090]);
T_dense[index4] = T_dense[index4] + (placeholder_d_shared[index5 + 3] * placeholder_shared[index6 + 3]);
T_dense[index4 + 2] = T_dense[index4 + 2] + (placeholder_d_shared[index5 + 3] * placeholder_shared[index6 + 25091]);
T_dense[index4] = T_dense[index4] + (placeholder_d_shared[index5 + 4] * placeholder_shared[index6 + 4]);
T_dense[index4 + 2] = T_dense[index4 + 2] + (placeholder_d_shared[index5 + 4] * placeholder_shared[index6 + 25092]);
T_dense[index4] = T_dense[index4] + (placeholder_d_shared[index5 + 5] * placeholder_shared[index6 + 5]);
T_dense[index4 + 2] = T_dense[index4 + 2] + (placeholder_d_shared[index5 + 5] * placeholder_shared[index6 + 25093]);
T_dense[index4] = T_dense[index4] + (placeholder_d_shared[index5 + 6] * placeholder_shared[index6 + 6]);
T_dense[index4 + 2] = T_dense[index4 + 2] + (placeholder_d_shared[index5 + 6] * placeholder_shared[index6 + 25094]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index7 = (int)threadIdx.x * 2 + ax1_inner;
int index8 = ax1_inner;
T_relu[index7] = max(T_dense[index8] + placeholder2[index7], 0.000000e+00f);
T_relu[index7 + 64] = max(T_dense[index8 + 2] + placeholder2[index7 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp47pvbozh | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.local .align 16 .b8 __local_depot0[64];
.reg .b64 %SP;
.reg .b64 %SPL;
.reg .pred %p<5>;
.reg .f32 %f<227>;
.reg .b32 %r<298>;
.reg .b64 %rd<22>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[401408];
mov.u64 %SPL, __local_depot0;
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd12, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd9;
cvta.to.global.u64 %rd2, %rd11;
add.u64 %rd3, %SPL, 0;
mov.f32 %f29, 0f00000000;
for.size.2 i in range(0, 2, 1):
for.size.1 j in range(0, 2, 1):
st.local.v4.f32 [%rd3+(0+i*16+j*32)], {%f29, %f29, %f29, %f29};
cvta.to.global.u64 %rd5, %rd10;
cvta.to.global.u64 %rd6, %rd12;
mov.u32 %r293, 0;
mov.u32 %r1, %tid.x;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
$L__BB0_1:
shl.b32 %r16, %r293, 3;
add.s32 %r17, %r16, %r1;
mul.wide.s32 %rd14, %r17, 4;
add.s64 %rd15, %rd1, %rd14;
ld.global.nc.f32 %f30, [%rd15];
shl.b32 %r18, %r17, 2;
add.s32 %r20, %r19, %r18;
st.shared.f32 [%r20], %f30;
ld.global.nc.f32 %f31, [%rd15+32];
st.shared.f32 [%r20+32], %f31;
add.s32 %r293, %r293, 2;
setp.ne.s32 %p1, %r293, 98;
@%p1 bra $L__BB0_1;
shl.b32 %r4, %r1, 3;
mov.u32 %r294, 0;
$L__BB0_3:
shl.b32 %r22, %r294, 6;
add.s32 %r23, %r22, %r4;
shl.b32 %r24, %r23, 2;
mov.u32 %r25, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r26, %r25, %r24;
mul.wide.s32 %rd16, %r23, 4;
add.s64 %rd17, %rd5, %rd16;
for.size.4 i in range(0, 16, 1):
for.size.1 j in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(27+i*16+j*8), %r(28+i*16+j*8), %r(29+i*16+j*8), %r(30+i*16+j*8)}, [%rd17+(0+i*256+j*16)];
for.size.1 j in range(0, 2, 1):
st.shared.v4.u32 [%r26+(0+i*256+j*16)], {%r(27+i*16+j*8), %r(28+i*16+j*8), %r(29+i*16+j*8), %r(30+i*16+j*8)};
add.s32 %r294, %r294, 16;
setp.ne.s32 %p2, %r294, 1568;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r285, %r1, 25088, %r25;
add.s32 %r7, %r285, 200812;
mov.u32 %r295, 0;
$L__BB0_5:
mov.u32 %r292, _ZZ24default_function_kernel0E20placeholder_d_shared;
mul.lo.s32 %r287, %r295, 112;
add.s32 %r289, %r292, %r287;
for.size.1 i in range(0, 28, 1):
ld.shared.f32 %f(1+i*1), [%r289+(0+i*4)];
add.s32 %r296, %r7, %r287;
mov.u32 %r297, 200704;
mov.u64 %rd21, %rd3;
$L__BB0_6:
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(32+i*3), [%r296+(-200812+i*200704)];
ld.local.f32 %f(33+i*3), [%rd21+(0+i*32)];
fma.rn.f32 %f(34+i*3), %f1, %f(32+i*3), %f(33+i*3);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(38+i*2), [%r296+(-200808+i*200704)];
fma.rn.f32 %f(39+i*2), %f2, %f(38+i*2), %f(34+i*3);
for.size.3 i in range(0, 26, 1):
for.size.2 j in range(0, 2, 1):
ld.shared.f32 %f(42+i*4+j*2), [%r296+(-200804+i*4+j*200704)];
fma.rn.f32 %f(43+i*4+j*2), %f(3+j*1), %f(42+i*4+j*2), %f(39+i*4+j*2);
for.size.1 i in range(0, 2, 1):
st.local.f32 [%rd21+(0+i*32)], %f(143+i*2);
add.s64 %rd21, %rd21, 4;
for.size.1 i in range(0, 2, 1):
add.s32 %r(296+i*1), %r(296+i*1), 3136;
setp.ne.s32 %p3, %r297, 225792;
@%p3 bra $L__BB0_6;
add.s32 %r295, %r295, 1;
setp.lt.u32 %p4, %r295, 28;
@%p4 bra $L__BB0_5;
mov.u32 %r291, %tid.x;
shl.b32 %r290, %r291, 3;
ld.local.v4.f32 {%f146, %f147, %f148, %f149}, [%rd3];
mul.wide.s32 %rd18, %r290, 4;
add.s64 %rd19, %rd6, %rd18;
ld.global.nc.f32 %f154, [%rd19];
add.f32 %f155, %f146, %f154;
mov.f32 %f156, 0f00000000;
max.f32 %f157, %f155, %f156;
add.s64 %rd20, %rd2, %rd18;
st.global.f32 [%rd20], %f157;
ld.local.v4.f32 {%f158, %f159, %f160, %f161}, [%rd3+32];
for.size.5 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
ld.global.nc.f32 %f(166+i*6+j*3), [%rd19+(256+i*4+j*-252)];
add.f32 %f(167+i*6+j*3), %f(158+i*1+j*-11), %f(166+i*6+j*3);
max.f32 %f(168+i*6+j*3), %f(167+i*6+j*3), %f156;
st.global.f32 [%rd20+(256+i*4+j*-252)], %f(168+i*6+j*3);
for.size.5 i in range(0, 2, 1):
ld.global.nc.f32 %f(184+i*11), [%rd19+(268+i*-252)];
add.f32 %f(185+i*11), %f(161+i*26), %f(184+i*11);
max.f32 %f(186+i*11), %f(185+i*11), %f156;
st.global.f32 [%rd20+(268+i*-252)], %f(186+i*11);
ld.local.v4.f32 {%f(187+i*11), %f(188+i*11), %f(189+i*11), %f(190+i*11)}, [%rd3+(16+i*32)];
for.size.5 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
ld.global.nc.f32 %f(206+i*6+j*3), [%rd19+(272+i*4+j*-252)];
add.f32 %f(207+i*6+j*3), %f(198+i*1+j*-10), %f(206+i*6+j*3);
max.f32 %f(208+i*6+j*3), %f(207+i*6+j*3), %f156;
st.global.f32 [%rd20+(272+i*4+j*-252)], %f(208+i*6+j*3);
ld.global.nc.f32 %f224, [%rd19+284];
add.f32 %f225, %f201, %f224;
max.f32 %f226, %f225, %f156;
st.global.f32 [%rd20+284], %f226;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[100352];
for (int j_outer_inner_init = 0; j_outer_inner_init < 8; ++j_outer_inner_init) {
int index0 = j_outer_inner_init;
T_dense[index0] = 0.000000e+00f;
T_dense[index0 + 8] = 0.000000e+00f;
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index1];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 1568; ++ax0_ax1_fused_outer_outer1) {
ulonglong4*(placeholder_shared + ax0_ax1_fused_outer_outer1 * 64 + (int)threadIdx.x * 8)[0] = ulonglong4*(placeholder1 + ax0_ax1_fused_outer_outer1 * 64 + (int)threadIdx.x * 8)[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 28; ++k_outer_inner) {
for (int j_outer_inner = 0; j_outer_inner < 8; ++j_outer_inner) {
for (int k_inner = 0; k_inner < 28; ++k_inner) {
int index2 = j_outer_inner;
int index3 = k_outer_inner * 28 + k_inner;
int index4 = (int)threadIdx.x * 6272 + j_outer_inner * 784 + k_outer_inner * 28 + k_inner;
T_dense[index2] = T_dense[index2] + placeholder_d_shared[index3] * placeholder_shared[index4];
T_dense[index2 + 8] = T_dense[index2 + 8] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 50176]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 8; ++ax1_inner) {
int index5 = (int)threadIdx.x * 8 + ax1_inner;
int index6 = ax1_inner;
T_relu[index5] = max(T_dense[index6] + placeholder2[index5], 0.000000e+00f);
T_relu[index5 + 64] = max(T_dense[index6 + 8] + placeholder2[index5 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp2kzu6hye | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<95>;
.reg .b32 %r<84>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[64];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[8192];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd11, [default_function_kernel0_param_1];
ld.param.u64 %rd12, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd11;
cvta.to.global.u64 %rd1, %rd12;
mov.u32 %r7, %tid.x;
shl.b32 %r8, %r7, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r1, %r9, %r8;
shr.s32 %r10, %r7, 3;
shl.b32 %r11, %r7, 1;
and.b32 %r12, %r11, 14;
shl.b32 %r13, %r7, 3;
mov.u32 %r14, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r2, %r14, %r13;
shl.b32 %r15, %r7, 6;
add.s32 %r3, %r14, %r15;
cvt.s64.s32 %rd2, %r7;
cvta.to.global.u64 %rd14, %rd10;
mul.wide.s32 %rd15, %r7, 4;
add.s64 %rd23, %rd14, %rd15;
mul.lo.s32 %r16, %r10, 784;
or.b32 %r17, %r12, %r16;
mul.wide.s32 %rd16, %r17, 4;
add.s64 %rd17, %rd13, %rd16;
add.s64 %rd22, %rd17, 200704;
mov.f32 %f93, 0f00000000;
mov.u32 %r83, 0;
mov.f32 %f94, %f93;
$L__BB0_1:
cvt.u32.u64 %r18, %rd2;
setp.gt.s32 %p1, %r18, 15;
bar.sync 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f7, [%rd23];
st.shared.f32 [%r1], %f7;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.v2.u32 {%r(19+i*4), %r(20+i*4)}, [%rd22+(-200704+i*25088)];
st.shared.v2.u32 [%r2+(0+i*512)], {%r(19+i*4), %r(20+i*4)};
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(8+i*5), [%r3+(0+i*4)];
ld.shared.f32 %f(9+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(10+i*5), %f(9+i*5), %f(8+i*5), %f(93+i*-83);
ld.shared.f32 %f(11+i*5), [%r3+(4096+i*4)];
fma.rn.f32 %f(12+i*5), %f(9+i*5), %f(11+i*5), %f(94+i*-82);
for.size.5 i in range(0, 13, 1):
ld.shared.f32 %f(18+i*5), [%r3+(8+i*4)];
ld.shared.f32 %f(19+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(20+i*5), %f(19+i*5), %f(18+i*5), %f(15+i*5);
ld.shared.f32 %f(21+i*5), [%r3+(4104+i*4)];
fma.rn.f32 %f(22+i*5), %f(19+i*5), %f(21+i*5), %f(17+i*5);
ld.shared.f32 %f83, [%r3+60];
ld.shared.f32 %f84, [_ZZ24default_function_kernel0E20placeholder_d_shared+60];
fma.rn.f32 %f93, %f84, %f83, %f80;
ld.shared.f32 %f85, [%r3+4156];
fma.rn.f32 %f94, %f84, %f85, %f82;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(23+i*-1), %rd(23+i*-1), 64;
add.s32 %r83, %r83, 1;
setp.ne.s32 %p2, %r83, 49;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd18, %rd9;
shl.b64 %rd19, %rd2, 2;
add.s64 %rd20, %rd18, %rd19;
ld.global.nc.f32 %f86, [%rd20];
add.f32 %f87, %f93, %f86;
mov.f32 %f88, 0f00000000;
max.f32 %f89, %f87, %f88;
add.s64 %rd21, %rd1, %rd19;
st.global.f32 [%rd21], %f89;
ld.global.nc.f32 %f90, [%rd20+256];
add.f32 %f91, %f94, %f90;
max.f32 %f92, %f91, %f88;
st.global.f32 [%rd21+256], %f92;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[16];
__shared__ float placeholder_shared[2048];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 49; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 16 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 16; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer * 128 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder1 + ax0_ax1_fused_outer_outer * 6272 + ((int)threadIdx.x >> 3) * 784 + k_outer_outer * 16 + ((int)threadIdx.x & 7) * 2))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 16; ++k_outer_inner) {
int index2 = k_outer_inner;
int index3 = (int)threadIdx.x * 16 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 1024]);
}
}
int index0 = (int)threadIdx.x;
T_relu[index0] = max(T_dense[0] + placeholder2[index0], 0.000000e+00f);
T_relu[index0 + 64] = max(T_dense[1] + placeholder2[index0 + 64], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp3ecfhtar | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<94>;
.reg .b32 %r<221>;
.reg .b64 %rd<47>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[112];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[1792];
ld.param.u64 %rd22, [default_function_kernel0_param_0];
ld.param.u64 %rd23, [default_function_kernel0_param_1];
ld.param.u64 %rd24, [default_function_kernel0_param_2];
ld.param.u64 %rd21, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd46, %rd23;
cvta.to.global.u64 %rd2, %rd24;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(9+i*1), %r1, (1+i*2);
mov.u32 %r11, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r11, %r10;
mov.u32 %r12, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r12, %r10;
mul.hi.s32 %r13, %r1, -1840700269;
add.s32 %r14, %r13, %r1;
shr.u32 %r15, %r14, 31;
shr.s32 %r16, %r14, 3;
add.s32 %r17, %r16, %r15;
mul.lo.s32 %r18, %r17, 14;
sub.s32 %r19, %r1, %r18;
shl.b32 %r20, %r19, 1;
for.size.6 i in range(0, 2, 1):
add.s32 %r(21+i*6), %r9, (32+i*-28);
mul.hi.s32 %r(22+i*6), %r(21+i*6), -1840700269;
add.s32 %r(23+i*6), %r(22+i*6), %r(21+i*6);
shr.u32 %r(24+i*6), %r(23+i*6), 31;
shr.s32 %r(25+i*6), %r(23+i*6), 4;
add.s32 %r(26+i*6), %r(25+i*6), %r(24+i*6);
for.size.14 i in range(0, 5, 1):
mul.lo.s32 %r(33+i*14), %r(32+i*14), 28;
sub.s32 %r(34+i*14), %r(27+i*14), %r(33+i*14);
add.s32 %r(35+i*14), %r9, (64+i*32);
mul.hi.s32 %r(36+i*14), %r(35+i*14), -1840700269;
add.s32 %r(37+i*14), %r(36+i*14), %r(35+i*14);
shr.u32 %r(38+i*14), %r(37+i*14), 31;
shr.s32 %r(39+i*14), %r(37+i*14), 4;
add.s32 %r(40+i*14), %r(39+i*14), %r(38+i*14);
add.s32 %r(41+i*14), %r9, (8+i*4);
mul.hi.s32 %r(42+i*14), %r(41+i*14), -1840700269;
add.s32 %r(43+i*14), %r(42+i*14), %r(41+i*14);
shr.u32 %r(44+i*14), %r(43+i*14), 31;
shr.s32 %r(45+i*14), %r(43+i*14), 4;
add.s32 %r(46+i*14), %r(45+i*14), %r(44+i*14);
mul.lo.s32 %r103, %r102, 28;
sub.s32 %r104, %r97, %r103;
for.size.6 i in range(0, 6, 1):
add.s32 %r(105+i*6), %r9, (256+i*32);
mul.hi.s32 %r(106+i*6), %r(105+i*6), -1840700269;
add.s32 %r(107+i*6), %r(106+i*6), %r(105+i*6);
shr.u32 %r(108+i*6), %r(107+i*6), 31;
shr.s32 %r(109+i*6), %r(107+i*6), 4;
add.s32 %r(110+i*6), %r(109+i*6), %r(108+i*6);
mad.lo.s32 %r4, %r1, 112, %r12;
mov.u32 %r5, %ctaid.x;
mul.lo.s32 %r141, %r5, 12544;
for.size.3 i in range(0, 6, 1):
add.s32 %r(142+i*2), %r(104+i*-14), %r141;
mad.lo.s32 %r(143+i*2), %r(140+i*-6), 784, %r(142+i*2);
mul.wide.s32 %rd(3+i*1), %r(143+i*2), 4;
for.size.2 i in range(0, 6, 1):
mad.lo.s32 %r(154+i*1), %r(96+i*-14), 784, %r(142+i*2);
mul.wide.s32 %rd(9+i*1), %r(154+i*1), 4;
add.s32 %r160, %r20, %r141;
mad.lo.s32 %r161, %r17, 784, %r160;
mul.wide.s32 %rd15, %r161, 4;
cvta.to.global.u64 %rd25, %rd22;
mul.wide.s32 %rd26, %r9, 4;
add.s64 %rd27, %rd25, %rd26;
add.s64 %rd45, %rd27, 4;
mov.f32 %f93, 0f00000000;
mov.u32 %r220, 0;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 13;
@%p1 bra $L__BB0_3;
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(4+i*1), [%rd45+(-4+i*4)];
st.shared.f32 [%r2+(0+i*4)], %f(4+i*1);
$L__BB0_3:
for.size.3 i in range(0, 7, 1):
add.s64 %rd(28+i*1), %rd46, %rd(15+i*-1);
ld.global.nc.v2.u32 {%r(162+i*4), %r(163+i*4)}, [%rd(28+i*1)+0];
st.shared.v2.u32 [%r3+(0+i*128)], {%r(162+i*4), %r(163+i*4)};
ld.global.nc.v2.u32 {%r190, %r191}, [%rd28+25088];
for.size.3 i in range(0, 6, 1):
st.shared.v2.u32 [%r3+(896+i*128)], {%r(190+i*4), %r(191+i*4)};
add.s64 %rd(35+i*1), %rd46, %rd(8+i*-1);
ld.global.nc.v2.u32 {%r(194+i*4), %r(195+i*4)}, [%rd(35+i*1)+0];
st.shared.v2.u32 [%r3+1664], {%r214, %r215};
bar.sync 0;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(6+i*3), [%r4+(0+i*4)];
ld.shared.f32 %f(7+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(8+i*3), %f(7+i*3), %f(6+i*3), %f(93+i*-85);
for.size.3 i in range(0, 25, 1):
ld.shared.f32 %f(12+i*3), [%r4+(8+i*4)];
ld.shared.f32 %f(13+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(14+i*3), %f(13+i*3), %f(12+i*3), %f(11+i*3);
ld.shared.f32 %f87, [%r4+108];
ld.shared.f32 %f88, [_ZZ24default_function_kernel0E20placeholder_d_shared+108];
fma.rn.f32 %f93, %f88, %f87, %f86;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(46+i*-1), %rd(46+i*-1), 112;
add.s32 %r220, %r220, 1;
setp.ne.s32 %p2, %r220, 28;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd41, %rd21;
shl.b32 %r218, %r5, 4;
add.s32 %r219, %r218, %r1;
mul.wide.s32 %rd42, %r219, 4;
add.s64 %rd43, %rd41, %rd42;
ld.global.nc.f32 %f89, [%rd43];
add.f32 %f90, %f93, %f89;
mov.f32 %f91, 0f00000000;
max.f32 %f92, %f90, %f91;
add.s64 %rd44, %rd2, %rd42;
st.global.f32 [%rd44], %f92;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[28];
__shared__ float placeholder_shared[448];
T_dense[0] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 28; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 2;
int index1 = k_outer_outer * 28 + (int)threadIdx.x * 2;
placeholder_d_shared[index0] = placeholder[index1];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 2;
int index1 = k_outer_outer * 28 + (int)threadIdx.x * 2;
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
}
((float2*)(placeholder_shared + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + (int)threadIdx.x / 14 * 784 + k_outer_outer * 28 + (int)threadIdx.x % 14 * 2))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 32))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 32) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 4) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 64))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 64) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 8) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 96))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 96) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 12) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 128))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 128) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 16) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 160))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 160) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 20) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 192))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 192) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 24) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 224))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + (int)threadIdx.x / 14 * 784 + k_outer_outer * 28 + (int)threadIdx.x % 14 * 2 + 6272))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 256))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 256) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 4) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 288))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 288) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 8) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 320))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 320) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 12) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 352))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 352) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 16) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 384))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 384) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 20) % 28))[0];
((float2*)(placeholder_shared + (int)threadIdx.x * 2 + 416))[0] = ((float2*)(placeholder1 + (int)blockIdx.x * 12544 + ((int)threadIdx.x * 2 + 416) / 28 * 784 + k_outer_outer * 28 + ((int)threadIdx.x * 2 + 24) % 28))[0];
__syncthreads();
int index2 = (int)threadIdx.x * 28;
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index2];
T_dense[0] = T_dense[0] + (placeholder_d_shared[1] * placeholder_shared[index2 + 1]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[2] * placeholder_shared[index2 + 2]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[3] * placeholder_shared[index2 + 3]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[4] * placeholder_shared[index2 + 4]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[5] * placeholder_shared[index2 + 5]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[6] * placeholder_shared[index2 + 6]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[7] * placeholder_shared[index2 + 7]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[8] * placeholder_shared[index2 + 8]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[9] * placeholder_shared[index2 + 9]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[10] * placeholder_shared[index2 + 10]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[11] * placeholder_shared[index2 + 11]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[12] * placeholder_shared[index2 + 12]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[13] * placeholder_shared[index2 + 13]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[14] * placeholder_shared[index2 + 14]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[15] * placeholder_shared[index2 + 15]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[16] * placeholder_shared[index2 + 16]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[17] * placeholder_shared[index2 + 17]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[18] * placeholder_shared[index2 + 18]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[19] * placeholder_shared[index2 + 19]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[20] * placeholder_shared[index2 + 20]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[21] * placeholder_shared[index2 + 21]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[22] * placeholder_shared[index2 + 22]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[23] * placeholder_shared[index2 + 23]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[24] * placeholder_shared[index2 + 24]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[25] * placeholder_shared[index2 + 25]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[26] * placeholder_shared[index2 + 26]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[27] * placeholder_shared[index2 + 27]);
}
int index3 = (int)blockIdx.x * 16 + (int)threadIdx.x;
T_relu[index3] = max(T_dense[0] + placeholder2[index3], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp0z5qj862 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<66>;
.reg .b32 %r<20>;
.reg .b64 %rd<20>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E7T_dense[256];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E8red_buf0[256];
ld.param.u64 %rd7, [default_function_kernel0_param_0];
ld.param.u64 %rd8, [default_function_kernel0_param_1];
ld.param.u64 %rd9, [default_function_kernel0_param_2];
ld.param.u64 %rd6, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd10, %rd8;
cvta.to.global.u64 %rd1, %rd9;
mov.u32 %r1, %tid.x;
cvta.to.global.u64 %rd11, %rd7;
mul.wide.s32 %rd12, %r1, 4;
add.s64 %rd2, %rd11, %rd12;
for.size.1 i in range(0, 12, 1):
ld.global.nc.f32 %f(1+i*1), [%rd2+(0+i*256)];
shl.b32 %r10, %r1, 2;
mov.u32 %r11, _ZZ24default_function_kernel0E8red_buf0;
add.s32 %r2, %r11, %r10;
mov.u32 %r3, %ctaid.x;
mad.lo.s32 %r12, %r3, 50176, %r1;
mul.wide.s32 %rd13, %r12, 4;
add.s64 %rd14, %rd10, %rd13;
add.s64 %rd19, %rd14, 1536;
mov.u32 %r19, 0;
mov.u32 %r18, _ZZ24default_function_kernel0E7T_dense;
$L__BB0_1:
ld.global.nc.f32 %f16, [%rd19+-1536];
fma.rn.f32 %f17, %f1, %f16, 0f00000000;
for.size.2 i in range(0, 10, 1):
ld.global.nc.f32 %f(18+i*2), [%rd19+(-1280+i*256)];
fma.rn.f32 %f(19+i*2), %f(2+i*1), %f(18+i*2), %f(17+i*2);
ld.global.nc.f32 %f38, [%rd19+1280];
fma.rn.f32 %f65, %f12, %f38, %f37;
setp.gt.s32 %p1, %r1, 15;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f39, [%rd19+1536];
ld.global.nc.f32 %f40, [%rd2+3072];
fma.rn.f32 %f65, %f40, %f39, %f65;
$L__BB0_3:
bar.sync 0;
st.volatile.shared.f32 [%r2], %f65;
bar.sync 0;
setp.gt.s32 %p2, %r1, 31;
@%p2 bra $L__BB0_5;
for.size.1 i in range(0, 2, 1):
ld.volatile.shared.f32 %f(41+i*1), [%r2+(0+i*128)];
add.f32 %f43, %f41, %f42;
st.volatile.shared.f32 [%r2], %f43;
$L__BB0_5:
bar.sync 0;
@%p1 bra $L__BB0_7;
for.size.1 i in range(0, 2, 1):
ld.volatile.shared.f32 %f(44+i*1), [%r2+(0+i*64)];
for.size.4 i in range(0, 2, 1):
add.f32 %f(46+i*3), %f(44+i*4), %f(45+i*2);
st.volatile.shared.f32 [%r2+(0+i*0)], %f(46+i*3);
ld.volatile.shared.f32 %f(47+i*3), [%r2+(32+i*-16)];
ld.volatile.shared.f32 %f(48+i*3), [%r2+(0+i*0)];
for.size.4 i in range(0, 2, 1):
add.f32 %f(52+i*3), %f(51+i*3), %f(50+i*3);
st.volatile.shared.f32 [%r2+(0+i*0)], %f(52+i*3);
ld.volatile.shared.f32 %f(53+i*3), [%r2+(8+i*-4)];
ld.volatile.shared.f32 %f(54+i*3), [%r2+(0+i*0)];
add.f32 %f58, %f57, %f56;
st.volatile.shared.f32 [%r2], %f58;
$L__BB0_7:
bar.sync 0;
ld.volatile.shared.f32 %f59, [_ZZ24default_function_kernel0E8red_buf0];
st.shared.f32 [%r18], %f59;
add.s32 %r18, %r18, 4;
add.s64 %rd19, %rd19, 3136;
add.s32 %r19, %r19, 1;
setp.ne.s32 %p4, %r19, 64;
@%p4 bra $L__BB0_1;
cvta.to.global.u64 %rd15, %rd6;
bar.sync 0;
shl.b32 %r13, %r3, 6;
add.s32 %r14, %r13, %r1;
mul.wide.s32 %rd16, %r14, 4;
add.s64 %rd17, %rd15, %rd16;
ld.global.nc.f32 %f60, [%rd17];
mov.u32 %r16, _ZZ24default_function_kernel0E7T_dense;
add.s32 %r17, %r16, %r10;
ld.shared.f32 %f61, [%r17];
add.f32 %f62, %f61, %f60;
mov.f32 %f63, 0f00000000;
max.f32 %f64, %f62, %f63;
add.s64 %rd18, %rd1, %rd16;
st.global.f32 [%rd18], %f64;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
__shared__ float T_dense[64];
float normal_reduce_temp0[1];
__shared__ float red_buf0[64];
for (int j = 0; j < 64; ++j) {
normal_reduce_temp0[0] = 0.000000e+00f;
int index0 = (int)threadIdx.x;
int index1 = (int)blockIdx.x * 50176 + j * 784 + (int)threadIdx.x;
normal_reduce_temp0[0] = normal_reduce_temp0[0] + placeholder[index0] * placeholder1[index1];
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 64] * placeholder1[index1 + 64]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 128] * placeholder1[index1 + 128]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 192] * placeholder1[index1 + 192]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 256] * placeholder1[index1 + 256]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 320] * placeholder1[index1 + 320]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 384] * placeholder1[index1 + 384]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 448] * placeholder1[index1 + 448]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 512] * placeholder1[index1 + 512]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 576] * placeholder1[index1 + 576]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 640] * placeholder1[index1 + 640]);
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 704] * placeholder1[index1 + 704]);
if ((int)threadIdx.x < 16) {
normal_reduce_temp0[0] = normal_reduce_temp0[0] + (placeholder[index0 + 768] * placeholder1[index1 + 768]);
}
__syncthreads();
((volatile float*)red_buf0)[index0] = normal_reduce_temp0[0];
__syncthreads();
if ((int)threadIdx.x < 32) {
((volatile float*)red_buf0)[index0] = ((volatile float*)red_buf0)[index0] + ((volatile float*)red_buf0)[index0 + 32];
}
__syncthreads();
if ((int)threadIdx.x < 16) {
((volatile float*)red_buf0)[index0] = ((volatile float*)red_buf0)[index0] + ((volatile float*)red_buf0)[index0 + 16];
((volatile float*)red_buf0)[index0] = ((volatile float*)red_buf0)[index0] + ((volatile float*)red_buf0)[index0 + 8];
((volatile float*)red_buf0)[index0] = ((volatile float*)red_buf0)[index0] + ((volatile float*)red_buf0)[index0 + 4];
((volatile float*)red_buf0)[index0] = ((volatile float*)red_buf0)[index0] + ((volatile float*)red_buf0)[index0 + 2];
((volatile float*)red_buf0)[index0] = ((volatile float*)red_buf0)[index0] + ((volatile float*)red_buf0)[index0 + 1];
}
__syncthreads();
int index2 = j;
T_dense[index2] = ((volatile float*)red_buf0)[0];
}
__syncthreads();
int index0 = (int)threadIdx.x;
int index3 = (int)blockIdx.x * 64 + (int)threadIdx.x;
T_relu[index3] = max(T_dense[index0] + placeholder2[index3], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmphn1275jx | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<121>;
.reg .b32 %r<111>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[100352];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd7, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd5;
cvta.to.global.u64 %rd3, %rd7;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 7;
shl.b32 %r21, %r2, 2;
mov.u32 %r22, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r22, %r21;
mul.lo.s32 %r4, %r1, 784;
add.s32 %r5, %r1, 224;
shl.b32 %r23, %r1, 2;
mov.u32 %r24, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r6, %r24, %r23;
mov.u32 %r106, 0;
mov.f32 %f113, 0f00000000;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(114+i*1), %f113;
$L__BB0_1:
bar.sync 0;
setp.lt.s32 %p1, %r1, 28;
@%p1 bra $L__BB0_2;
bra.uni $L__BB0_3;
$L__BB0_2:
mad.lo.s32 %r25, %r106, 196, %r2;
mul.wide.s32 %rd9, %r25, 4;
add.s64 %rd10, %rd2, %rd9;
for.size.2 i in range(0, 7, 1):
ld.global.nc.f32 %f(21+i*1), [%rd10+(0+i*4)];
st.shared.f32 [%r3+(0+i*4)], %f(21+i*1);
$L__BB0_3:
mul.lo.s32 %r8, %r106, 196;
mov.u32 %r109, 512;
mov.u32 %r107, %r1;
mov.u32 %r108, %r5;
$L__BB0_4:
mul.hi.s32 %r27, %r107, 1402438301;
shr.u32 %r28, %r27, 31;
shr.s32 %r29, %r27, 6;
add.s32 %r30, %r29, %r28;
mul.lo.s32 %r31, %r30, 196;
sub.s32 %r32, %r107, %r31;
add.s32 %r33, %r32, %r8;
mad.lo.s32 %r34, %r30, 784, %r33;
mul.wide.s32 %rd11, %r34, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f28, [%rd12];
add.s32 %r35, %r6, %r109;
for.size.13 i in range(0, 4, 1):
st.shared.f32 [%r35+(-512+i*128)], %f(28+i*1);
add.s32 %r(36+i*9), %r107, (32+i*32);
mul.hi.s32 %r(37+i*9), %r(36+i*9), 1402438301;
shr.u32 %r(38+i*9), %r(37+i*9), 31;
shr.s32 %r(39+i*9), %r(37+i*9), 6;
add.s32 %r(40+i*9), %r(39+i*9), %r(38+i*9);
mul.lo.s32 %r(41+i*9), %r(40+i*9), 196;
sub.s32 %r(42+i*9), %r(36+i*9), %r(41+i*9);
add.s32 %r(43+i*9), %r(42+i*9), %r8;
mad.lo.s32 %r(44+i*9), %r(40+i*9), 784, %r(43+i*9);
mul.wide.s32 %rd(13+i*2), %r(44+i*9), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(29+i*1), [%rd(14+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.f32 [%r35+(0+i*128)], %f(32+i*1);
add.s32 %r(72+i*9), %r108, (-64+i*32);
mul.hi.s32 %r(73+i*9), %r(72+i*9), 1402438301;
shr.u32 %r(74+i*9), %r(73+i*9), 31;
shr.s32 %r(75+i*9), %r(73+i*9), 6;
add.s32 %r(76+i*9), %r(75+i*9), %r(74+i*9);
mul.lo.s32 %r(77+i*9), %r(76+i*9), 196;
sub.s32 %r(78+i*9), %r(72+i*9), %r(77+i*9);
add.s32 %r(79+i*9), %r(78+i*9), %r8;
mad.lo.s32 %r(80+i*9), %r(76+i*9), 784, %r(79+i*9);
mul.wide.s32 %rd(21+i*2), %r(80+i*9), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(33+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r35+256], %f34;
mul.hi.s32 %r90, %r108, 1402438301;
shr.u32 %r91, %r90, 31;
shr.s32 %r92, %r90, 6;
add.s32 %r93, %r92, %r91;
mul.lo.s32 %r94, %r93, 196;
sub.s32 %r95, %r108, %r94;
add.s32 %r96, %r95, %r8;
mad.lo.s32 %r97, %r93, 784, %r96;
mul.wide.s32 %rd25, %r97, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f35, [%rd26];
st.shared.f32 [%r35+384], %f35;
for.size.1 i in range(0, 2, 1):
add.s32 %r(108+i*-1), %r(108+i*-1), 256;
add.s32 %r109, %r109, 1024;
setp.ne.s32 %p2, %r109, 100864;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r110, 0;
bra.uni $L__BB0_6;
$L__BB0_7:
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(68+i*1), [%r(17+i*-1)+16];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(70+i*2), %f69, %f(68+i*3), %f(116+i*-1);
ld.shared.f32 %f(71+i*2), [%r17+(800+i*784)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(74+i*2), %f69, %f(73+i*2), %f(114+i*-1);
ld.shared.f32 %f(75+i*2), [%r17+(2368+i*-2348)];
for.size.9 i in range(0, 2, 1):
ld.shared.f32 %f(78+i*9), [%r16+(20+i*4)];
fma.rn.f32 %f(79+i*9), %f(78+i*9), %f(77+i*9), %f(70+i*9);
ld.shared.f32 %f(80+i*9), [%r17+(804+i*4)];
fma.rn.f32 %f(81+i*9), %f(78+i*9), %f(80+i*9), %f(72+i*9);
ld.shared.f32 %f(82+i*9), [%r17+(1588+i*4)];
fma.rn.f32 %f(83+i*9), %f(78+i*9), %f(82+i*9), %f(74+i*9);
ld.shared.f32 %f(84+i*9), [%r17+(2372+i*4)];
fma.rn.f32 %f(85+i*9), %f(78+i*9), %f(84+i*9), %f(76+i*9);
ld.shared.f32 %f(86+i*9), [%r17+(24+i*4)];
ld.shared.f32 %f96, [%r16+28];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(116+i*-1), %f96, %f(95+i*2), %f(88+i*2);
ld.shared.f32 %f(97+i*1), [%r17+(812+i*784)];
fma.rn.f32 %f114, %f96, %f98, %f92;
ld.shared.f32 %f99, [%r17+2380];
fma.rn.f32 %f113, %f96, %f99, %f94;
add.s32 %r110, %r110, 2;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(99+i*1), %r110, (2+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(16+i*86), %r(22+i*77), %r(100+i*-96);
shl.b32 %r103, %r102, 2;
add.s32 %r17, %r24, %r103;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(36+i*1), [%r(17+i*-1)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(38+i*2), %f37, %f(36+i*3), %f(116+i*-1);
ld.shared.f32 %f(39+i*2), [%r17+(784+i*784)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(42+i*2), %f37, %f(41+i*2), %f(114+i*-1);
ld.shared.f32 %f(43+i*2), [%r17+(2352+i*-2348)];
for.size.9 i in range(0, 2, 1):
ld.shared.f32 %f(46+i*9), [%r16+(4+i*4)];
fma.rn.f32 %f(47+i*9), %f(46+i*9), %f(45+i*9), %f(38+i*9);
ld.shared.f32 %f(48+i*9), [%r17+(788+i*4)];
fma.rn.f32 %f(49+i*9), %f(46+i*9), %f(48+i*9), %f(40+i*9);
ld.shared.f32 %f(50+i*9), [%r17+(1572+i*4)];
fma.rn.f32 %f(51+i*9), %f(46+i*9), %f(50+i*9), %f(42+i*9);
ld.shared.f32 %f(52+i*9), [%r17+(2356+i*4)];
fma.rn.f32 %f(53+i*9), %f(46+i*9), %f(52+i*9), %f(44+i*9);
ld.shared.f32 %f(54+i*9), [%r17+(8+i*4)];
ld.shared.f32 %f64, [%r16+12];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(116+i*-1), %f64, %f(63+i*2), %f(56+i*2);
ld.shared.f32 %f(65+i*1), [%r17+(796+i*784)];
fma.rn.f32 %f114, %f64, %f66, %f60;
ld.shared.f32 %f67, [%r17+2364];
fma.rn.f32 %f113, %f64, %f67, %f62;
setp.eq.s32 %p3, %r110, 48;
@%p3 bra $L__BB0_8;
bra.uni $L__BB0_7;
$L__BB0_8:
add.s32 %r106, %r106, 1;
setp.lt.u32 %p4, %r106, 4;
@%p4 bra $L__BB0_1;
mul.wide.s32 %rd27, %r23, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f100, [%rd28];
add.f32 %f101, %f116, %f100;
mov.f32 %f102, 0f00000000;
max.f32 %f103, %f101, %f102;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd29+(0+i*4)], %f(103+i*3);
ld.global.nc.f32 %f(104+i*3), [%rd28+(4+i*4)];
add.f32 %f(105+i*3), %f(115+i*-1), %f(104+i*3);
max.f32 %f(106+i*3), %f(105+i*3), %f102;
st.global.f32 [%rd29+12], %f112;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[25088];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 28) {
int index0 = (int)threadIdx.x * 7;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 7;
placeholder_d_shared[index0] = placeholder[index1];
}
if ((int)threadIdx.x < 28) {
int index0 = (int)threadIdx.x * 7;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 7;
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
}
if ((int)threadIdx.x < 28) {
int index0 = (int)threadIdx.x * 7;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 7;
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
}
if ((int)threadIdx.x < 28) {
int index0 = (int)threadIdx.x * 7;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 7;
placeholder_d_shared[index0 + 3] = placeholder[index1 + 3];
}
if ((int)threadIdx.x < 28) {
int index0 = (int)threadIdx.x * 7;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 7;
placeholder_d_shared[index0 + 4] = placeholder[index1 + 4];
}
if ((int)threadIdx.x < 28) {
int index0 = (int)threadIdx.x * 7;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 7;
placeholder_d_shared[index0 + 5] = placeholder[index1 + 5];
}
if ((int)threadIdx.x < 28) {
int index0 = (int)threadIdx.x * 7;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 7;
placeholder_d_shared[index0 + 6] = placeholder[index1 + 6];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 784; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) % 196;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 49; ++k_outer_inner) {
int index4 = k_outer_inner * 4;
int index5 = (int)threadIdx.x * 784 + k_outer_inner * 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 196]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 392]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 588]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 197]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 393]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 589]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 2] * placeholder_shared[index5 + 2]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4 + 2] * placeholder_shared[index5 + 198]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4 + 2] * placeholder_shared[index5 + 394]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4 + 2] * placeholder_shared[index5 + 590]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 3] * placeholder_shared[index5 + 3]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4 + 3] * placeholder_shared[index5 + 199]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4 + 3] * placeholder_shared[index5 + 395]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4 + 3] * placeholder_shared[index5 + 591]);
}
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index6 = (int)threadIdx.x * 4 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp4gkbdxo7 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<102>;
.reg .b32 %r<83>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[4096];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd1, %rd11;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(8+i*1), %r1, (2+i*2);
mov.u32 %r10, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r10, %r9;
mov.u32 %r11, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r11, %r9;
shr.s32 %r12, %r1, 1;
and.b32 %r13, %r8, 4;
shl.b32 %r14, %r1, 6;
add.s32 %r4, %r11, %r14;
cvta.to.global.u64 %rd13, %rd9;
mul.wide.s32 %rd14, %r8, 4;
add.s64 %rd15, %rd13, %rd14;
add.s64 %rd23, %rd15, 8;
mul.lo.s32 %r15, %r12, 784;
or.b32 %r16, %r13, %r15;
mul.wide.s32 %rd16, %r16, 4;
add.s64 %rd17, %rd12, %rd16;
add.s64 %rd22, %rd17, 200704;
mov.u32 %r82, 0;
mov.f32 %f98, 0f00000000;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(99+i*1), %f98;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_3;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(13+i*1), [%rd23+(-8+i*4)];
st.shared.f32 [%r2+(0+i*4)], %f(13+i*1);
$L__BB0_3:
for.size.2 i in range(0, 8, 1):
ld.global.nc.v4.u32 {%r(17+i*8), %r(18+i*8), %r(19+i*8), %r(20+i*8)}, [%rd22+(-200704+i*50176)];
st.shared.v4.u32 [%r3+(0+i*512)], {%r(17+i*8), %r(18+i*8), %r(19+i*8), %r(20+i*8)};
bar.sync 0;
for.size.9 i in range(0, 2, 1):
ld.shared.f32 %f(17+i*9), [%r4+(0+i*4)];
ld.shared.f32 %f(18+i*9), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(19+i*9), %f(18+i*9), %f(17+i*9), %f(101+i*-82);
ld.shared.f32 %f(20+i*9), [%r4+(2048+i*4)];
fma.rn.f32 %f(21+i*9), %f(18+i*9), %f(20+i*9), %f(100+i*-79);
ld.shared.f32 %f(22+i*9), [%r4+(32+i*4)];
fma.rn.f32 %f(23+i*9), %f(18+i*9), %f(22+i*9), %f(99+i*-76);
ld.shared.f32 %f(24+i*9), [%r4+(2080+i*4)];
fma.rn.f32 %f(25+i*9), %f(18+i*9), %f(24+i*9), %f(98+i*-73);
for.size.9 i in range(0, 5, 1):
ld.shared.f32 %f(35+i*9), [%r4+(8+i*4)];
ld.shared.f32 %f(36+i*9), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(37+i*9), %f(36+i*9), %f(35+i*9), %f(28+i*9);
ld.shared.f32 %f(38+i*9), [%r4+(2056+i*4)];
fma.rn.f32 %f(39+i*9), %f(36+i*9), %f(38+i*9), %f(30+i*9);
ld.shared.f32 %f(40+i*9), [%r4+(40+i*4)];
fma.rn.f32 %f(41+i*9), %f(36+i*9), %f(40+i*9), %f(32+i*9);
ld.shared.f32 %f(42+i*9), [%r4+(2088+i*4)];
fma.rn.f32 %f(43+i*9), %f(36+i*9), %f(42+i*9), %f(34+i*9);
ld.shared.f32 %f80, [%r4+28];
ld.shared.f32 %f81, [_ZZ24default_function_kernel0E20placeholder_d_shared+28];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(101+i*-1), %f81, %f(80+i*2), %f(73+i*2);
ld.shared.f32 %f(82+i*1), [%r4+(2076+i*-2016)];
fma.rn.f32 %f99, %f81, %f83, %f77;
ld.shared.f32 %f84, [%r4+2108];
fma.rn.f32 %f98, %f81, %f84, %f79;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(23+i*-1), %rd(23+i*-1), 32;
add.s32 %r82, %r82, 1;
setp.ne.s32 %p2, %r82, 98;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd18, %rd8;
shl.b32 %r81, %r1, 1;
mul.wide.s32 %rd19, %r81, 4;
add.s64 %rd20, %rd18, %rd19;
ld.global.nc.f32 %f85, [%rd20];
add.f32 %f86, %f101, %f85;
mov.f32 %f87, 0f00000000;
max.f32 %f88, %f86, %f87;
add.s64 %rd21, %rd1, %rd19;
for.size.4 i in range(0, 2, 1):
st.global.f32 [%rd21+(0+i*256)], %f(88+i*3);
ld.global.nc.f32 %f(89+i*3), [%rd20+(256+i*-252)];
add.f32 %f(90+i*3), %f(100+i*-1), %f(89+i*3);
max.f32 %f(91+i*3), %f(90+i*3), %f87;
st.global.f32 [%rd21+4], %f94;
ld.global.nc.f32 %f95, [%rd20+260];
add.f32 %f96, %f98, %f95;
max.f32 %f97, %f96, %f87;
st.global.f32 [%rd21+260], %f97;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[1024];
T_dense[0] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 8 + (int)threadIdx.x * 4;
placeholder_d_shared[index0] = placeholder[index1];
}
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 8 + (int)threadIdx.x * 4;
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
}
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 8 + (int)threadIdx.x * 4;
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
}
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 8 + (int)threadIdx.x * 4;
placeholder_d_shared[index0 + 3] = placeholder[index1 + 3];
}
((float4*)(placeholder_shared + (int)threadIdx.x * 4))[0] = ((float4*)(placeholder1 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 1) * 4))[0];
((float4*)(placeholder_shared + (int)threadIdx.x * 4 + 128))[0] = ((float4*)(placeholder1 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 1) * 4 + 12544))[0];
((float4*)(placeholder_shared + (int)threadIdx.x * 4 + 256))[0] = ((float4*)(placeholder1 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 1) * 4 + 25088))[0];
((float4*)(placeholder_shared + (int)threadIdx.x * 4 + 384))[0] = ((float4*)(placeholder1 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 1) * 4 + 37632))[0];
((float4*)(placeholder_shared + (int)threadIdx.x * 4 + 512))[0] = ((float4*)(placeholder1 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 1) * 4 + 50176))[0];
((float4*)(placeholder_shared + (int)threadIdx.x * 4 + 640))[0] = ((float4*)(placeholder1 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 1) * 4 + 62720))[0];
((float4*)(placeholder_shared + (int)threadIdx.x * 4 + 768))[0] = ((float4*)(placeholder1 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 1) * 4 + 75264))[0];
((float4*)(placeholder_shared + (int)threadIdx.x * 4 + 896))[0] = ((float4*)(placeholder1 + ((int)threadIdx.x >> 1) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 1) * 4 + 87808))[0];
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 8; ++k_outer_inner) {
int index2 = k_outer_inner;
int index3 = (int)threadIdx.x * 16 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 512]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 8]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 520]);
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index4 = (int)threadIdx.x * 2 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[index5 + 2] + placeholder2[index4 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmprdwasgvk | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<65>;
.reg .b32 %r<163>;
.reg .b64 %rd<46>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[56];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[1792];
ld.param.u64 %rd22, [default_function_kernel0_param_0];
ld.param.u64 %rd23, [default_function_kernel0_param_1];
ld.param.u64 %rd24, [default_function_kernel0_param_2];
ld.param.u64 %rd21, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd45, %rd23;
cvta.to.global.u64 %rd2, %rd24;
mov.u32 %r1, %tid.x;
shl.b32 %r9, %r1, 2;
mov.u32 %r10, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r10, %r9;
mul.hi.s32 %r11, %r1, -1840700269;
add.s32 %r12, %r11, %r1;
shr.u32 %r13, %r12, 31;
shr.s32 %r14, %r12, 3;
add.s32 %r15, %r14, %r13;
mul.lo.s32 %r16, %r15, 14;
sub.s32 %r17, %r1, %r16;
mov.u32 %r18, _ZZ24default_function_kernel0E18placeholder_shared;
for.size.6 i in range(0, 2, 1):
add.s32 %r(3+i*21), %r(18+i*5), %r(9+i*13);
add.s32 %r(19+i*6), %r1, (32+i*-28);
mul.hi.s32 %r(20+i*6), %r(19+i*6), -1840700269;
add.s32 %r(21+i*6), %r(20+i*6), %r(19+i*6);
shr.u32 %r(22+i*6), %r(21+i*6), 31;
shr.s32 %r(23+i*6), %r(21+i*6), 3;
for.size.14 i in range(0, 2, 1):
add.s32 %r(30+i*14), %r(29+i*14), %r(28+i*14);
mul.lo.s32 %r(31+i*14), %r(30+i*14), 14;
sub.s32 %r(32+i*14), %r(25+i*14), %r(31+i*14);
add.s32 %r(33+i*14), %r1, (64+i*32);
mul.hi.s32 %r(34+i*14), %r(33+i*14), -1840700269;
add.s32 %r(35+i*14), %r(34+i*14), %r(33+i*14);
shr.u32 %r(36+i*14), %r(35+i*14), 31;
shr.s32 %r(37+i*14), %r(35+i*14), 3;
add.s32 %r(38+i*14), %r(37+i*14), %r(36+i*14);
add.s32 %r(39+i*14), %r1, (8+i*4);
mul.hi.s32 %r(40+i*14), %r(39+i*14), -1840700269;
add.s32 %r(41+i*14), %r(40+i*14), %r(39+i*14);
shr.u32 %r(42+i*14), %r(41+i*14), 31;
shr.s32 %r(43+i*14), %r(41+i*14), 3;
for.size.14 i in range(0, 3, 1):
add.s32 %r(58+i*14), %r(57+i*14), %r(56+i*14);
mul.lo.s32 %r(59+i*14), %r(58+i*14), 14;
sub.s32 %r(60+i*14), %r(53+i*14), %r(59+i*14);
add.s32 %r(61+i*14), %r1, (128+i*32);
mul.hi.s32 %r(62+i*14), %r(61+i*14), -1840700269;
add.s32 %r(63+i*14), %r(62+i*14), %r(61+i*14);
shr.u32 %r(64+i*14), %r(63+i*14), 31;
shr.s32 %r(65+i*14), %r(63+i*14), 3;
add.s32 %r(66+i*14), %r(65+i*14), %r(64+i*14);
add.s32 %r(67+i*14), %r1, (2+i*4);
mul.hi.s32 %r(68+i*14), %r(67+i*14), -1840700269;
add.s32 %r(69+i*14), %r(68+i*14), %r(67+i*14);
shr.u32 %r(70+i*14), %r(69+i*14), 31;
shr.s32 %r(71+i*14), %r(69+i*14), 3;
add.s32 %r100, %r99, %r98;
mul.lo.s32 %r101, %r100, 14;
sub.s32 %r102, %r95, %r101;
for.size.6 i in range(0, 6, 1):
add.s32 %r(103+i*6), %r1, (256+i*32);
mul.hi.s32 %r(104+i*6), %r(103+i*6), -1840700269;
add.s32 %r(105+i*6), %r(104+i*6), %r(103+i*6);
shr.u32 %r(106+i*6), %r(105+i*6), 31;
shr.s32 %r(107+i*6), %r(105+i*6), 3;
add.s32 %r(108+i*6), %r(107+i*6), %r(106+i*6);
mad.lo.s32 %r4, %r1, 56, %r18;
mov.u32 %r5, %ctaid.x;
mul.lo.s32 %r139, %r5, 25088;
for.size.3 i in range(0, 6, 1):
add.s32 %r(140+i*2), %r(102+i*-14), %r139;
mad.lo.s32 %r(141+i*2), %r(138+i*-6), 784, %r(140+i*2);
mul.wide.s32 %rd(3+i*1), %r(141+i*2), 4;
for.size.2 i in range(0, 6, 1):
mad.lo.s32 %r(152+i*1), %r(94+i*-14), 784, %r(140+i*2);
mul.wide.s32 %rd(9+i*1), %r(152+i*1), 4;
add.s32 %r158, %r17, %r139;
mad.lo.s32 %r159, %r15, 784, %r158;
mul.wide.s32 %rd15, %r159, 4;
cvta.to.global.u64 %rd25, %rd22;
mul.wide.s32 %rd26, %r1, 4;
add.s64 %rd44, %rd25, %rd26;
mov.f32 %f64, 0f00000000;
mov.u32 %r162, 0;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 13;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f4, [%rd44];
st.shared.f32 [%r2], %f4;
$L__BB0_3:
for.size.3 i in range(0, 7, 1):
add.s64 %rd(27+i*1), %rd45, %rd(15+i*-1);
ld.global.nc.f32 %f(5+i*1), [%rd(27+i*1)+0];
st.shared.f32 [%r3+(0+i*128)], %f(5+i*1);
ld.global.nc.f32 %f12, [%rd27+50176];
for.size.3 i in range(0, 6, 1):
st.shared.f32 [%r3+(896+i*128)], %f(12+i*1);
add.s64 %rd(34+i*1), %rd45, %rd(8+i*-1);
ld.global.nc.f32 %f(13+i*1), [%rd(34+i*1)+0];
st.shared.f32 [%r3+1664], %f18;
bar.sync 0;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(19+i*3), [%r4+(0+i*4)];
ld.shared.f32 %f(20+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(21+i*3), %f(20+i*3), %f(19+i*3), %f(64+i*-43);
for.size.3 i in range(0, 11, 1):
ld.shared.f32 %f(25+i*3), [%r4+(8+i*4)];
ld.shared.f32 %f(26+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(27+i*3), %f(26+i*3), %f(25+i*3), %f(24+i*3);
ld.shared.f32 %f58, [%r4+52];
ld.shared.f32 %f59, [_ZZ24default_function_kernel0E20placeholder_d_shared+52];
fma.rn.f32 %f64, %f59, %f58, %f57;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(45+i*-1), %rd(45+i*-1), 56;
add.s32 %r162, %r162, 1;
setp.ne.s32 %p2, %r162, 56;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd40, %rd21;
shl.b32 %r160, %r5, 5;
add.s32 %r161, %r160, %r1;
mul.wide.s32 %rd41, %r161, 4;
add.s64 %rd42, %rd40, %rd41;
ld.global.nc.f32 %f60, [%rd42];
add.f32 %f61, %f64, %f60;
mov.f32 %f62, 0f00000000;
max.f32 %f63, %f61, %f62;
add.s64 %rd43, %rd2, %rd41;
st.global.f32 [%rd43], %f63;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[14];
__shared__ float placeholder_shared[448];
T_dense[0] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 56; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 14 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index2 = (int)blockIdx.x * 25088 + (int)threadIdx.x / 14 * 784 + k_outer_outer * 14 + (int)threadIdx.x % 14;
placeholder_shared[index0] = placeholder1[index2];
int index3 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 32) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 4) % 14;
placeholder_shared[index0 + 32] = placeholder1[index3];
int index4 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 64) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 8) % 14;
placeholder_shared[index0 + 64] = placeholder1[index4];
int index5 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 96) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 12) % 14;
placeholder_shared[index0 + 96] = placeholder1[index5];
int index6 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 128) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 2) % 14;
placeholder_shared[index0 + 128] = placeholder1[index6];
int index7 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 160) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 6) % 14;
placeholder_shared[index0 + 160] = placeholder1[index7];
int index8 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 192) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 10) % 14;
placeholder_shared[index0 + 192] = placeholder1[index8];
placeholder_shared[index0 + 224] = placeholder1[index2 + 12544];
int index9 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 256) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 4) % 14;
placeholder_shared[index0 + 256] = placeholder1[index9];
int index10 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 288) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 8) % 14;
placeholder_shared[index0 + 288] = placeholder1[index10];
int index11 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 320) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 12) % 14;
placeholder_shared[index0 + 320] = placeholder1[index11];
int index12 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 352) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 2) % 14;
placeholder_shared[index0 + 352] = placeholder1[index12];
int index13 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 384) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 6) % 14;
placeholder_shared[index0 + 384] = placeholder1[index13];
int index14 = (int)blockIdx.x * 25088 + ((int)threadIdx.x + 416) / 14 * 784 + k_outer_outer * 14 + ((int)threadIdx.x + 10) % 14;
placeholder_shared[index0 + 416] = placeholder1[index14];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 14];
int index15 = (int)threadIdx.x * 14 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 14 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index16];
int index17 = (int)threadIdx.x * 14 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index17];
int index18 = (int)threadIdx.x * 14 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index18];
int index19 = (int)threadIdx.x * 14 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index19];
int index20 = (int)threadIdx.x * 14 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index20];
int index21 = (int)threadIdx.x * 14 + 7;
T_dense[0] = T_dense[0] + placeholder_d_shared[7] * placeholder_shared[index21];
int index22 = (int)threadIdx.x * 14 + 8;
T_dense[0] = T_dense[0] + placeholder_d_shared[8] * placeholder_shared[index22];
int index23 = (int)threadIdx.x * 14 + 9;
T_dense[0] = T_dense[0] + placeholder_d_shared[9] * placeholder_shared[index23];
int index24 = (int)threadIdx.x * 14 + 10;
T_dense[0] = T_dense[0] + placeholder_d_shared[10] * placeholder_shared[index24];
int index25 = (int)threadIdx.x * 14 + 11;
T_dense[0] = T_dense[0] + placeholder_d_shared[11] * placeholder_shared[index25];
int index26 = (int)threadIdx.x * 14 + 12;
T_dense[0] = T_dense[0] + placeholder_d_shared[12] * placeholder_shared[index26];
int index27 = (int)threadIdx.x * 14 + 13;
T_dense[0] = T_dense[0] + placeholder_d_shared[13] * placeholder_shared[index27];
}
int index28 = (int)blockIdx.x * 32 + (int)threadIdx.x;
T_relu[index28] = max(T_dense[0] + placeholder2[index28], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpoqvhlkw8 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<292>;
.reg .b32 %r<15>;
.reg .b64 %rd<22>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[4];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[512];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd12, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd10;
cvta.to.global.u64 %rd1, %rd11;
cvta.to.global.u64 %rd2, %rd12;
mov.u32 %r1, %tid.x;
shl.b32 %r9, %r1, 2;
mov.u32 %r10, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r10, %r9;
shl.b32 %r3, %r1, 5;
mov.u32 %r11, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r4, %r11, %r9;
shl.b32 %r12, %r1, 7;
add.s32 %r5, %r11, %r12;
mul.lo.s32 %r13, %r1, 784;
cvta.to.global.u64 %rd14, %rd9;
mul.wide.s32 %rd15, %r13, 4;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(21+i*-5), %rd(14+i*-1), %rd15;
add.s64 %rd20, %rd16, 200704;
mov.u32 %r14, 0;
mov.f32 %f260, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(261+i*1), %f260;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f97, [%rd21];
st.shared.f32 [%r2], %f97;
$L__BB0_3:
for.size.2 i in range(0, 32, 1):
ld.global.nc.f32 %f(98+i*1), [%rd20+(-200704+i*12544)];
st.shared.f32 [%r4+(0+i*16)], %f(98+i*1);
bar.sync 0;
ld.shared.f32 %f130, [%r5];
ld.shared.f32 %f131, [_ZZ24default_function_kernel0E20placeholder_d_shared];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(291+i*-1), %f131, %f(130+i*2), %f(291+i*-1);
ld.shared.f32 %f(132+i*1), [%r5+(4+i*4)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(289+i*-1), %f131, %f(133+i*1), %f(289+i*-1);
ld.shared.f32 %f(134+i*1), [%r5+(12+i*4)];
fma.rn.f32 %f260, %f131, %f162, %f260;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(21+i*-1), %rd(21+i*-1), 4;
add.s32 %r14, %r14, 1;
setp.ne.s32 %p2, %r14, 784;
@%p2 bra $L__BB0_1;
mul.wide.s32 %rd17, %r3, 4;
add.s64 %rd18, %rd2, %rd17;
ld.global.nc.f32 %f163, [%rd18];
add.f32 %f164, %f291, %f163;
mov.f32 %f165, 0f00000000;
max.f32 %f166, %f164, %f165;
add.s64 %rd19, %rd1, %rd17;
for.size.4 i in range(0, 31, 1):
st.global.f32 [%rd19+(0+i*4)], %f(166+i*3);
ld.global.nc.f32 %f(167+i*3), [%rd18+(4+i*4)];
add.f32 %f(168+i*3), %f(290+i*-1), %f(167+i*3);
max.f32 %f(169+i*3), %f(168+i*3), %f165;
st.global.f32 [%rd19+124], %f259;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[128];
for (int j_inner_init = 0; j_inner_init < 32; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
int index1 = (int)threadIdx.x;
int index2 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 32; ++ax0_ax1_fused_outer_outer) {
int index3 = ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x;
int index4 = ax0_ax1_fused_outer_outer * 3136 + (int)threadIdx.x * 784 + k_outer_outer;
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int j_inner = 0; j_inner < 32; ++j_inner) {
int index5 = j_inner;
int index6 = (int)threadIdx.x * 32 + j_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[0] * placeholder_shared[index6];
}
}
for (int ax1_inner = 0; ax1_inner < 32; ++ax1_inner) {
int index7 = (int)threadIdx.x * 32 + ax1_inner;
int index8 = ax1_inner;
T_relu[index7] = max(T_dense[index8] + placeholder2[index7], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpu_fi60q4 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<65>;
.reg .b32 %r<68>;
.reg .b64 %rd<18>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[196];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[784];
ld.param.u64 %rd3, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd5, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
mov.u32 %r1, %tid.x;
mov.u32 %r21, %ctaid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(2+i*1), %r(21+i*-20), (3136+i*-3087);
add.s32 %r4, %r1, 2;
shl.b32 %r22, %r1, 2;
mov.u32 %r23, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r23, %r22;
mov.f32 %f61, 0f00000000;
mov.u32 %r63, 0;
cvta.to.global.u64 %rd7, %rd3;
mov.f32 %f62, %f61;
$L__BB0_1:
setp.gt.s32 %p1, %r1, 0;
bar.sync 0;
mul.lo.s32 %r7, %r63, 49;
add.s32 %r26, %r7, %r22;
mul.wide.s32 %rd8, %r26, 4;
add.s64 %rd2, %rd7, %rd8;
ld.global.nc.f32 %f11, [%rd2];
shl.b32 %r27, %r1, 4;
mov.u32 %r28, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r29, %r28, %r27;
for.size.2 i in range(0, 3, 1):
st.shared.f32 [%r29+(0+i*4)], %f(11+i*1);
ld.global.nc.f32 %f(12+i*1), [%rd2+(4+i*4)];
for.size.6 i in range(0, 5, 1):
for.size.2 j in range(0, 2, 1):
st.shared.f32 [%r29+(12+i*32+j*20)], %f(14+i*4+j*1);
ld.global.nc.f32 %f(15+i*4+j*1), [%rd2+(32+i*32+j*4)];
for.size.2 j in range(0, 2, 1):
st.shared.f32 [%r29+(36+i*32+j*4)], %f(16+i*4+j*1);
ld.global.nc.f32 %f(17+i*4+j*1), [%rd2+(40+i*32+j*4)];
st.shared.f32 [%r29+172], %f34;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f35, [%rd2+192];
st.shared.f32 [%r29+192], %f35;
$L__BB0_3:
add.s32 %r8, %r7, %r2;
mov.u32 %r66, 8;
mov.u32 %r64, 0;
mov.u32 %r65, %r4;
$L__BB0_4:
add.s32 %r36, %r65, -2;
mul.hi.s32 %r37, %r36, 1402438301;
shr.u32 %r38, %r37, 31;
shr.s32 %r39, %r37, 4;
add.s32 %r40, %r39, %r38;
mul.lo.s32 %r41, %r40, 49;
sub.s32 %r42, %r36, %r41;
add.s32 %r43, %r8, %r42;
mad.lo.s32 %r44, %r40, 784, %r43;
mul.wide.s32 %rd9, %r44, 4;
add.s64 %rd10, %rd1, %rd9;
ld.global.nc.f32 %f36, [%rd10];
add.s32 %r45, %r5, %r64;
st.shared.f32 [%r45], %f36;
mul.hi.s32 %r46, %r65, 1402438301;
shr.u32 %r47, %r46, 31;
shr.s32 %r48, %r46, 4;
add.s32 %r49, %r48, %r47;
mul.lo.s32 %r50, %r49, 49;
sub.s32 %r51, %r65, %r50;
add.s32 %r52, %r8, %r51;
mad.lo.s32 %r53, %r49, 784, %r52;
mul.wide.s32 %rd11, %r53, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f37, [%rd12];
st.shared.f32 [%r45+8], %f37;
for.size.1 i in range(0, 2, 1):
add.s32 %r(65+i*-1), %r(65+i*-1), (4+i*12);
add.s32 %r66, %r66, 16;
setp.ne.s32 %p2, %r66, 792;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r67, 0;
bra.uni $L__BB0_6;
$L__BB0_7:
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(41+i*1), [%r(16+i*1)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(43+i*2), %f42, %f(41+i*3), %f(61+i*1);
ld.shared.f32 %f(44+i*2), [%r16+(396+i*-388)];
ld.shared.f32 %f47, [%r17+8];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(48+i*2), %f47, %f(46+i*3), %f(43+i*2);
ld.shared.f32 %f(49+i*2), [%r16+(400+i*-388)];
ld.shared.f32 %f52, [%r17+12];
fma.rn.f32 %f61, %f52, %f51, %f48;
ld.shared.f32 %f53, [%r16+404];
fma.rn.f32 %f62, %f52, %f53, %f50;
add.s32 %r67, %r67, 4;
$L__BB0_6:
add.s32 %r55, %r67, %r3;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(56+i*2), %r(55+i*12), 2;
add.s32 %r(16+i*1), %r(23+i*5), %r(56+i*2);
ld.shared.f32 %f(38+i*1), [%r(16+i*1)+0];
fma.rn.f32 %f61, %f39, %f38, %f61;
ld.shared.f32 %f40, [%r16+392];
fma.rn.f32 %f62, %f39, %f40, %f62;
setp.eq.s32 %p3, %r67, 48;
@%p3 bra $L__BB0_8;
bra.uni $L__BB0_7;
$L__BB0_8:
add.s32 %r63, %r63, 1;
setp.lt.u32 %p4, %r63, 16;
@%p4 bra $L__BB0_1;
shl.b32 %r61, %r21, 2;
add.s32 %r62, %r61, %r1;
cvta.to.global.u64 %rd13, %rd5;
mul.wide.s32 %rd14, %r62, 4;
add.s64 %rd15, %rd13, %rd14;
ld.global.nc.f32 %f54, [%rd15];
add.f32 %f55, %f61, %f54;
mov.f32 %f56, 0f00000000;
max.f32 %f57, %f55, %f56;
cvta.to.global.u64 %rd16, %rd4;
add.s64 %rd17, %rd16, %rd14;
st.global.f32 [%rd17], %f57;
ld.global.nc.f32 %f58, [%rd15+8];
add.f32 %f59, %f62, %f58;
max.f32 %f60, %f59, %f56;
st.global.f32 [%rd17+8], %f60;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[49];
__shared__ float placeholder_shared[196];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 16; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x * 4;
int index1 = k_outer_outer * 49 + (int)threadIdx.x * 4;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
placeholder_d_shared[index0 + 2] = placeholder[index1 + 2];
placeholder_d_shared[index0 + 3] = placeholder[index1 + 3];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 9] = placeholder[index1 + 9];
placeholder_d_shared[index0 + 10] = placeholder[index1 + 10];
placeholder_d_shared[index0 + 11] = placeholder[index1 + 11];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 17] = placeholder[index1 + 17];
placeholder_d_shared[index0 + 18] = placeholder[index1 + 18];
placeholder_d_shared[index0 + 19] = placeholder[index1 + 19];
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
placeholder_d_shared[index0 + 25] = placeholder[index1 + 25];
placeholder_d_shared[index0 + 26] = placeholder[index1 + 26];
placeholder_d_shared[index0 + 27] = placeholder[index1 + 27];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 33] = placeholder[index1 + 33];
placeholder_d_shared[index0 + 34] = placeholder[index1 + 34];
placeholder_d_shared[index0 + 35] = placeholder[index1 + 35];
placeholder_d_shared[index0 + 40] = placeholder[index1 + 40];
placeholder_d_shared[index0 + 41] = placeholder[index1 + 41];
placeholder_d_shared[index0 + 42] = placeholder[index1 + 42];
placeholder_d_shared[index0 + 43] = placeholder[index1 + 43];
if ((int)threadIdx.x < 1) {
placeholder_d_shared[index0 + 48] = placeholder[index1 + 48];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 3136 + (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x) / 49 * 784 + k_outer_outer * 49 + (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x) % 49;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_inner = 0; k_inner < 49; ++k_inner) {
int index4 = k_inner;
int index5 = (int)threadIdx.x * 49 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 98]);
}
}
int index6 = (int)blockIdx.x * 4 + (int)threadIdx.x;
T_relu[index6] = max(T_dense[0] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 2] = max(T_dense[1] + placeholder2[index6 + 2], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp7k5e04rj | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<148>;
.reg .b32 %r<141>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[100352];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(2+i*17), %r1, (1+i*2);
mov.u32 %r20, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r20, %r19;
mul.lo.s32 %r4, %r1, 196;
add.s32 %r5, %r2, 224;
mov.u32 %r21, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r6, %r21, %r19;
mov.f32 %f132, 0f00000000;
mov.u32 %r136, 0;
for.size.1 i in range(0, 7, 1):
mov.f32 %f(133+i*1), %f132;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r136, 196;
add.s32 %r22, %r8, %r2;
mul.wide.s32 %rd10, %r22, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.3 i in range(0, 6, 1):
for.size.2 j in range(0, 2, 1):
ld.global.nc.f32 %f(33+i*2+j*1), [%rd5+(0+i*128+j*4)];
st.shared.f32 [%r3+(0+i*128+j*4)], %f(33+i*2+j*1);
setp.gt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_3;
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(45+i*1), [%rd5+(768+i*4)];
st.shared.f32 [%r3+(768+i*4)], %f(45+i*1);
$L__BB0_3:
mov.u32 %r139, 512;
mov.u32 %r137, %r2;
mov.u32 %r138, %r5;
$L__BB0_4:
mul.hi.s32 %r24, %r137, 1402438301;
shr.u32 %r25, %r24, 31;
shr.s32 %r26, %r24, 6;
add.s32 %r27, %r26, %r25;
mul.lo.s32 %r28, %r27, 196;
sub.s32 %r29, %r137, %r28;
add.s32 %r30, %r29, %r8;
mad.lo.s32 %r31, %r27, 784, %r30;
mul.wide.s32 %rd11, %r31, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.v2.u32 {%r32, %r33}, [%rd12];
add.s32 %r36, %r6, %r139;
for.size.13 i in range(0, 2, 1):
st.shared.v2.u32 [%r36+(-512+i*128)], {%r(32+i*14), %r(33+i*14)};
add.s32 %r(37+i*13), %r137, (32+i*32);
mul.hi.s32 %r(38+i*13), %r(37+i*13), 1402438301;
shr.u32 %r(39+i*13), %r(38+i*13), 31;
shr.s32 %r(40+i*13), %r(38+i*13), 6;
add.s32 %r(41+i*13), %r(40+i*13), %r(39+i*13);
mul.lo.s32 %r(42+i*13), %r(41+i*13), 196;
sub.s32 %r(43+i*13), %r(37+i*13), %r(42+i*13);
add.s32 %r(44+i*13), %r(43+i*13), %r8;
mad.lo.s32 %r(45+i*13), %r(41+i*13), 784, %r(44+i*13);
mul.wide.s32 %rd(13+i*2), %r(45+i*13), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.v2.u32 {%r(46+i*13), %r(47+i*13)}, [%rd(14+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.v2.u32 [%r36+(-256+i*128)], {%r(59+i*13), %r(60+i*13)};
add.s32 %r(63+i*13), %r137, (96+i*32);
mul.hi.s32 %r(64+i*13), %r(63+i*13), 1402438301;
shr.u32 %r(65+i*13), %r(64+i*13), 31;
shr.s32 %r(66+i*13), %r(64+i*13), 6;
add.s32 %r(67+i*13), %r(66+i*13), %r(65+i*13);
mul.lo.s32 %r(68+i*13), %r(67+i*13), 196;
sub.s32 %r(69+i*13), %r(63+i*13), %r(68+i*13);
add.s32 %r(70+i*13), %r(69+i*13), %r8;
mad.lo.s32 %r(71+i*13), %r(67+i*13), 784, %r(70+i*13);
mul.wide.s32 %rd(17+i*2), %r(71+i*13), 4;
add.s64 %rd(18+i*2), %rd1, %rd(17+i*2);
ld.global.nc.v2.u32 {%r(72+i*13), %r(73+i*13)}, [%rd(18+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.v2.u32 [%r36+(0+i*128)], {%r(85+i*13), %r(86+i*13)};
add.s32 %r(89+i*13), %r138, (-64+i*32);
mul.hi.s32 %r(90+i*13), %r(89+i*13), 1402438301;
shr.u32 %r(91+i*13), %r(90+i*13), 31;
shr.s32 %r(92+i*13), %r(90+i*13), 6;
add.s32 %r(93+i*13), %r(92+i*13), %r(91+i*13);
mul.lo.s32 %r(94+i*13), %r(93+i*13), 196;
sub.s32 %r(95+i*13), %r(89+i*13), %r(94+i*13);
add.s32 %r(96+i*13), %r(95+i*13), %r8;
mad.lo.s32 %r(97+i*13), %r(93+i*13), 784, %r(96+i*13);
mul.wide.s32 %rd(21+i*2), %r(97+i*13), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.v2.u32 {%r(98+i*13), %r(99+i*13)}, [%rd(22+i*2)+0];
st.shared.v2.u32 [%r36+256], {%r111, %r112};
mul.hi.s32 %r115, %r138, 1402438301;
shr.u32 %r116, %r115, 31;
shr.s32 %r117, %r115, 6;
add.s32 %r118, %r117, %r116;
mul.lo.s32 %r119, %r118, 196;
sub.s32 %r120, %r138, %r119;
add.s32 %r121, %r120, %r8;
mad.lo.s32 %r122, %r118, 784, %r121;
mul.wide.s32 %rd25, %r122, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.v2.u32 {%r123, %r124}, [%rd26];
st.shared.v2.u32 [%r36+384], {%r123, %r124};
add.s32 %r139, %r139, 1024;
setp.ne.s32 %p2, %r139, 100864;
for.size.1 i in range(0, 2, 1):
add.s32 %r(13+i*124), %r138, (256+i*-224);
mov.u32 %r138, %r13;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r140, 0;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(128+i*1), %r140, (2+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(131+i*1), %r(20+i*108), %r(129+i*-125);
shl.b32 %r133, %r132, 2;
add.s32 %r135, %r21, %r133;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(47+i*1), [%r(135+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(49+i*2), %f48, %f(47+i*3), %f(132+i*1);
ld.shared.f32 %f(50+i*2), [%r135+(12544+i*12544)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(53+i*2), %f48, %f(52+i*2), %f(134+i*1);
ld.shared.f32 %f(54+i*2), [%r135+(37632+i*12544)];
for.size.17 i in range(0, 2, 1):
fma.rn.f32 %f(63+i*17), %f(48+i*17), %f(62+i*17), %f(139+i*-76);
ld.shared.f32 %f(64+i*17), [%r135+(4+i*4)];
ld.shared.f32 %f(65+i*17), [%r131+(4+i*4)];
fma.rn.f32 %f(66+i*17), %f(65+i*17), %f(64+i*17), %f(49+i*17);
ld.shared.f32 %f(67+i*17), [%r135+(12548+i*4)];
fma.rn.f32 %f(68+i*17), %f(65+i*17), %f(67+i*17), %f(51+i*17);
ld.shared.f32 %f(69+i*17), [%r135+(25092+i*4)];
fma.rn.f32 %f(70+i*17), %f(65+i*17), %f(69+i*17), %f(53+i*17);
ld.shared.f32 %f(71+i*17), [%r135+(37636+i*4)];
fma.rn.f32 %f(72+i*17), %f(65+i*17), %f(71+i*17), %f(55+i*17);
ld.shared.f32 %f(73+i*17), [%r135+(50180+i*4)];
fma.rn.f32 %f(74+i*17), %f(65+i*17), %f(73+i*17), %f(57+i*17);
ld.shared.f32 %f(75+i*17), [%r135+(62724+i*4)];
fma.rn.f32 %f(76+i*17), %f(65+i*17), %f(75+i*17), %f(59+i*17);
ld.shared.f32 %f(77+i*17), [%r135+(75268+i*4)];
fma.rn.f32 %f(78+i*17), %f(65+i*17), %f(77+i*17), %f(61+i*17);
ld.shared.f32 %f(79+i*17), [%r135+(87812+i*4)];
fma.rn.f32 %f97, %f82, %f96, %f80;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(98+i*1), [%r(135+i*-4)+12];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(132+i*1), %f99, %f(98+i*2), %f(83+i*2);
ld.shared.f32 %f(100+i*1), [%r135+(12556+i*12544)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(134+i*1), %f99, %f(101+i*1), %f(87+i*2);
ld.shared.f32 %f(102+i*1), [%r135+(37644+i*12544)];
fma.rn.f32 %f139, %f99, %f106, %f97;
add.s32 %r140, %r140, 1;
setp.ne.s32 %p3, %r140, 49;
@%p3 bra $L__BB0_6;
add.s32 %r136, %r136, 1;
setp.lt.u32 %p4, %r136, 4;
@%p4 bra $L__BB0_1;
mul.wide.s32 %rd27, %r1, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f107, [%rd28];
add.f32 %f108, %f132, %f107;
mov.f32 %f109, 0f00000000;
max.f32 %f110, %f108, %f109;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd29+(0+i*64)], %f(110+i*3);
ld.global.nc.f32 %f(111+i*3), [%rd28+(64+i*64)];
add.f32 %f(112+i*3), %f(133+i*1), %f(111+i*3);
max.f32 %f(113+i*3), %f(112+i*3), %f109;
st.global.f32 [%rd29+448], %f131;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[8];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[25088];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x * 2;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 2;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 33] = placeholder[index1 + 33];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 65] = placeholder[index1 + 65];
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
placeholder_d_shared[index0 + 97] = placeholder[index1 + 97];
placeholder_d_shared[index0 + 128] = placeholder[index1 + 128];
placeholder_d_shared[index0 + 129] = placeholder[index1 + 129];
placeholder_d_shared[index0 + 160] = placeholder[index1 + 160];
placeholder_d_shared[index0 + 161] = placeholder[index1 + 161];
if ((int)threadIdx.x < 2) {
placeholder_d_shared[index0 + 192] = placeholder[index1 + 192];
}
if ((int)threadIdx.x < 2) {
placeholder_d_shared[index0 + 193] = placeholder[index1 + 193];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 784; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder1 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 2) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 2) % 196))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 49; ++k_outer_inner) {
int index2 = k_outer_inner * 4;
int index3 = (int)threadIdx.x * 196 + k_outer_inner * 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 3136]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 6272]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 9408]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 12544]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 15680]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 18816]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 21952]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 3137]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 6273]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 9409]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 12545]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 15681]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 18817]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 21953]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 2]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 3138]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 6274]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 9410]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 12546]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 15682]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 18818]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2 + 2] * placeholder_shared[index3 + 21954]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 3]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 3139]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 6275]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 9411]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 12547]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 15683]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 18819]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2 + 3] * placeholder_shared[index3 + 21955]);
}
}
int index4 = (int)threadIdx.x;
T_relu[index4] = max(T_dense[0] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 16] = max(T_dense[1] + placeholder2[index4 + 16], 0.000000e+00f);
T_relu[index4 + 32] = max(T_dense[2] + placeholder2[index4 + 32], 0.000000e+00f);
T_relu[index4 + 48] = max(T_dense[3] + placeholder2[index4 + 48], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[4] + placeholder2[index4 + 64], 0.000000e+00f);
T_relu[index4 + 80] = max(T_dense[5] + placeholder2[index4 + 80], 0.000000e+00f);
T_relu[index4 + 96] = max(T_dense[6] + placeholder2[index4 + 96], 0.000000e+00f);
T_relu[index4 + 112] = max(T_dense[7] + placeholder2[index4 + 112], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpt07dw9ff | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<26>;
.reg .f32 %f<345>;
.reg .b32 %r<145>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[196];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[25088];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd8;
cvta.to.global.u64 %rd3, %rd7;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
shl.b32 %r2, %r1, 1;
mul.lo.s32 %r3, %r1, 1568;
shl.b32 %r22, %r1, 3;
mov.u32 %r23, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r4, %r23, %r22;
add.s32 %r5, %r2, 25;
mov.u32 %r24, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r25, %r24, %r22;
add.s32 %r6, %r25, 64;
mov.u32 %r139, 0;
mov.f32 %f281, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(282+i*1), %f281;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r139, 49;
add.s32 %r26, %r8, %r2;
mul.wide.s32 %rd10, %r26, 4;
add.s64 %rd5, %rd1, %rd10;
setp.gt.s32 %p1, %r1, 24;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f129, [%rd5];
st.shared.f32 [%r4], %f129;
$L__BB0_3:
setp.gt.s32 %p2, %r1, 23;
@%p2 bra $L__BB0_5;
ld.global.nc.f32 %f130, [%rd5+4];
st.shared.f32 [%r4+4], %f130;
$L__BB0_5:
setp.gt.s32 %p3, %r1, 20;
@%p3 bra $L__BB0_7;
ld.global.nc.f32 %f131, [%rd5+32];
st.shared.f32 [%r4+32], %f131;
$L__BB0_7:
setp.gt.s32 %p4, %r1, 19;
@%p4 bra $L__BB0_9;
ld.global.nc.f32 %f132, [%rd5+36];
st.shared.f32 [%r4+36], %f132;
$L__BB0_9:
setp.gt.s32 %p5, %r1, 16;
@%p5 bra $L__BB0_11;
ld.global.nc.f32 %f133, [%rd5+64];
st.shared.f32 [%r4+64], %f133;
$L__BB0_11:
setp.gt.s32 %p6, %r1, 15;
@%p6 bra $L__BB0_13;
ld.global.nc.f32 %f134, [%rd5+68];
st.shared.f32 [%r4+68], %f134;
$L__BB0_13:
setp.gt.s32 %p7, %r1, 12;
@%p7 bra $L__BB0_15;
ld.global.nc.f32 %f135, [%rd5+96];
st.shared.f32 [%r4+96], %f135;
$L__BB0_15:
setp.gt.s32 %p8, %r1, 11;
@%p8 bra $L__BB0_17;
ld.global.nc.f32 %f136, [%rd5+100];
st.shared.f32 [%r4+100], %f136;
$L__BB0_17:
setp.gt.s32 %p9, %r1, 8;
@%p9 bra $L__BB0_19;
ld.global.nc.f32 %f137, [%rd5+128];
st.shared.f32 [%r4+128], %f137;
$L__BB0_19:
setp.gt.s32 %p10, %r1, 7;
@%p10 bra $L__BB0_21;
ld.global.nc.f32 %f138, [%rd5+132];
st.shared.f32 [%r4+132], %f138;
$L__BB0_21:
setp.gt.s32 %p11, %r1, 4;
@%p11 bra $L__BB0_23;
ld.global.nc.f32 %f139, [%rd5+160];
st.shared.f32 [%r4+160], %f139;
$L__BB0_23:
setp.gt.s32 %p12, %r1, 3;
@%p12 bra $L__BB0_25;
ld.global.nc.f32 %f140, [%rd5+164];
st.shared.f32 [%r4+164], %f140;
$L__BB0_25:
setp.gt.s32 %p13, %r1, 0;
@%p13 bra $L__BB0_27;
ld.global.nc.f32 %f141, [%rd5+192];
st.shared.f32 [%r4+192], %f141;
$L__BB0_27:
setp.gt.s32 %p14, %r1, -1;
@%p14 bra $L__BB0_29;
ld.global.nc.f32 %f142, [%rd5+196];
st.shared.f32 [%r4+196], %f142;
$L__BB0_29:
mov.u32 %r143, 64;
mov.u32 %r140, %r6;
mov.u32 %r141, %r2;
mov.u32 %r142, %r5;
$L__BB0_30:
add.s32 %r28, %r141, 1;
for.size.6 i in range(0, 2, 1):
mul.hi.s32 %r(29+i*6), %r(28+i*113), 1402438301;
shr.u32 %r(30+i*6), %r(29+i*6), 31;
shr.s32 %r(31+i*6), %r(29+i*6), 4;
add.s32 %r(32+i*6), %r(31+i*6), %r(30+i*6);
mul.lo.s32 %r(33+i*6), %r(32+i*6), 49;
sub.s32 %r(34+i*6), %r(28+i*113), %r(33+i*6);
for.size.1 i in range(0, 2, 1):
setp.gt.s32 %p(15+i*1), %r(40+i*-6), -1;
for.size.2 i in range(0, 2, 1):
shr.s32 %r(41+i*2), %r(40+i*-6), 31;
add.s32 %r(42+i*2), %r(38+i*-6), %r(41+i*2);
for.size.1 i in range(0, 2, 1):
selp.b32 %r(45+i*1), 0, 49, %p(15+i*1);
for.size.1 i in range(0, 2, 1):
add.s32 %r(47+i*1), %r(40+i*7), %r(8+i*37);
for.size.35 i in range(0, 2, 1):
mad.lo.s32 %r(49+i*26), %r(42+i*26), 784, %r(48+i*26);
add.s32 %r(50+i*26), %r(34+i*32), %r8;
add.s32 %r(51+i*26), %r(50+i*26), %r(46+i*26);
mad.lo.s32 %r(52+i*26), %r(44+i*26), 784, %r(51+i*26);
mul.wide.s32 %rd(11+i*4), %r(49+i*26), 4;
add.s64 %rd(12+i*4), %rd3, %rd(11+i*4);
mul.wide.s32 %rd(13+i*4), %r(52+i*26), 4;
add.s64 %rd(14+i*4), %rd3, %rd(13+i*4);
ld.global.nc.f32 %f(143+i*2), [%rd(12+i*6)+0];
ld.global.nc.f32 %f(144+i*2), [%rd(14+i*2)+0];
st.shared.v2.f32 [%r140+(-64+i*32)], {%f(143+i*3), %f(144+i*1)};
add.s32 %r(53+i*26), %r141, (8+i*8);
mul.hi.s32 %r(54+i*26), %r(53+i*26), 1402438301;
shr.u32 %r(55+i*26), %r(54+i*26), 31;
shr.s32 %r(56+i*26), %r(54+i*26), 4;
add.s32 %r(57+i*26), %r(56+i*26), %r(55+i*26);
mul.lo.s32 %r(58+i*26), %r(57+i*26), 49;
sub.s32 %r(59+i*26), %r(53+i*26), %r(58+i*26);
add.s32 %r(60+i*26), %r(141+i*1), (9+i*-17);
mul.hi.s32 %r(61+i*26), %r(60+i*26), 1402438301;
shr.u32 %r(62+i*26), %r(61+i*26), 31;
shr.s32 %r(63+i*26), %r(61+i*26), 4;
add.s32 %r(64+i*26), %r(63+i*26), %r(62+i*26);
mul.lo.s32 %r(65+i*26), %r(64+i*26), 49;
sub.s32 %r(66+i*26), %r(60+i*26), %r(65+i*26);
setp.gt.s32 %p(17+i*2), %r(59+i*26), -1;
setp.gt.s32 %p(18+i*2), %r(66+i*26), -1;
shr.s32 %r(67+i*26), %r(59+i*26), 31;
add.s32 %r(68+i*26), %r(57+i*26), %r(67+i*26);
shr.s32 %r(69+i*26), %r(66+i*26), 31;
add.s32 %r(70+i*26), %r(64+i*26), %r(69+i*26);
selp.b32 %r(71+i*26), 0, 49, %p(17+i*2);
selp.b32 %r(72+i*26), 0, 49, %p(18+i*2);
add.s32 %r(73+i*26), %r(59+i*26), %r8;
add.s32 %r(74+i*26), %r(73+i*26), %r(71+i*26);
mad.lo.s32 %r101, %r94, 784, %r100;
for.size.1 i in range(0, 2, 1):
add.s32 %r(102+i*1), %r(92+i*10), %r(8+i*90);
mad.lo.s32 %r104, %r96, 784, %r103;
for.size.2 i in range(0, 2, 1):
mul.wide.s32 %rd(19+i*2), %r(101+i*3), 4;
add.s64 %rd(20+i*2), %rd3, %rd(19+i*2);
for.size.1 i in range(0, 2, 1):
ld.global.nc.f32 %f(147+i*1), [%rd(22+i*-2)+0];
st.shared.v2.f32 [%r140], {%f148, %f147};
add.s32 %r105, %r142, -1;
for.size.7 i in range(0, 2, 1):
mul.hi.s32 %r(106+i*6), %r(105+i*37), 1402438301;
shr.u32 %r(107+i*6), %r(106+i*6), 31;
shr.s32 %r(108+i*6), %r(106+i*6), 4;
add.s32 %r(109+i*6), %r(108+i*6), %r(107+i*6);
mul.lo.s32 %r(110+i*6), %r(109+i*6), 49;
sub.s32 %r(111+i*6), %r(105+i*37), %r(110+i*6);
setp.gt.s32 %p(21+i*1), %r(111+i*6), -1;
for.size.2 i in range(0, 2, 1):
shr.s32 %r(118+i*2), %r(111+i*6), 31;
add.s32 %r(119+i*2), %r(109+i*6), %r(118+i*2);
for.size.1 i in range(0, 2, 1):
selp.b32 %r(122+i*1), 0, 49, %p(21+i*1);
for.size.1 i in range(0, 2, 1):
add.s32 %r(124+i*1), %r(111+i*13), %r(8+i*114);
mad.lo.s32 %r126, %r119, 784, %r125;
for.size.1 i in range(0, 2, 1):
add.s32 %r(127+i*1), %r(117+i*10), %r(8+i*115);
mad.lo.s32 %r129, %r121, 784, %r128;
for.size.2 i in range(0, 2, 1):
mul.wide.s32 %rd(23+i*2), %r(126+i*3), 4;
add.s64 %rd(24+i*2), %rd3, %rd(23+i*2);
for.size.1 i in range(0, 2, 1):
ld.global.nc.f32 %f(149+i*1), [%rd(26+i*-2)+0];
st.shared.v2.f32 [%r140+32], {%f150, %f149};
for.size.1 i in range(0, 2, 1):
add.s32 %r(140+i*3), %r(140+i*3), 128;
setp.ne.s32 %p23, %r143, 25152;
for.size.1 i in range(0, 2, 1):
add.s32 %r(16+i*125), %r142, (32+i*-25);
mov.u32 %r142, %r16;
@%p23 bra $L__BB0_30;
bar.sync 0;
mov.u32 %r144, 0;
$L__BB0_32:
add.s32 %r131, %r144, %r3;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(132+i*3), %r(131+i*13), 2;
add.s32 %r(134+i*3), %r(24+i*-1), %r(132+i*3);
ld.shared.f32 %f(151+i*1), [%r(134+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(312+i*-1), %f152, %f(151+i*2), %f(312+i*-1);
ld.shared.f32 %f(153+i*1), [%r134+(196+i*196)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(310+i*-1), %f152, %f(154+i*1), %f(310+i*-1);
ld.shared.f32 %f(155+i*1), [%r134+(588+i*196)];
fma.rn.f32 %f281, %f152, %f183, %f281;
add.s32 %r144, %r144, 1;
setp.ne.s32 %p24, %r144, 49;
@%p24 bra $L__BB0_32;
add.s32 %r139, %r139, 1;
setp.lt.u32 %p25, %r139, 16;
@%p25 bra $L__BB0_1;
shl.b32 %r138, %r1, 5;
mul.wide.s32 %rd27, %r138, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f184, [%rd28];
add.f32 %f185, %f312, %f184;
mov.f32 %f186, 0f00000000;
max.f32 %f187, %f185, %f186;
add.s64 %rd29, %rd2, %rd27;
for.size.4 i in range(0, 31, 1):
st.global.f32 [%rd29+(0+i*4)], %f(187+i*3);
ld.global.nc.f32 %f(188+i*3), [%rd28+(4+i*4)];
add.f32 %f(189+i*3), %f(311+i*-1), %f(188+i*3);
max.f32 %f(190+i*3), %f(189+i*3), %f186;
st.global.f32 [%rd29+124], %f280;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[49];
__shared__ float placeholder_shared[6272];
for (int j_inner_init = 0; j_inner_init < 32; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 16; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 7; ++ax0_ax1_fused_outer_outer) {
for (int ax0_ax1_fused_inner_s = 0; ax0_ax1_fused_inner_s < 2; ++ax0_ax1_fused_inner_s) {
if (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x * 2 + ax0_ax1_fused_inner_s < 49) {
if (ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x < 25) {
int index1 = ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x * 2 + ax0_ax1_fused_inner_s;
int index2 = k_outer_outer * 49 + ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x * 2 + ax0_ax1_fused_inner_s;
placeholder_d_shared[index1] = placeholder[index2];
}
}
}
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 784; ++ax0_ax1_fused_outer_outer1) {
int2 _1;
int2 _2;
int2 _3;
int2 _4;
int2 _5 = make_int2(ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 2+1*0, ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 2+1*1);
int2 _6 = make_int2(49, 49);
_4.x = _5.x%_6.x;
_4.y = _5.y%_6.y;
int2 _7;
int2 _8 = make_int2(ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 2+1*0, ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 2+1*1);
int2 _9 = make_int2(49, 49);
_7.x = _8.x/_9.x;
_7.y = _8.y/_9.y;
int2 _10;
ushort2 _11;
ushort2 _12;
ushort2 _13;
int2 _14 = make_int2(49, 49);
int2 _15 = make_int2(0, 0);
_13.x = _14.x>=_15.x;
_13.y = _14.y>=_15.y;
ushort2 _16;
int2 _17 = make_int2(0, 0);
_16.x = _4.x>=_17.x;
_16.y = _4.y>=_17.y;
_12.x = _13.x&&_16.x;
_12.y = _13.y&&_16.y;
ushort2 _18;
ushort2 _19;
int2 _20 = make_int2(49, 49);
int2 _21 = make_int2(0, 0);
_19.x = _20.x<_21.x;
_19.y = _20.y<_21.y;
ushort2 _22;
int2 _23 = make_int2(0, 0);
_22.x = _4.x<=_23.x;
_22.y = _4.y<=_23.y;
_18.x = _19.x&&_22.x;
_18.y = _19.y&&_22.y;
_11.x = _12.x||_18.x;
_11.y = _12.y||_18.y;
int2 _24;
int2 _25 = make_int2(1, 1);
_24.x = _7.x-_25.x;
_24.y = _7.y-_25.y;
_10.x = bool(_11.x)?_7.x:_24.x;
_10.y = bool(_11.y)?_7.y:_24.y;
int2 _26 = make_int2(784, 784);
_3.x = _10.x*_26.x;
_3.y = _10.y*_26.y;
int2 _27 = make_int2(k_outer_outer * 49, k_outer_outer * 49);
_2.x = _3.x+_27.x;
_2.y = _3.y+_27.y;
int2 _28;
int2 _29 = make_int2(ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 2+1*0, ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 2+1*1);
int2 _30 = make_int2(49, 49);
_28.x = _29.x%_30.x;
_28.y = _29.y%_30.y;
int2 _31;
ushort2 _32;
ushort2 _33;
ushort2 _34;
int2 _35 = make_int2(49, 49);
int2 _36 = make_int2(0, 0);
_34.x = _35.x>=_36.x;
_34.y = _35.y>=_36.y;
ushort2 _37;
int2 _38 = make_int2(0, 0);
_37.x = _28.x>=_38.x;
_37.y = _28.y>=_38.y;
_33.x = _34.x&&_37.x;
_33.y = _34.y&&_37.y;
ushort2 _39;
ushort2 _40;
int2 _41 = make_int2(49, 49);
int2 _42 = make_int2(0, 0);
_40.x = _41.x<_42.x;
_40.y = _41.y<_42.y;
ushort2 _43;
int2 _44 = make_int2(0, 0);
_43.x = _28.x<=_44.x;
_43.y = _28.y<=_44.y;
_39.x = _40.x&&_43.x;
_39.y = _40.y&&_43.y;
_32.x = _33.x||_39.x;
_32.y = _33.y||_39.y;
int2 _45;
int2 _46 = make_int2(49, 49);
_45.x = _28.x+_46.x;
_45.y = _28.y+_46.y;
_31.x = bool(_32.x)?_28.x:_45.x;
_31.y = bool(_32.y)?_28.y:_45.y;
_1.x = _2.x+_31.x;
_1.y = _2.y+_31.y;
int index3 = _1.x;
int index4 = _1.y;
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 2))[0] = make_float2(placeholder1[index3],placeholder1[index4]);
}
__syncthreads();
for (int k_inner = 0; k_inner < 49; ++k_inner) {
for (int j_inner = 0; j_inner < 32; ++j_inner) {
int index5 = j_inner;
int index6 = k_inner;
int index7 = (int)threadIdx.x * 1568 + j_inner * 49 + k_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
for (int ax1_inner = 0; ax1_inner < 32; ++ax1_inner) {
int index8 = (int)threadIdx.x * 32 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpuylgrfj4 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<356>;
.reg .b32 %r<106>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[448];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[14336];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
shl.b32 %r13, %r1, 2;
mov.u32 %r14, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r14, %r13;
mov.u32 %r3, %ctaid.x;
mul.lo.s32 %r4, %r3, 25088;
mov.u32 %r15, _ZZ24default_function_kernel0E18placeholder_shared;
mad.lo.s32 %r5, %r1, 448, %r15;
mov.f32 %f355, 0f00000000;
mov.u32 %r104, 0;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r7, %r104, 112;
add.s32 %r16, %r7, %r1;
mul.wide.s32 %rd10, %r16, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.2 i in range(0, 3, 1):
ld.global.nc.f32 %f(4+i*1), [%rd5+(0+i*128)];
st.shared.f32 [%r2+(0+i*128)], %f(4+i*1);
setp.gt.s32 %p1, %r1, 15;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f7, [%rd5+384];
st.shared.f32 [%r2+384], %f7;
$L__BB0_3:
add.s32 %r8, %r7, %r4;
mov.u32 %r105, 0;
$L__BB0_4:
shl.b32 %r18, %r105, 5;
add.s32 %r19, %r18, %r1;
mul.hi.s32 %r20, %r19, -1840700269;
add.s32 %r21, %r20, %r19;
shr.u32 %r22, %r21, 31;
shr.s32 %r23, %r21, 6;
add.s32 %r24, %r23, %r22;
mul.lo.s32 %r25, %r24, 112;
sub.s32 %r26, %r19, %r25;
add.s32 %r27, %r8, %r26;
mad.lo.s32 %r28, %r24, 784, %r27;
mul.wide.s32 %rd11, %r28, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f8, [%rd12];
shl.b32 %r29, %r19, 2;
add.s32 %r31, %r15, %r29;
for.size.14 i in range(0, 7, 1):
st.shared.f32 [%r31+(0+i*128)], %f(8+i*1);
add.s32 %r(32+i*10), %r19, (32+i*32);
mul.hi.s32 %r(33+i*10), %r(32+i*10), -1840700269;
add.s32 %r(34+i*10), %r(33+i*10), %r(32+i*10);
shr.u32 %r(35+i*10), %r(34+i*10), 31;
shr.s32 %r(36+i*10), %r(34+i*10), 6;
add.s32 %r(37+i*10), %r(36+i*10), %r(35+i*10);
mul.lo.s32 %r(38+i*10), %r(37+i*10), 112;
sub.s32 %r(39+i*10), %r(32+i*10), %r(38+i*10);
add.s32 %r(40+i*10), %r8, %r(39+i*10);
mad.lo.s32 %r(41+i*10), %r(37+i*10), 784, %r(40+i*10);
mul.wide.s32 %rd(13+i*2), %r(41+i*10), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(9+i*1), [%rd(14+i*2)+0];
st.shared.f32 [%r31+896], %f15;
add.s32 %r105, %r105, 8;
setp.ne.s32 %p2, %r105, 112;
@%p2 bra $L__BB0_4;
bar.sync 0;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(16+i*3), [%r5+(0+i*4)];
ld.shared.f32 %f(17+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(18+i*3), %f(17+i*3), %f(16+i*3), %f(355+i*-337);
for.size.3 i in range(0, 109, 1):
ld.shared.f32 %f(22+i*3), [%r5+(8+i*4)];
ld.shared.f32 %f(23+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(24+i*3), %f(23+i*3), %f(22+i*3), %f(21+i*3);
ld.shared.f32 %f349, [%r5+444];
ld.shared.f32 %f350, [_ZZ24default_function_kernel0E20placeholder_d_shared+444];
fma.rn.f32 %f355, %f350, %f349, %f348;
add.s32 %r104, %r104, 1;
setp.lt.u32 %p3, %r104, 7;
@%p3 bra $L__BB0_1;
shl.b32 %r102, %r3, 5;
add.s32 %r103, %r102, %r1;
mul.wide.s32 %rd27, %r103, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f351, [%rd28];
add.f32 %f352, %f355, %f351;
mov.f32 %f353, 0f00000000;
max.f32 %f354, %f352, %f353;
add.s64 %rd29, %rd3, %rd27;
st.global.f32 [%rd29], %f354;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[112];
__shared__ float placeholder_shared[3584];
T_dense[0] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 7; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 112 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
if ((int)threadIdx.x < 16) {
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 112; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 25088 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) / 112 * 784 + k_outer_outer * 112 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) % 112;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 56; ++k_inner) {
int index4 = k_outer_inner * 56 + k_inner;
int index5 = (int)threadIdx.x * 112 + k_outer_inner * 56 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
}
}
}
int index6 = (int)blockIdx.x * 32 + (int)threadIdx.x;
T_relu[index6] = max(T_dense[0] + placeholder2[index6], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpijfqes39 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<227>;
.reg .b32 %r<32>;
.reg .b64 %rd<20>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[50176];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
mov.u32 %r1, %tid.x;
mul.lo.s32 %r14, %r1, 28;
cvta.to.global.u64 %rd11, %rd10;
mul.wide.s32 %rd12, %r14, 4;
add.s64 %rd1, %rd11, %rd12;
ld.global.nc.f32 %f3, [%rd1];
shl.b32 %r15, %r14, 2;
mov.u32 %r30, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r30, %r15;
for.size.2 i in range(0, 27, 1):
st.shared.f32 [%r2+(0+i*4)], %f(3+i*1);
ld.global.nc.f32 %f(4+i*1), [%rd1+(4+i*4)];
st.shared.f32 [%r2+108], %f30;
setp.gt.s32 %p1, %r1, 11;
@%p1 bra $L__BB0_2;
for.size.2 i in range(0, 28, 1):
ld.global.nc.f32 %f(31+i*1), [%rd1+(1792+i*4)];
st.shared.f32 [%r2+(1792+i*4)], %f(31+i*1);
$L__BB0_2:
mov.u32 %r3, %ctaid.x;
mad.lo.s32 %r18, %r3, 12544, %r1;
cvta.to.global.u64 %rd13, %rd7;
mul.wide.s32 %rd14, %r18, 4;
add.s64 %rd15, %rd13, %rd14;
add.s64 %rd19, %rd15, 512;
shl.b32 %r19, %r1, 2;
mov.u32 %r20, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r21, %r20, %r19;
add.s32 %r28, %r21, 512;
cvta.to.global.u64 %rd3, %rd9;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r29, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(59+i*1), [%rd19+(-512+i*64)];
st.shared.f32 [%r28+(-512+i*64)], %f(59+i*1);
add.s64 %rd19, %rd19, 1024;
for.size.1 i in range(0, 2, 1):
add.s32 %r(28+i*1), %r(28+i*1), (1024+i*-1008);
setp.ne.s32 %p2, %r29, 784;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r9, %r1, 3136, %r20;
mov.f32 %f226, 0f00000000;
mov.u32 %r31, 96;
$L__BB0_5:
add.s32 %r25, %r9, %r31;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(76+i*3), [%r25+(-96+i*4)];
ld.shared.f32 %f(77+i*3), [%r30+(0+i*4)];
fma.rn.f32 %f(78+i*3), %f(77+i*3), %f(76+i*3), %f(226+i*-148);
for.size.3 i in range(0, 46, 1):
ld.shared.f32 %f(82+i*3), [%r25+(-88+i*4)];
ld.shared.f32 %f(83+i*3), [%r30+(8+i*4)];
fma.rn.f32 %f(84+i*3), %f(83+i*3), %f(82+i*3), %f(81+i*3);
ld.shared.f32 %f220, [%r25+96];
ld.shared.f32 %f221, [%r30+192];
fma.rn.f32 %f226, %f221, %f220, %f219;
for.size.1 i in range(0, 2, 1):
add.s32 %r(30+i*1), %r(30+i*1), 196;
setp.ne.s32 %p3, %r31, 3232;
@%p3 bra $L__BB0_5;
shl.b32 %r26, %r3, 4;
add.s32 %r27, %r26, %r1;
mul.wide.s32 %rd16, %r27, 4;
add.s64 %rd17, %rd3, %rd16;
ld.global.nc.f32 %f222, [%rd17];
add.f32 %f223, %f226, %f222;
mov.f32 %f224, 0f00000000;
max.f32 %f225, %f223, %f224;
add.s64 %rd18, %rd4, %rd16;
st.global.f32 [%rd18], %f225;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[12544];
T_dense[0] = 0.000000e+00f;
int index0 = (int)threadIdx.x * 28;
placeholder_d_shared[index0] = placeholder[index0];
placeholder_d_shared[index0 + 1] = placeholder[index0 + 1];
placeholder_d_shared[index0 + 2] = placeholder[index0 + 2];
placeholder_d_shared[index0 + 3] = placeholder[index0 + 3];
placeholder_d_shared[index0 + 4] = placeholder[index0 + 4];
placeholder_d_shared[index0 + 5] = placeholder[index0 + 5];
placeholder_d_shared[index0 + 6] = placeholder[index0 + 6];
placeholder_d_shared[index0 + 7] = placeholder[index0 + 7];
placeholder_d_shared[index0 + 8] = placeholder[index0 + 8];
placeholder_d_shared[index0 + 9] = placeholder[index0 + 9];
placeholder_d_shared[index0 + 10] = placeholder[index0 + 10];
placeholder_d_shared[index0 + 11] = placeholder[index0 + 11];
placeholder_d_shared[index0 + 12] = placeholder[index0 + 12];
placeholder_d_shared[index0 + 13] = placeholder[index0 + 13];
placeholder_d_shared[index0 + 14] = placeholder[index0 + 14];
placeholder_d_shared[index0 + 15] = placeholder[index0 + 15];
placeholder_d_shared[index0 + 16] = placeholder[index0 + 16];
placeholder_d_shared[index0 + 17] = placeholder[index0 + 17];
placeholder_d_shared[index0 + 18] = placeholder[index0 + 18];
placeholder_d_shared[index0 + 19] = placeholder[index0 + 19];
placeholder_d_shared[index0 + 20] = placeholder[index0 + 20];
placeholder_d_shared[index0 + 21] = placeholder[index0 + 21];
placeholder_d_shared[index0 + 22] = placeholder[index0 + 22];
placeholder_d_shared[index0 + 23] = placeholder[index0 + 23];
placeholder_d_shared[index0 + 24] = placeholder[index0 + 24];
placeholder_d_shared[index0 + 25] = placeholder[index0 + 25];
placeholder_d_shared[index0 + 26] = placeholder[index0 + 26];
placeholder_d_shared[index0 + 27] = placeholder[index0 + 27];
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 448] = placeholder[index0 + 448];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 449] = placeholder[index0 + 449];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 450] = placeholder[index0 + 450];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 451] = placeholder[index0 + 451];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 452] = placeholder[index0 + 452];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 453] = placeholder[index0 + 453];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 454] = placeholder[index0 + 454];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 455] = placeholder[index0 + 455];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 456] = placeholder[index0 + 456];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 457] = placeholder[index0 + 457];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 458] = placeholder[index0 + 458];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 459] = placeholder[index0 + 459];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 460] = placeholder[index0 + 460];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 461] = placeholder[index0 + 461];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 462] = placeholder[index0 + 462];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 463] = placeholder[index0 + 463];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 464] = placeholder[index0 + 464];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 465] = placeholder[index0 + 465];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 466] = placeholder[index0 + 466];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 467] = placeholder[index0 + 467];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 468] = placeholder[index0 + 468];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 469] = placeholder[index0 + 469];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 470] = placeholder[index0 + 470];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 471] = placeholder[index0 + 471];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 472] = placeholder[index0 + 472];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 473] = placeholder[index0 + 473];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 474] = placeholder[index0 + 474];
}
if ((int)threadIdx.x < 12) {
placeholder_d_shared[index0 + 475] = placeholder[index0 + 475];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 784; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 16 + (int)threadIdx.x;
int index2 = (int)blockIdx.x * 12544 + ax0_ax1_fused_outer_outer * 16 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index2];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 16; ++k_outer_inner) {
int index3 = k_outer_inner * 49;
int index4 = (int)threadIdx.x * 784 + k_outer_inner * 49;
T_dense[0] = T_dense[0] + placeholder_d_shared[index3] * placeholder_shared[index4];
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 1] * placeholder_shared[index4 + 1]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 2] * placeholder_shared[index4 + 2]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 3] * placeholder_shared[index4 + 3]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 4] * placeholder_shared[index4 + 4]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 5] * placeholder_shared[index4 + 5]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 6] * placeholder_shared[index4 + 6]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 7] * placeholder_shared[index4 + 7]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 8] * placeholder_shared[index4 + 8]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 9] * placeholder_shared[index4 + 9]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 10] * placeholder_shared[index4 + 10]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 11] * placeholder_shared[index4 + 11]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 12] * placeholder_shared[index4 + 12]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 13] * placeholder_shared[index4 + 13]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 14] * placeholder_shared[index4 + 14]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 15] * placeholder_shared[index4 + 15]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 16] * placeholder_shared[index4 + 16]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 17] * placeholder_shared[index4 + 17]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 18] * placeholder_shared[index4 + 18]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 19] * placeholder_shared[index4 + 19]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 20] * placeholder_shared[index4 + 20]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 21] * placeholder_shared[index4 + 21]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 22] * placeholder_shared[index4 + 22]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 23] * placeholder_shared[index4 + 23]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 24] * placeholder_shared[index4 + 24]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 25] * placeholder_shared[index4 + 25]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 26] * placeholder_shared[index4 + 26]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 27] * placeholder_shared[index4 + 27]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 28] * placeholder_shared[index4 + 28]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 29] * placeholder_shared[index4 + 29]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 30] * placeholder_shared[index4 + 30]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 31] * placeholder_shared[index4 + 31]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 32] * placeholder_shared[index4 + 32]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 33] * placeholder_shared[index4 + 33]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 34] * placeholder_shared[index4 + 34]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 35] * placeholder_shared[index4 + 35]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 36] * placeholder_shared[index4 + 36]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 37] * placeholder_shared[index4 + 37]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 38] * placeholder_shared[index4 + 38]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 39] * placeholder_shared[index4 + 39]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 40] * placeholder_shared[index4 + 40]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 41] * placeholder_shared[index4 + 41]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 42] * placeholder_shared[index4 + 42]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 43] * placeholder_shared[index4 + 43]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 44] * placeholder_shared[index4 + 44]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 45] * placeholder_shared[index4 + 45]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 46] * placeholder_shared[index4 + 46]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 47] * placeholder_shared[index4 + 47]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 48] * placeholder_shared[index4 + 48]);
}
int index5 = (int)blockIdx.x * 16 + (int)threadIdx.x;
T_relu[index5] = max(T_dense[0] + placeholder2[index5], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpwtkvgh63 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<63>;
.reg .b32 %r<116>;
.reg .b64 %rd<16>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd3, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd5, [default_function_kernel0_param_3];
mov.u32 %r1, %tid.x;
cvta.to.global.u64 %rd7, %rd6;
mul.wide.s32 %rd8, %r1, 4;
add.s64 %rd1, %rd7, %rd8;
ld.global.nc.f32 %f9, [%rd1];
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r19, %r18;
for.size.2 i in range(0, 23, 1):
st.shared.f32 [%r2+(0+i*128)], %f(9+i*1);
ld.global.nc.f32 %f(10+i*1), [%rd1+(128+i*128)];
st.shared.f32 [%r2+2944], %f32;
setp.gt.s32 %p1, %r1, 15;
@%p1 bra $L__BB0_2;
ld.global.nc.f32 %f33, [%rd1+3072];
st.shared.f32 [%r2+3072], %f33;
$L__BB0_2:
shl.b32 %r3, %r1, 3;
mov.u32 %r21, %ctaid.x;
mad.lo.s32 %r4, %r21, 50176, %r3;
cvta.to.global.u64 %rd2, %rd3;
mov.u32 %r112, 0;
$L__BB0_3:
shl.b32 %r22, %r112, 8;
add.s32 %r23, %r22, %r3;
shl.b32 %r24, %r23, 2;
mov.u32 %r25, _ZZ24default_function_kernel0E18placeholder_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(26+i*1), %r(25+i*-21), %r(24+i*-2);
mul.wide.s32 %rd9, %r27, 4;
add.s64 %rd10, %rd2, %rd9;
for.size.4 i in range(0, 4, 1):
for.size.1 j in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(28+i*16+j*8), %r(29+i*16+j*8), %r(30+i*16+j*8), %r(31+i*16+j*8)}, [%rd10+(0+i*1024+j*16)];
for.size.1 j in range(0, 2, 1):
st.shared.v4.u32 [%r26+(0+i*1024+j*16)], {%r(28+i*16+j*8), %r(29+i*16+j*8), %r(30+i*16+j*8), %r(31+i*16+j*8)};
add.s32 %r112, %r112, 4;
setp.ne.s32 %p2, %r112, 196;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r7, %r1, 6272, %r25;
mov.u32 %r113, 0;
mov.f32 %f61, 0f00000000;
mov.f32 %f62, %f61;
$L__BB0_5:
mul.lo.s32 %r95, %r113, 196;
for.size.1 i in range(0, 2, 1):
add.s32 %r(9+i*88), %r(7+i*12), %r95;
add.s32 %r114, %r97, 8;
mov.u32 %r115, 3136;
bra.uni $L__BB0_6;
$L__BB0_7:
for.size.2 i in range(0, 2, 1):
add.s32 %r(103+i*1), %r(98+i*16), (-3136+i*3128);
ld.shared.f32 %f(39+i*1), [%r(103+i*1)+4];
fma.rn.f32 %f41, %f40, %f39, %f62;
add.s32 %r105, %r98, -3136;
ld.shared.f32 %f42, [%r105+3140];
fma.rn.f32 %f43, %f40, %f42, %f61;
for.size.2 i in range(0, 2, 1):
add.s32 %r(106+i*1), %r(98+i*16), (-3136+i*3128);
ld.shared.f32 %f(44+i*1), [%r(106+i*1)+8];
fma.rn.f32 %f46, %f45, %f44, %f41;
add.s32 %r108, %r98, -3136;
ld.shared.f32 %f47, [%r108+3144];
fma.rn.f32 %f48, %f45, %f47, %f43;
for.size.2 i in range(0, 2, 1):
add.s32 %r(109+i*1), %r(98+i*16), (-3136+i*3128);
ld.shared.f32 %f(49+i*1), [%r(109+i*1)+12];
fma.rn.f32 %f62, %f50, %f49, %f46;
add.s32 %r111, %r98, -3136;
ld.shared.f32 %f51, [%r111+3148];
fma.rn.f32 %f61, %f50, %f51, %f48;
for.size.1 i in range(0, 2, 1):
add.s32 %r(115+i*-1), %r(115+i*-1), 16;
$L__BB0_6:
add.s32 %r98, %r9, %r115;
ld.shared.f32 %f36, [%r98+-3136];
ld.shared.f32 %f37, [%r114+-8];
fma.rn.f32 %f62, %f37, %f36, %f62;
ld.shared.f32 %f38, [%r98];
fma.rn.f32 %f61, %f37, %f38, %f61;
setp.eq.s32 %p3, %r115, 3328;
@%p3 bra $L__BB0_8;
bra.uni $L__BB0_7;
$L__BB0_8:
add.s32 %r113, %r113, 1;
setp.lt.u32 %p4, %r113, 16;
@%p4 bra $L__BB0_5;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(100+i*1), %r(21+i*-20), (6+i*-5);
add.s32 %r102, %r100, %r101;
cvta.to.global.u64 %rd11, %rd5;
mul.wide.s32 %rd12, %r102, 4;
add.s64 %rd13, %rd11, %rd12;
ld.global.nc.f32 %f52, [%rd13];
add.f32 %f53, %f62, %f52;
mov.f32 %f54, 0f00000000;
max.f32 %f55, %f53, %f54;
cvta.to.global.u64 %rd14, %rd4;
add.s64 %rd15, %rd14, %rd12;
st.global.f32 [%rd15], %f55;
ld.global.nc.f32 %f56, [%rd13+4];
add.f32 %f57, %f61, %f56;
max.f32 %f58, %f57, %f54;
st.global.f32 [%rd15+4], %f58;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[50176];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
int index0 = (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index0];
placeholder_d_shared[index0 + 32] = placeholder[index0 + 32];
placeholder_d_shared[index0 + 64] = placeholder[index0 + 64];
placeholder_d_shared[index0 + 96] = placeholder[index0 + 96];
placeholder_d_shared[index0 + 128] = placeholder[index0 + 128];
placeholder_d_shared[index0 + 160] = placeholder[index0 + 160];
placeholder_d_shared[index0 + 192] = placeholder[index0 + 192];
placeholder_d_shared[index0 + 224] = placeholder[index0 + 224];
placeholder_d_shared[index0 + 256] = placeholder[index0 + 256];
placeholder_d_shared[index0 + 288] = placeholder[index0 + 288];
placeholder_d_shared[index0 + 320] = placeholder[index0 + 320];
placeholder_d_shared[index0 + 352] = placeholder[index0 + 352];
placeholder_d_shared[index0 + 384] = placeholder[index0 + 384];
placeholder_d_shared[index0 + 416] = placeholder[index0 + 416];
placeholder_d_shared[index0 + 448] = placeholder[index0 + 448];
placeholder_d_shared[index0 + 480] = placeholder[index0 + 480];
placeholder_d_shared[index0 + 512] = placeholder[index0 + 512];
placeholder_d_shared[index0 + 544] = placeholder[index0 + 544];
placeholder_d_shared[index0 + 576] = placeholder[index0 + 576];
placeholder_d_shared[index0 + 608] = placeholder[index0 + 608];
placeholder_d_shared[index0 + 640] = placeholder[index0 + 640];
placeholder_d_shared[index0 + 672] = placeholder[index0 + 672];
placeholder_d_shared[index0 + 704] = placeholder[index0 + 704];
placeholder_d_shared[index0 + 736] = placeholder[index0 + 736];
if ((int)threadIdx.x < 16) {
placeholder_d_shared[index0 + 768] = placeholder[index0 + 768];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 196; ++ax0_ax1_fused_outer_outer) {
ulonglong4*(placeholder_shared + ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 8)[0] = ulonglong4*(placeholder1 + (int)blockIdx.x * 50176 + ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 8)[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 16; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 49; ++k_inner) {
int index1 = k_outer_inner * 49 + k_inner;
int index2 = (int)threadIdx.x * 1568 + k_outer_inner * 49 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index1] * placeholder_shared[index2];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index1] * placeholder_shared[index2 + 784]);
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index3 = (int)blockIdx.x * 64 + (int)threadIdx.x * 2 + ax1_inner;
int index4 = ax1_inner;
T_relu[index3] = max(T_dense[index4] + placeholder2[index3], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmppet5lnpj | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<218>;
.reg .b32 %r<117>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[448];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[57344];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd7, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd5;
cvta.to.global.u64 %rd3, %rd7;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r1, %tid.x;
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r19, %r18;
mul.lo.s32 %r3, %r1, 896;
add.s32 %r4, %r1, 56;
mov.u32 %r20, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r20, %r18;
mov.u32 %r112, 0;
mov.f32 %f186, 0f00000000;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(187+i*1), %f186;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r7, %r112, 112;
add.s32 %r22, %r7, %r1;
mul.wide.s32 %rd9, %r22, 4;
add.s64 %rd10, %rd2, %rd9;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(65+i*1), [%rd10+(0+i*32)];
st.shared.f32 [%r2+(0+i*32)], %f(65+i*1);
ld.global.nc.f32 %f69, [%rd10+128];
mov.u32 %r115, 128;
for.size.2 i in range(0, 9, 1):
st.shared.f32 [%r2+(128+i*32)], %f(69+i*1);
ld.global.nc.f32 %f(70+i*1), [%rd10+(160+i*32)];
st.shared.f32 [%r2+416], %f78;
mov.u32 %r113, %r1;
mov.u32 %r114, %r4;
$L__BB0_2:
mul.hi.s32 %r23, %r113, -1840700269;
add.s32 %r24, %r23, %r113;
shr.u32 %r25, %r24, 31;
shr.s32 %r26, %r24, 6;
add.s32 %r27, %r26, %r25;
mul.lo.s32 %r28, %r27, 112;
sub.s32 %r29, %r113, %r28;
add.s32 %r30, %r29, %r7;
mad.lo.s32 %r31, %r27, 784, %r30;
mul.wide.s32 %rd11, %r31, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f79, [%rd12];
add.s32 %r32, %r5, %r115;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r32+(-128+i*32)], %f(79+i*1);
add.s32 %r(33+i*10), %r113, (8+i*8);
mul.hi.s32 %r(34+i*10), %r(33+i*10), -1840700269;
add.s32 %r(35+i*10), %r(34+i*10), %r(33+i*10);
shr.u32 %r(36+i*10), %r(35+i*10), 31;
shr.s32 %r(37+i*10), %r(35+i*10), 6;
add.s32 %r(38+i*10), %r(37+i*10), %r(36+i*10);
mul.lo.s32 %r(39+i*10), %r(38+i*10), 112;
sub.s32 %r(40+i*10), %r(33+i*10), %r(39+i*10);
add.s32 %r(41+i*10), %r(40+i*10), %r7;
mad.lo.s32 %r(42+i*10), %r(38+i*10), 784, %r(41+i*10);
mul.wide.s32 %rd(13+i*2), %r(42+i*10), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(80+i*1), [%rd(14+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r32+(0+i*32)], %f(83+i*1);
add.s32 %r(73+i*10), %r114, (-16+i*8);
mul.hi.s32 %r(74+i*10), %r(73+i*10), -1840700269;
add.s32 %r(75+i*10), %r(74+i*10), %r(73+i*10);
shr.u32 %r(76+i*10), %r(75+i*10), 31;
shr.s32 %r(77+i*10), %r(75+i*10), 6;
add.s32 %r(78+i*10), %r(77+i*10), %r(76+i*10);
mul.lo.s32 %r(79+i*10), %r(78+i*10), 112;
sub.s32 %r(80+i*10), %r(73+i*10), %r(79+i*10);
add.s32 %r(81+i*10), %r(80+i*10), %r7;
mad.lo.s32 %r(82+i*10), %r(78+i*10), 784, %r(81+i*10);
mul.wide.s32 %rd(21+i*2), %r(82+i*10), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(84+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r32+64], %f85;
mul.hi.s32 %r93, %r114, -1840700269;
add.s32 %r94, %r93, %r114;
shr.u32 %r95, %r94, 31;
shr.s32 %r96, %r94, 6;
add.s32 %r97, %r96, %r95;
mul.lo.s32 %r98, %r97, 112;
sub.s32 %r99, %r114, %r98;
add.s32 %r100, %r99, %r7;
mad.lo.s32 %r101, %r97, 784, %r100;
mul.wide.s32 %rd25, %r101, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f86, [%rd26];
st.shared.f32 [%r32+96], %f86;
add.s32 %r115, %r115, 256;
setp.ne.s32 %p1, %r115, 57472;
for.size.1 i in range(0, 2, 1):
add.s32 %r(12+i*101), %r114, (64+i*-56);
mov.u32 %r114, %r12;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r116, 0;
$L__BB0_4:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(103+i*1), %r116, (1+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(106+i*1), %r(19+i*84), %r(104+i*-101);
shl.b32 %r108, %r107, 2;
add.s32 %r110, %r20, %r108;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(87+i*1), [%r(110+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(89+i*2), %f88, %f(87+i*3), %f(201+i*-1);
ld.shared.f32 %f(90+i*2), [%r110+(28672+i*-28668)];
ld.shared.f32 %f93, [%r106+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(201+i*-1), %f93, %f(92+i*2), %f(89+i*2);
ld.shared.f32 %f(94+i*1), [%r110+(28676+i*-28228)];
for.size.6 i in range(0, 6, 1):
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(96+i*6+j*2), %f88, %f(95+i*6+j*2), %f(199+i*-2+j*-1);
ld.shared.f32 %f(97+i*6+j*2), [%r110+(29120+i*448+j*-28668)];
for.size.2 j in range(0, 2, 1):
fma.rn.f32 %f(199+i*-2+j*-1), %f93, %f(99+i*6+j*1), %f(96+i*6+j*2);
ld.shared.f32 %f(100+i*6+j*1), [%r110+(29124+i*448+j*-28228)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(132+i*2), %f88, %f(131+i*2), %f(187+i*-1);
ld.shared.f32 %f(133+i*2), [%r110+(31808+i*-28668)];
fma.rn.f32 %f187, %f93, %f135, %f132;
ld.shared.f32 %f136, [%r110+31812];
fma.rn.f32 %f186, %f93, %f136, %f134;
add.s32 %r116, %r116, 1;
setp.ne.s32 %p2, %r116, 56;
@%p2 bra $L__BB0_4;
add.s32 %r112, %r112, 1;
setp.lt.u32 %p3, %r112, 7;
@%p3 bra $L__BB0_1;
shl.b32 %r111, %r1, 3;
mul.wide.s32 %rd27, %r111, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f137, [%rd28];
add.f32 %f138, %f201, %f137;
mov.f32 %f139, 0f00000000;
max.f32 %f140, %f138, %f139;
add.s64 %rd29, %rd3, %rd27;
for.size.5 i in range(0, 7, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(0+i*4+j*256)], %f(140+i*6+j*3);
ld.global.nc.f32 %f(141+i*6+j*3), [%rd28+(256+i*4+j*-252)];
add.f32 %f(142+i*6+j*3), %f(200+i*-2+j*-1), %f(141+i*6+j*3);
max.f32 %f(143+i*6+j*3), %f(142+i*6+j*3), %f139;
st.global.f32 [%rd29+28], %f182;
ld.global.nc.f32 %f183, [%rd28+284];
add.f32 %f184, %f186, %f183;
max.f32 %f185, %f184, %f139;
st.global.f32 [%rd29+284], %f185;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[112];
__shared__ float placeholder_shared[14336];
T_dense[0] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 7; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 112 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 40] = placeholder[index1 + 40];
placeholder_d_shared[index0 + 48] = placeholder[index1 + 48];
placeholder_d_shared[index0 + 56] = placeholder[index1 + 56];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 72] = placeholder[index1 + 72];
placeholder_d_shared[index0 + 80] = placeholder[index1 + 80];
placeholder_d_shared[index0 + 88] = placeholder[index1 + 88];
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
placeholder_d_shared[index0 + 104] = placeholder[index1 + 104];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1792; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x) / 112 * 784 + k_outer_outer * 112 + (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x) % 112;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 56; ++k_outer_inner) {
int index4 = k_outer_inner * 2;
int index5 = (int)threadIdx.x * 896 + k_outer_inner * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7168]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7169]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 112]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7280]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 113]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7281]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 224]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7392]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 225]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7393]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 336]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7504]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 337]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7505]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 448]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7616]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 449]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7617]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 560]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7728]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 561]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7729]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 672]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7840]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 673]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7841]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7952]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 785]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 7953]);
}
}
for (int ax1_inner = 0; ax1_inner < 8; ++ax1_inner) {
int index6 = (int)threadIdx.x * 8 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 64] = max(T_dense[index7 + 8] + placeholder2[index6 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp0pv1s1pa | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.local .align 16 .b8 __local_depot0[256];
.reg .b64 %SP;
.reg .b64 %SPL;
.reg .pred %p<4>;
.reg .f32 %f<472>;
.reg .b32 %r<114>;
.reg .b64 %rd<35>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[56];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[7168];
mov.u64 %SPL, __local_depot0;
ld.param.u64 %rd7, [default_function_kernel0_param_0];
ld.param.u64 %rd8, [default_function_kernel0_param_1];
ld.param.u64 %rd9, [default_function_kernel0_param_2];
ld.param.u64 %rd10, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd9;
add.u64 %rd3, %SPL, 0;
mov.f32 %f15, 0f00000000;
for.size.1 i in range(0, 16, 1):
st.local.v4.f32 [%rd3+(0+i*16)], {%f15, %f15, %f15, %f15};
cvta.to.global.u64 %rd5, %rd8;
cvta.to.global.u64 %rd6, %rd10;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 896;
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r19, %r18;
add.s32 %r4, %r1, 14;
mov.u32 %r20, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r20, %r18;
mov.u32 %r109, 0;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r7, %r109, 14;
add.s32 %r22, %r7, %r1;
mul.wide.s32 %rd12, %r22, 4;
add.s64 %rd13, %rd1, %rd12;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(16+i*1), [%rd13+(0+i*8)];
st.shared.f32 [%r3+(0+i*8)], %f(16+i*1);
ld.global.nc.f32 %f20, [%rd13+32];
mov.u32 %r112, 32;
for.size.2 i in range(0, 2, 1):
st.shared.f32 [%r3+(32+i*8)], %f(20+i*1);
ld.global.nc.f32 %f(21+i*1), [%rd13+(40+i*8)];
st.shared.f32 [%r3+48], %f22;
mov.u32 %r110, %r1;
mov.u32 %r111, %r4;
$L__BB0_2:
mul.hi.s32 %r23, %r110, -1840700269;
add.s32 %r24, %r23, %r110;
shr.u32 %r25, %r24, 31;
shr.s32 %r26, %r24, 3;
add.s32 %r27, %r26, %r25;
mul.lo.s32 %r28, %r27, 14;
sub.s32 %r29, %r110, %r28;
add.s32 %r30, %r29, %r7;
mad.lo.s32 %r31, %r27, 784, %r30;
mul.wide.s32 %rd14, %r31, 4;
add.s64 %rd15, %rd5, %rd14;
ld.global.nc.f32 %f23, [%rd15];
add.s32 %r32, %r5, %r112;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r32+(-32+i*8)], %f(23+i*1);
add.s32 %r(33+i*10), %r110, (2+i*2);
mul.hi.s32 %r(34+i*10), %r(33+i*10), -1840700269;
add.s32 %r(35+i*10), %r(34+i*10), %r(33+i*10);
shr.u32 %r(36+i*10), %r(35+i*10), 31;
shr.s32 %r(37+i*10), %r(35+i*10), 3;
add.s32 %r(38+i*10), %r(37+i*10), %r(36+i*10);
mul.lo.s32 %r(39+i*10), %r(38+i*10), 14;
sub.s32 %r(40+i*10), %r(33+i*10), %r(39+i*10);
add.s32 %r(41+i*10), %r(40+i*10), %r7;
mad.lo.s32 %r(42+i*10), %r(38+i*10), 784, %r(41+i*10);
mul.wide.s32 %rd(16+i*2), %r(42+i*10), 4;
add.s64 %rd(17+i*2), %rd5, %rd(16+i*2);
ld.global.nc.f32 %f(24+i*1), [%rd(17+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r32+(0+i*8)], %f(27+i*1);
add.s32 %r(73+i*10), %r111, (-4+i*2);
mul.hi.s32 %r(74+i*10), %r(73+i*10), -1840700269;
add.s32 %r(75+i*10), %r(74+i*10), %r(73+i*10);
shr.u32 %r(76+i*10), %r(75+i*10), 31;
shr.s32 %r(77+i*10), %r(75+i*10), 3;
add.s32 %r(78+i*10), %r(77+i*10), %r(76+i*10);
mul.lo.s32 %r(79+i*10), %r(78+i*10), 14;
sub.s32 %r(80+i*10), %r(73+i*10), %r(79+i*10);
add.s32 %r(81+i*10), %r(80+i*10), %r7;
mad.lo.s32 %r(82+i*10), %r(78+i*10), 784, %r(81+i*10);
mul.wide.s32 %rd(24+i*2), %r(82+i*10), 4;
add.s64 %rd(25+i*2), %rd5, %rd(24+i*2);
ld.global.nc.f32 %f(28+i*1), [%rd(25+i*2)+0];
st.shared.f32 [%r32+16], %f29;
mul.hi.s32 %r93, %r111, -1840700269;
add.s32 %r94, %r93, %r111;
shr.u32 %r95, %r94, 31;
shr.s32 %r96, %r94, 3;
add.s32 %r97, %r96, %r95;
mul.lo.s32 %r98, %r97, 14;
sub.s32 %r99, %r111, %r98;
add.s32 %r100, %r99, %r7;
mad.lo.s32 %r101, %r97, 784, %r100;
mul.wide.s32 %rd28, %r101, 4;
add.s64 %rd29, %rd5, %rd28;
ld.global.nc.f32 %f30, [%rd29];
st.shared.f32 [%r32+24], %f30;
for.size.1 i in range(0, 2, 1):
add.s32 %r(111+i*-1), %r(111+i*-1), 16;
add.s32 %r112, %r112, 64;
setp.ne.s32 %p1, %r112, 7200;
@%p1 bra $L__BB0_2;
bar.sync 0;
ld.shared.f32 %f1, [_ZZ24default_function_kernel0E20placeholder_d_shared];
mov.u32 %r113, 0;
for.size.1 i in range(0, 13, 1):
ld.shared.f32 %f(2+i*1), [_ZZ24default_function_kernel0E20placeholder_d_shared+(4+i*4)];
$L__BB0_4:
mad.lo.s32 %r103, %r113, 56, %r2;
shl.b32 %r104, %r113, 2;
mul.wide.s32 %rd30, %r104, 4;
add.s64 %rd31, %rd3, %rd30;
ld.local.v4.f32 {%f31, %f32, %f33, %f34}, [%rd31];
shl.b32 %r105, %r103, 2;
add.s32 %r107, %r20, %r105;
for.size.2 i in range(0, 4, 1):
ld.shared.f32 %f(39+i*2), [%r107+(0+i*56)];
fma.rn.f32 %f(40+i*2), %f1, %f(39+i*2), %f(31+i*1);
for.size.3 i in range(0, 12, 1):
for.size.2 j in range(0, 4, 1):
ld.shared.f32 %f(47+i*8+j*2), [%r107+(4+i*4+j*56)];
fma.rn.f32 %f(48+i*8+j*2), %f(2+j*1), %f(47+i*8+j*2), %f(40+i*8+j*2);
for.size.1 i in range(0, 4, 1):
ld.shared.f32 %f(143+i*1), [%r107+(52+i*56)];
for.size.1 i in range(0, 4, 1):
fma.rn.f32 %f(147+i*1), %f14, %f(146+i*-1), %f(142+i*-2);
st.local.v4.f32 [%rd31], {%f150, %f149, %f148, %f147};
add.s32 %r113, %r113, 1;
setp.ne.s32 %p2, %r113, 16;
@%p2 bra $L__BB0_4;
add.s32 %r109, %r109, 1;
setp.lt.u32 %p3, %r109, 56;
@%p3 bra $L__BB0_1;
shl.b32 %r108, %r1, 6;
ld.local.v4.f32 {%f151, %f152, %f153, %f154}, [%rd3];
mul.wide.s32 %rd32, %r108, 4;
add.s64 %rd33, %rd6, %rd32;
ld.global.nc.f32 %f159, [%rd33];
add.f32 %f160, %f151, %f159;
mov.f32 %f161, 0f00000000;
max.f32 %f162, %f160, %f161;
add.s64 %rd34, %rd2, %rd32;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd34+(0+i*4)], %f(162+i*3);
ld.global.nc.f32 %f(163+i*3), [%rd33+(4+i*4)];
add.f32 %f(164+i*3), %f(152+i*1), %f(163+i*3);
max.f32 %f(165+i*3), %f(164+i*3), %f161;
for.size.17 i in range(0, 15, 1):
st.global.f32 [%rd34+(12+i*16)], %f(171+i*20);
ld.local.v4.f32 {%f(172+i*20), %f(173+i*20), %f(174+i*20), %f(175+i*20)}, [%rd3+(16+i*16)];
ld.global.nc.f32 %f(180+i*20), [%rd33+(16+i*16)];
add.f32 %f(181+i*20), %f(172+i*20), %f(180+i*20);
max.f32 %f(182+i*20), %f(181+i*20), %f161;
st.global.f32 [%rd34+(16+i*16)], %f(182+i*20);
ld.global.nc.f32 %f(183+i*20), [%rd33+(20+i*16)];
add.f32 %f(184+i*20), %f(173+i*20), %f(183+i*20);
max.f32 %f(185+i*20), %f(184+i*20), %f161;
st.global.f32 [%rd34+(20+i*16)], %f(185+i*20);
ld.global.nc.f32 %f(186+i*20), [%rd33+(24+i*16)];
add.f32 %f(187+i*20), %f(174+i*20), %f(186+i*20);
max.f32 %f(188+i*20), %f(187+i*20), %f161;
st.global.f32 [%rd34+(24+i*16)], %f(188+i*20);
ld.global.nc.f32 %f(189+i*20), [%rd33+(28+i*16)];
add.f32 %f(190+i*20), %f(175+i*20), %f(189+i*20);
max.f32 %f(191+i*20), %f(190+i*20), %f161;
st.global.f32 [%rd34+252], %f471;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[14];
__shared__ float placeholder_shared[1792];
for (int j_outer_inner_init = 0; j_outer_inner_init < 16; ++j_outer_inner_init) {
for (int j_inner_init = 0; j_inner_init < 4; ++j_inner_init) {
int index0 = j_outer_inner_init * 4 + j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
}
for (int k_outer_outer = 0; k_outer_outer < 56; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 7; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index2 = k_outer_outer * 14 + ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 896; ++ax0_ax1_fused_outer_outer1) {
int index3 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
int index4 = (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) / 14 * 784 + k_outer_outer * 14 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) % 14;
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int j_outer_inner = 0; j_outer_inner < 16; ++j_outer_inner) {
for (int k_inner = 0; k_inner < 14; ++k_inner) {
for (int j_inner = 0; j_inner < 4; ++j_inner) {
int index5 = j_outer_inner * 4 + j_inner;
int index6 = k_inner;
int index7 = (int)threadIdx.x * 896 + j_outer_inner * 56 + j_inner * 14 + k_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index8 = (int)threadIdx.x * 64 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp6j96wr30 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<2>;
.reg .f32 %f<78>;
.reg .b32 %r<8>;
.reg .b64 %rd<21>;
// demoted variable
.shared .align 4 .f32 _ZZ24default_function_kernel0E20placeholder_d_shared_$_0;
// demoted variable
.shared .align 4 .f32 _ZZ24default_function_kernel0E18placeholder_shared_$_0;
// demoted variable
.shared .align 4 .f32 _ZZ24default_function_kernel0E18placeholder_shared_$_1;
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd19, %rd9;
cvta.to.global.u64 %rd2, %rd11;
mov.u32 %r1, %ctaid.x;
mul.lo.s32 %r5, %r1, 1568;
mul.wide.s32 %rd13, %r5, 4;
add.s64 %rd14, %rd12, %rd13;
add.s64 %rd20, %rd14, 3164;
mov.u32 %r7, 0;
mov.f32 %f76, 0f00000000;
mov.f32 %f77, %f76;
$L__BB0_1:
for.size.13 i in range(0, 2, 1):
ld.global.nc.f32 %f(7+i*8), [%rd19+(0+i*4)];
st.shared.f32 [_ZZ24default_function_kernel0E20placeholder_d_shared_$_0+0], %f(7+i*8);
bar.sync 0;
ld.global.nc.f32 %f(8+i*8), [%rd20+(-3164+i*4)];
st.shared.f32 [_ZZ24default_function_kernel0E18placeholder_shared_$_0+0], %f(8+i*8);
ld.global.nc.f32 %f(9+i*8), [%rd20+(-28+i*4)];
st.shared.f32 [_ZZ24default_function_kernel0E18placeholder_shared_$_1+0], %f(9+i*8);
bar.sync 0;
ld.shared.f32 %f(10+i*8), [_ZZ24default_function_kernel0E18placeholder_shared_$_0+0];
ld.shared.f32 %f(11+i*8), [_ZZ24default_function_kernel0E20placeholder_d_shared_$_0+0];
fma.rn.f32 %f(12+i*8), %f(11+i*8), %f(10+i*8), %f(77+i*-65);
ld.shared.f32 %f(13+i*8), [_ZZ24default_function_kernel0E18placeholder_shared_$_1+0];
fma.rn.f32 %f(14+i*8), %f(11+i*8), %f(13+i*8), %f(76+i*-62);
for.size.13 i in range(0, 5, 1):
ld.global.nc.f32 %f(23+i*8), [%rd19+(8+i*4)];
st.shared.f32 [_ZZ24default_function_kernel0E20placeholder_d_shared_$_0+0], %f(23+i*8);
bar.sync 0;
ld.global.nc.f32 %f(24+i*8), [%rd20+(-3156+i*4)];
st.shared.f32 [_ZZ24default_function_kernel0E18placeholder_shared_$_0+0], %f(24+i*8);
ld.global.nc.f32 %f(25+i*8), [%rd20+(-20+i*4)];
st.shared.f32 [_ZZ24default_function_kernel0E18placeholder_shared_$_1+0], %f(25+i*8);
bar.sync 0;
ld.shared.f32 %f(26+i*8), [_ZZ24default_function_kernel0E18placeholder_shared_$_0+0];
ld.shared.f32 %f(27+i*8), [_ZZ24default_function_kernel0E20placeholder_d_shared_$_0+0];
fma.rn.f32 %f(28+i*8), %f(27+i*8), %f(26+i*8), %f(20+i*8);
ld.shared.f32 %f(29+i*8), [_ZZ24default_function_kernel0E18placeholder_shared_$_1+0];
fma.rn.f32 %f(30+i*8), %f(27+i*8), %f(29+i*8), %f(22+i*8);
ld.global.nc.f32 %f63, [%rd19+28];
st.shared.f32 [_ZZ24default_function_kernel0E20placeholder_d_shared_$_0], %f63;
bar.sync 0;
ld.global.nc.f32 %f64, [%rd20+-3136];
st.shared.f32 [_ZZ24default_function_kernel0E18placeholder_shared_$_0], %f64;
ld.global.nc.f32 %f65, [%rd20];
st.shared.f32 [_ZZ24default_function_kernel0E18placeholder_shared_$_1], %f65;
bar.sync 0;
ld.shared.f32 %f66, [_ZZ24default_function_kernel0E18placeholder_shared_$_0];
ld.shared.f32 %f67, [_ZZ24default_function_kernel0E20placeholder_d_shared_$_0];
fma.rn.f32 %f77, %f67, %f66, %f60;
ld.shared.f32 %f68, [_ZZ24default_function_kernel0E18placeholder_shared_$_1];
fma.rn.f32 %f76, %f67, %f68, %f62;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(20+i*-1), %rd(20+i*-1), 32;
add.s32 %r7, %r7, 8;
setp.ne.s32 %p1, %r7, 784;
@%p1 bra $L__BB0_1;
cvta.to.global.u64 %rd15, %rd8;
shl.b32 %r6, %r1, 1;
mul.wide.s32 %rd16, %r6, 4;
add.s64 %rd17, %rd15, %rd16;
ld.global.nc.f32 %f69, [%rd17];
add.f32 %f70, %f77, %f69;
mov.f32 %f71, 0f00000000;
max.f32 %f72, %f70, %f71;
add.s64 %rd18, %rd2, %rd16;
st.global.f32 [%rd18], %f72;
ld.global.nc.f32 %f73, [%rd17+4];
add.f32 %f74, %f76, %f73;
max.f32 %f75, %f74, %f71;
st.global.f32 [%rd18+4], %f75;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[2];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
int index0 = k_outer_outer;
placeholder_d_shared[0] = placeholder[index0];
__syncthreads();
int index1 = (int)blockIdx.x * 1568 + k_outer_outer;
placeholder_shared[0] = placeholder1[index1];
placeholder_shared[1] = placeholder1[index1 + 784];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[0];
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[1];
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index2 = (int)blockIdx.x * 2 + ax1_inner;
int index3 = ax1_inner;
T_relu[index2] = max(T_dense[index3] + placeholder2[index2], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpnxhshjwm | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<6>;
.reg .f32 %f<298>;
.reg .b32 %r<191>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[112];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[14336];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
shl.b32 %r3, %r1, 2;
mov.u32 %r21, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r21, %r3;
mul.lo.s32 %r4, %r1, 112;
add.s32 %r5, %r3, 224;
shl.b32 %r22, %r1, 4;
mov.u32 %r23, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r6, %r23, %r22;
mov.u32 %r185, 0;
mov.f32 %f250, 0f00000000;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(251+i*1), %f250;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r185, 28;
add.s32 %r24, %r8, %r1;
mul.wide.s32 %rd10, %r24, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.2 i in range(0, 3, 1):
ld.global.nc.f32 %f(97+i*1), [%rd5+(0+i*32)];
st.shared.f32 [%r2+(0+i*32)], %f(97+i*1);
setp.gt.s32 %p1, %r1, 3;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f100, [%rd5+96];
st.shared.f32 [%r2+96], %f100;
$L__BB0_3:
mov.u32 %r188, 512;
mov.u32 %r186, %r3;
mov.u32 %r187, %r5;
$L__BB0_4:
mul.hi.s32 %r26, %r186, -1840700269;
add.s32 %r27, %r26, %r186;
shr.u32 %r28, %r27, 31;
shr.s32 %r29, %r27, 4;
add.s32 %r30, %r29, %r28;
mul.lo.s32 %r31, %r30, 28;
sub.s32 %r32, %r186, %r31;
add.s32 %r33, %r32, %r8;
mad.lo.s32 %r34, %r30, 784, %r33;
mul.wide.s32 %rd11, %r34, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.v4.u32 {%r35, %r36, %r37, %r38}, [%rd12];
add.s32 %r43, %r6, %r188;
for.size.14 i in range(0, 2, 1):
st.shared.v4.u32 [%r43+(-512+i*128)], {%r(35+i*19), %r(36+i*19), %r(37+i*19), %r(38+i*19)};
add.s32 %r(44+i*18), %r186, (32+i*32);
mul.hi.s32 %r(45+i*18), %r(44+i*18), -1840700269;
add.s32 %r(46+i*18), %r(45+i*18), %r(44+i*18);
shr.u32 %r(47+i*18), %r(46+i*18), 31;
shr.s32 %r(48+i*18), %r(46+i*18), 4;
add.s32 %r(49+i*18), %r(48+i*18), %r(47+i*18);
mul.lo.s32 %r(50+i*18), %r(49+i*18), 28;
sub.s32 %r(51+i*18), %r(44+i*18), %r(50+i*18);
add.s32 %r(52+i*18), %r(51+i*18), %r8;
mad.lo.s32 %r(53+i*18), %r(49+i*18), 784, %r(52+i*18);
mul.wide.s32 %rd(13+i*2), %r(53+i*18), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.v4.u32 {%r(54+i*18), %r(55+i*18), %r(56+i*18), %r(57+i*18)}, [%rd(14+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.v4.u32 [%r43+(-256+i*128)], {%r(72+i*18), %r(73+i*18), %r(74+i*18), %r(75+i*18)};
add.s32 %r(80+i*18), %r186, (96+i*32);
mul.hi.s32 %r(81+i*18), %r(80+i*18), -1840700269;
add.s32 %r(82+i*18), %r(81+i*18), %r(80+i*18);
shr.u32 %r(83+i*18), %r(82+i*18), 31;
shr.s32 %r(84+i*18), %r(82+i*18), 4;
add.s32 %r(85+i*18), %r(84+i*18), %r(83+i*18);
mul.lo.s32 %r(86+i*18), %r(85+i*18), 28;
sub.s32 %r(87+i*18), %r(80+i*18), %r(86+i*18);
add.s32 %r(88+i*18), %r(87+i*18), %r8;
mad.lo.s32 %r(89+i*18), %r(85+i*18), 784, %r(88+i*18);
mul.wide.s32 %rd(17+i*2), %r(89+i*18), 4;
add.s64 %rd(18+i*2), %rd1, %rd(17+i*2);
ld.global.nc.v4.u32 {%r(90+i*18), %r(91+i*18), %r(92+i*18), %r(93+i*18)}, [%rd(18+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.v4.u32 [%r43+(0+i*128)], {%r(108+i*18), %r(109+i*18), %r(110+i*18), %r(111+i*18)};
add.s32 %r(116+i*18), %r187, (-64+i*32);
mul.hi.s32 %r(117+i*18), %r(116+i*18), -1840700269;
add.s32 %r(118+i*18), %r(117+i*18), %r(116+i*18);
shr.u32 %r(119+i*18), %r(118+i*18), 31;
shr.s32 %r(120+i*18), %r(118+i*18), 4;
add.s32 %r(121+i*18), %r(120+i*18), %r(119+i*18);
mul.lo.s32 %r(122+i*18), %r(121+i*18), 28;
sub.s32 %r(123+i*18), %r(116+i*18), %r(122+i*18);
add.s32 %r(124+i*18), %r(123+i*18), %r8;
mad.lo.s32 %r(125+i*18), %r(121+i*18), 784, %r(124+i*18);
mul.wide.s32 %rd(21+i*2), %r(125+i*18), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.v4.u32 {%r(126+i*18), %r(127+i*18), %r(128+i*18), %r(129+i*18)}, [%rd(22+i*2)+0];
st.shared.v4.u32 [%r43+256], {%r144, %r145, %r146, %r147};
mul.hi.s32 %r152, %r187, -1840700269;
add.s32 %r153, %r152, %r187;
shr.u32 %r154, %r153, 31;
shr.s32 %r155, %r153, 4;
add.s32 %r156, %r155, %r154;
mul.lo.s32 %r157, %r156, 28;
sub.s32 %r158, %r187, %r157;
add.s32 %r159, %r158, %r8;
mad.lo.s32 %r160, %r156, 784, %r159;
mul.wide.s32 %rd25, %r160, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.v4.u32 {%r161, %r162, %r163, %r164}, [%rd26];
st.shared.v4.u32 [%r43+384], {%r161, %r162, %r163, %r164};
add.s32 %r188, %r188, 1024;
setp.ne.s32 %p2, %r188, 14848;
for.size.1 i in range(0, 2, 1):
add.s32 %r(13+i*173), %r187, (256+i*-224);
mov.u32 %r187, %r13;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r189, 0;
$L__BB0_6:
add.s32 %r170, %r4, %r189;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(171+i*3), %r(170+i*19), 2;
add.s32 %r(173+i*3), %r(23+i*-2), %r(171+i*3);
ld.shared.f32 %f(101+i*1), [%r(173+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(103+i*2), %f102, %f(101+i*3), %f(265+i*-1);
ld.shared.f32 %f(104+i*2), [%r173+(3584+i*3584)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(107+i*2), %f102, %f(106+i*2), %f(263+i*-1);
ld.shared.f32 %f(108+i*2), [%r173+(10752+i*-10640)];
for.size.2 i in range(0, 3, 1):
fma.rn.f32 %f(111+i*2), %f102, %f(110+i*2), %f(261+i*-1);
ld.shared.f32 %f(112+i*2), [%r173+(3696+i*3584)];
for.size.3 i in range(0, 2, 1):
for.size.2 j in range(0, 4, 1):
fma.rn.f32 %f(117+i*8+j*2), %f102, %f(116+i*8+j*2), %f(258+i*-4+j*-1);
ld.shared.f32 %f(118+i*8+j*2), [%r173+(224+i*112+j*3584)];
fma.rn.f32 %f133, %f102, %f132, %f250;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(134+i*1), [%r(173+i*3)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(265+i*-1), %f135, %f(134+i*2), %f(103+i*2);
ld.shared.f32 %f(136+i*1), [%r173+(3588+i*3584)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(263+i*-1), %f135, %f(137+i*1), %f(107+i*2);
ld.shared.f32 %f(138+i*1), [%r173+(10756+i*-10640)];
for.size.2 i in range(0, 3, 1):
fma.rn.f32 %f(261+i*-1), %f135, %f(139+i*1), %f(111+i*2);
ld.shared.f32 %f(140+i*1), [%r173+(3700+i*3584)];
for.size.3 i in range(0, 2, 1):
for.size.2 j in range(0, 4, 1):
fma.rn.f32 %f(258+i*-4+j*-1), %f135, %f(142+i*4+j*1), %f(117+i*8+j*2);
ld.shared.f32 %f(143+i*4+j*1), [%r173+(228+i*112+j*3584)];
fma.rn.f32 %f250, %f135, %f150, %f133;
add.s32 %r189, %r189, 2;
setp.ne.s32 %p3, %r189, 14;
@%p3 bra $L__BB0_6;
mov.u32 %r190, 0;
$L__BB0_8:
shl.b32 %r178, %r190, 2;
for.size.1 i in range(0, 2, 1):
add.s32 %r(180+i*1), %r(21+i*169), %r(178+i*-174);
shl.b32 %r182, %r181, 2;
add.s32 %r184, %r23, %r182;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(151+i*1), [%r(184+i*-4)+56];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(153+i*2), %f152, %f(151+i*3), %f(265+i*-1);
ld.shared.f32 %f(154+i*2), [%r184+(3640+i*3584)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(157+i*2), %f152, %f(156+i*2), %f(263+i*-1);
ld.shared.f32 %f(158+i*2), [%r184+(10808+i*-10640)];
for.size.2 i in range(0, 3, 1):
fma.rn.f32 %f(161+i*2), %f152, %f(160+i*2), %f(261+i*-1);
ld.shared.f32 %f(162+i*2), [%r184+(3752+i*3584)];
for.size.3 i in range(0, 2, 1):
for.size.2 j in range(0, 4, 1):
fma.rn.f32 %f(167+i*8+j*2), %f152, %f(166+i*8+j*2), %f(258+i*-4+j*-1);
ld.shared.f32 %f(168+i*8+j*2), [%r184+(280+i*112+j*3584)];
fma.rn.f32 %f183, %f152, %f182, %f250;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(184+i*1), [%r(184+i*-4)+60];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(265+i*-1), %f185, %f(184+i*2), %f(153+i*2);
ld.shared.f32 %f(186+i*1), [%r184+(3644+i*3584)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(263+i*-1), %f185, %f(187+i*1), %f(157+i*2);
ld.shared.f32 %f(188+i*1), [%r184+(10812+i*-10640)];
for.size.2 i in range(0, 3, 1):
fma.rn.f32 %f(261+i*-1), %f185, %f(189+i*1), %f(161+i*2);
ld.shared.f32 %f(190+i*1), [%r184+(3756+i*3584)];
for.size.3 i in range(0, 2, 1):
for.size.2 j in range(0, 4, 1):
fma.rn.f32 %f(258+i*-4+j*-1), %f185, %f(192+i*4+j*1), %f(167+i*8+j*2);
ld.shared.f32 %f(193+i*4+j*1), [%r184+(284+i*112+j*3584)];
fma.rn.f32 %f250, %f185, %f200, %f183;
add.s32 %r190, %r190, 2;
setp.ne.s32 %p4, %r190, 14;
@%p4 bra $L__BB0_8;
add.s32 %r185, %r185, 1;
setp.lt.u32 %p5, %r185, 28;
@%p5 bra $L__BB0_1;
mul.wide.s32 %rd27, %r3, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f201, [%rd28];
add.f32 %f202, %f265, %f201;
mov.f32 %f203, 0f00000000;
max.f32 %f204, %f202, %f203;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd29+(0+i*128)], %f(204+i*3);
ld.global.nc.f32 %f(205+i*3), [%rd28+(128+i*128)];
add.f32 %f(206+i*3), %f(264+i*-1), %f(205+i*3);
max.f32 %f(207+i*3), %f(206+i*3), %f203;
for.size.10 i in range(0, 3, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(384+i*4+j*-380)], %f(213+i*12+j*3);
ld.global.nc.f32 %f(214+i*12+j*3), [%rd28+(4+i*4+j*128)];
add.f32 %f(215+i*12+j*3), %f(261+i*-4+j*-1), %f(214+i*12+j*3);
max.f32 %f(216+i*12+j*3), %f(215+i*12+j*3), %f203;
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(132+i*4+j*128)], %f(219+i*12+j*3);
ld.global.nc.f32 %f(220+i*12+j*3), [%rd28+(260+i*4+j*128)];
add.f32 %f(221+i*12+j*3), %f(259+i*-4+j*-1), %f(220+i*12+j*3);
max.f32 %f(222+i*12+j*3), %f(221+i*12+j*3), %f203;
st.global.f32 [%rd29+396], %f249;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[28];
__shared__ float placeholder_shared[3584];
T_dense[0] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 28; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 28 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 112; ++ax0_ax1_fused_outer_outer) {
((float4*)(placeholder_shared + ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 4))[0] = ((float4*)(placeholder1 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 4) / 28 * 784 + k_outer_outer * 28 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 4) % 28))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 14; ++k_inner) {
int index2 = k_outer_inner * 14 + k_inner;
int index3 = (int)threadIdx.x * 112 + k_outer_inner * 14 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 896]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 1792]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 2688]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 28]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 924]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 1820]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 2716]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 56]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 952]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 1848]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 2744]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 84]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 980]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 1876]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 2772]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index4 = (int)threadIdx.x * 4 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 32] = max(T_dense[index5 + 4] + placeholder2[index4 + 32], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[index5 + 8] + placeholder2[index4 + 64], 0.000000e+00f);
T_relu[index4 + 96] = max(T_dense[index5 + 12] + placeholder2[index4 + 96], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp4mo41uxm | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<141>;
.reg .b32 %r<17>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[56];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[7168];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd11, [default_function_kernel0_param_1];
ld.param.u64 %rd12, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd11;
cvta.to.global.u64 %rd1, %rd12;
mov.u32 %r7, %tid.x;
shl.b32 %r8, %r7, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r1, %r9, %r8;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(10+i*1), %r7, (3136+i*-3080);
shl.b32 %r12, %r11, 2;
mov.u32 %r13, _ZZ24default_function_kernel0E18placeholder_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(2+i*1), %r13, %r(12+i*-1);
cvt.s64.s32 %rd2, %r7;
cvta.to.global.u64 %rd14, %rd10;
for.size.2 i in range(0, 2, 1):
mul.wide.s32 %rd(15+i*1), %r(7+i*3), 4;
add.s64 %rd(23+i*-6), %rd(14+i*-1), %rd(15+i*1);
add.s64 %rd22, %rd17, 6272;
mov.f32 %f139, 0f00000000;
mov.u32 %r16, 0;
mov.f32 %f140, %f139;
$L__BB0_1:
cvt.u32.u64 %r14, %rd2;
setp.gt.s32 %p1, %r14, 13;
bar.sync 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f7, [%rd23];
st.shared.f32 [%r1], %f7;
$L__BB0_3:
setp.gt.s32 %p2, %r14, 31;
@%p2 bra $L__BB0_5;
for.size.3 i in range(0, 4, 1):
for.size.2 j in range(0, 14, 1):
ld.global.nc.f32 %f(8+i*14+j*1), [%rd22+(-6272+i*3136+j*4)];
st.shared.f32 [%r2+(0+i*56+j*4)], %f(8+i*14+j*1);
$L__BB0_5:
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(64+i*5), [%r3+(0+i*4)];
ld.shared.f32 %f(65+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(66+i*5), %f(65+i*5), %f(64+i*5), %f(139+i*-73);
ld.shared.f32 %f(67+i*5), [%r3+(3584+i*4)];
fma.rn.f32 %f(68+i*5), %f(65+i*5), %f(67+i*5), %f(140+i*-72);
for.size.5 i in range(0, 11, 1):
ld.shared.f32 %f(74+i*5), [%r3+(8+i*4)];
ld.shared.f32 %f(75+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(76+i*5), %f(75+i*5), %f(74+i*5), %f(71+i*5);
ld.shared.f32 %f(77+i*5), [%r3+(3592+i*4)];
fma.rn.f32 %f(78+i*5), %f(75+i*5), %f(77+i*5), %f(73+i*5);
ld.shared.f32 %f129, [%r3+52];
ld.shared.f32 %f130, [_ZZ24default_function_kernel0E20placeholder_d_shared+52];
fma.rn.f32 %f139, %f130, %f129, %f126;
ld.shared.f32 %f131, [%r3+3636];
fma.rn.f32 %f140, %f130, %f131, %f128;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(23+i*-1), %rd(23+i*-1), 56;
add.s32 %r16, %r16, 1;
setp.ne.s32 %p3, %r16, 56;
@%p3 bra $L__BB0_1;
cvta.to.global.u64 %rd18, %rd9;
shl.b64 %rd19, %rd2, 2;
add.s64 %rd20, %rd18, %rd19;
ld.global.nc.f32 %f132, [%rd20];
add.f32 %f133, %f139, %f132;
mov.f32 %f134, 0f00000000;
max.f32 %f135, %f133, %f134;
add.s64 %rd21, %rd1, %rd19;
st.global.f32 [%rd21], %f135;
ld.global.nc.f32 %f136, [%rd20+256];
add.f32 %f137, %f140, %f136;
max.f32 %f138, %f137, %f134;
st.global.f32 [%rd21+256], %f138;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[14];
__shared__ float placeholder_shared[1792];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 56; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 14 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
if ((int)threadIdx.x < 32) {
int index0 = (int)threadIdx.x;
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
placeholder_shared[index0 * 56] = placeholder1[index2];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index3 = (int)threadIdx.x * 56 + 1;
placeholder_shared[index3] = placeholder1[index2 + 1];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index4 = (int)threadIdx.x * 56 + 2;
placeholder_shared[index4] = placeholder1[index2 + 2];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index5 = (int)threadIdx.x * 56 + 3;
placeholder_shared[index5] = placeholder1[index2 + 3];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index6 = (int)threadIdx.x * 56 + 4;
placeholder_shared[index6] = placeholder1[index2 + 4];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index7 = (int)threadIdx.x * 56 + 5;
placeholder_shared[index7] = placeholder1[index2 + 5];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index8 = (int)threadIdx.x * 56 + 6;
placeholder_shared[index8] = placeholder1[index2 + 6];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index9 = (int)threadIdx.x * 56 + 7;
placeholder_shared[index9] = placeholder1[index2 + 7];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index10 = (int)threadIdx.x * 56 + 8;
placeholder_shared[index10] = placeholder1[index2 + 8];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index11 = (int)threadIdx.x * 56 + 9;
placeholder_shared[index11] = placeholder1[index2 + 9];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index12 = (int)threadIdx.x * 56 + 10;
placeholder_shared[index12] = placeholder1[index2 + 10];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index13 = (int)threadIdx.x * 56 + 11;
placeholder_shared[index13] = placeholder1[index2 + 11];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index14 = (int)threadIdx.x * 56 + 12;
placeholder_shared[index14] = placeholder1[index2 + 12];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index15 = (int)threadIdx.x * 56 + 13;
placeholder_shared[index15] = placeholder1[index2 + 13];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index16 = (int)threadIdx.x * 56 + 14;
placeholder_shared[index16] = placeholder1[index2 + 784];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index17 = (int)threadIdx.x * 56 + 15;
placeholder_shared[index17] = placeholder1[index2 + 785];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index18 = (int)threadIdx.x * 56 + 16;
placeholder_shared[index18] = placeholder1[index2 + 786];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index19 = (int)threadIdx.x * 56 + 17;
placeholder_shared[index19] = placeholder1[index2 + 787];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index20 = (int)threadIdx.x * 56 + 18;
placeholder_shared[index20] = placeholder1[index2 + 788];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index21 = (int)threadIdx.x * 56 + 19;
placeholder_shared[index21] = placeholder1[index2 + 789];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index22 = (int)threadIdx.x * 56 + 20;
placeholder_shared[index22] = placeholder1[index2 + 790];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index23 = (int)threadIdx.x * 56 + 21;
placeholder_shared[index23] = placeholder1[index2 + 791];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index24 = (int)threadIdx.x * 56 + 22;
placeholder_shared[index24] = placeholder1[index2 + 792];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index25 = (int)threadIdx.x * 56 + 23;
placeholder_shared[index25] = placeholder1[index2 + 793];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index26 = (int)threadIdx.x * 56 + 24;
placeholder_shared[index26] = placeholder1[index2 + 794];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index27 = (int)threadIdx.x * 56 + 25;
placeholder_shared[index27] = placeholder1[index2 + 795];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index28 = (int)threadIdx.x * 56 + 26;
placeholder_shared[index28] = placeholder1[index2 + 796];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index29 = (int)threadIdx.x * 56 + 27;
placeholder_shared[index29] = placeholder1[index2 + 797];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index30 = (int)threadIdx.x * 56 + 28;
placeholder_shared[index30] = placeholder1[index2 + 1568];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index31 = (int)threadIdx.x * 56 + 29;
placeholder_shared[index31] = placeholder1[index2 + 1569];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index32 = (int)threadIdx.x * 56 + 30;
placeholder_shared[index32] = placeholder1[index2 + 1570];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index33 = (int)threadIdx.x * 56 + 31;
placeholder_shared[index33] = placeholder1[index2 + 1571];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index34 = (int)threadIdx.x * 56 + 32;
placeholder_shared[index34] = placeholder1[index2 + 1572];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index35 = (int)threadIdx.x * 56 + 33;
placeholder_shared[index35] = placeholder1[index2 + 1573];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index36 = (int)threadIdx.x * 56 + 34;
placeholder_shared[index36] = placeholder1[index2 + 1574];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index37 = (int)threadIdx.x * 56 + 35;
placeholder_shared[index37] = placeholder1[index2 + 1575];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index38 = (int)threadIdx.x * 56 + 36;
placeholder_shared[index38] = placeholder1[index2 + 1576];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index39 = (int)threadIdx.x * 56 + 37;
placeholder_shared[index39] = placeholder1[index2 + 1577];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index40 = (int)threadIdx.x * 56 + 38;
placeholder_shared[index40] = placeholder1[index2 + 1578];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index41 = (int)threadIdx.x * 56 + 39;
placeholder_shared[index41] = placeholder1[index2 + 1579];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index42 = (int)threadIdx.x * 56 + 40;
placeholder_shared[index42] = placeholder1[index2 + 1580];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index43 = (int)threadIdx.x * 56 + 41;
placeholder_shared[index43] = placeholder1[index2 + 1581];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index44 = (int)threadIdx.x * 56 + 42;
placeholder_shared[index44] = placeholder1[index2 + 2352];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index45 = (int)threadIdx.x * 56 + 43;
placeholder_shared[index45] = placeholder1[index2 + 2353];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index46 = (int)threadIdx.x * 56 + 44;
placeholder_shared[index46] = placeholder1[index2 + 2354];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index47 = (int)threadIdx.x * 56 + 45;
placeholder_shared[index47] = placeholder1[index2 + 2355];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index48 = (int)threadIdx.x * 56 + 46;
placeholder_shared[index48] = placeholder1[index2 + 2356];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index49 = (int)threadIdx.x * 56 + 47;
placeholder_shared[index49] = placeholder1[index2 + 2357];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index50 = (int)threadIdx.x * 56 + 48;
placeholder_shared[index50] = placeholder1[index2 + 2358];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index51 = (int)threadIdx.x * 56 + 49;
placeholder_shared[index51] = placeholder1[index2 + 2359];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index52 = (int)threadIdx.x * 56 + 50;
placeholder_shared[index52] = placeholder1[index2 + 2360];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index53 = (int)threadIdx.x * 56 + 51;
placeholder_shared[index53] = placeholder1[index2 + 2361];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index54 = (int)threadIdx.x * 56 + 52;
placeholder_shared[index54] = placeholder1[index2 + 2362];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index55 = (int)threadIdx.x * 56 + 53;
placeholder_shared[index55] = placeholder1[index2 + 2363];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index56 = (int)threadIdx.x * 56 + 54;
placeholder_shared[index56] = placeholder1[index2 + 2364];
}
if ((int)threadIdx.x < 32) {
int index2 = (int)threadIdx.x * 3136 + k_outer_outer * 14;
int index57 = (int)threadIdx.x * 56 + 55;
placeholder_shared[index57] = placeholder1[index2 + 2365];
}
__syncthreads();
int index0 = (int)threadIdx.x;
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 14];
int index58 = (int)threadIdx.x * 14 + 896;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index58];
int index59 = (int)threadIdx.x * 14 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index59];
int index60 = (int)threadIdx.x * 14 + 897;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index60];
int index61 = (int)threadIdx.x * 14 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index61];
int index62 = (int)threadIdx.x * 14 + 898;
T_dense[1] = T_dense[1] + placeholder_d_shared[2] * placeholder_shared[index62];
int index63 = (int)threadIdx.x * 14 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index63];
int index64 = (int)threadIdx.x * 14 + 899;
T_dense[1] = T_dense[1] + placeholder_d_shared[3] * placeholder_shared[index64];
int index65 = (int)threadIdx.x * 14 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index65];
int index66 = (int)threadIdx.x * 14 + 900;
T_dense[1] = T_dense[1] + placeholder_d_shared[4] * placeholder_shared[index66];
int index67 = (int)threadIdx.x * 14 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index67];
int index68 = (int)threadIdx.x * 14 + 901;
T_dense[1] = T_dense[1] + placeholder_d_shared[5] * placeholder_shared[index68];
int index69 = (int)threadIdx.x * 14 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index69];
int index70 = (int)threadIdx.x * 14 + 902;
T_dense[1] = T_dense[1] + placeholder_d_shared[6] * placeholder_shared[index70];
int index71 = (int)threadIdx.x * 14 + 7;
T_dense[0] = T_dense[0] + placeholder_d_shared[7] * placeholder_shared[index71];
int index72 = (int)threadIdx.x * 14 + 903;
T_dense[1] = T_dense[1] + placeholder_d_shared[7] * placeholder_shared[index72];
int index73 = (int)threadIdx.x * 14 + 8;
T_dense[0] = T_dense[0] + placeholder_d_shared[8] * placeholder_shared[index73];
int index74 = (int)threadIdx.x * 14 + 904;
T_dense[1] = T_dense[1] + placeholder_d_shared[8] * placeholder_shared[index74];
int index75 = (int)threadIdx.x * 14 + 9;
T_dense[0] = T_dense[0] + placeholder_d_shared[9] * placeholder_shared[index75];
int index76 = (int)threadIdx.x * 14 + 905;
T_dense[1] = T_dense[1] + placeholder_d_shared[9] * placeholder_shared[index76];
int index77 = (int)threadIdx.x * 14 + 10;
T_dense[0] = T_dense[0] + placeholder_d_shared[10] * placeholder_shared[index77];
int index78 = (int)threadIdx.x * 14 + 906;
T_dense[1] = T_dense[1] + placeholder_d_shared[10] * placeholder_shared[index78];
int index79 = (int)threadIdx.x * 14 + 11;
T_dense[0] = T_dense[0] + placeholder_d_shared[11] * placeholder_shared[index79];
int index80 = (int)threadIdx.x * 14 + 907;
T_dense[1] = T_dense[1] + placeholder_d_shared[11] * placeholder_shared[index80];
int index81 = (int)threadIdx.x * 14 + 12;
T_dense[0] = T_dense[0] + placeholder_d_shared[12] * placeholder_shared[index81];
int index82 = (int)threadIdx.x * 14 + 908;
T_dense[1] = T_dense[1] + placeholder_d_shared[12] * placeholder_shared[index82];
int index83 = (int)threadIdx.x * 14 + 13;
T_dense[0] = T_dense[0] + placeholder_d_shared[13] * placeholder_shared[index83];
int index84 = (int)threadIdx.x * 14 + 909;
T_dense[1] = T_dense[1] + placeholder_d_shared[13] * placeholder_shared[index84];
}
int index0 = (int)threadIdx.x;
T_relu[index0] = max(T_dense[0] + placeholder2[index0], 0.000000e+00f);
T_relu[index0 + 64] = max(T_dense[1] + placeholder2[index0 + 64], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpyqbuds92 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<142>;
.reg .b32 %r<141>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[392];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[50176];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(2+i*19), %r1, (1+i*2);
mov.u32 %r22, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r22, %r21;
mul.lo.s32 %r4, %r1, 98;
add.s32 %r5, %r2, 224;
mov.u32 %r23, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r6, %r23, %r21;
mov.f32 %f126, 0f00000000;
mov.u32 %r136, 0;
for.size.1 i in range(0, 7, 1):
mov.f32 %f(127+i*1), %f126;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r136, 98;
add.s32 %r24, %r8, %r2;
mul.wide.s32 %rd10, %r24, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.3 i in range(0, 3, 1):
for.size.2 j in range(0, 2, 1):
ld.global.nc.f32 %f(41+i*2+j*1), [%rd5+(0+i*128+j*4)];
st.shared.f32 [%r3+(0+i*128+j*4)], %f(41+i*2+j*1);
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(47+i*1), [%rd5+(384+i*4)];
st.shared.f32 [%r3+(384+i*4)], %f(47+i*1);
$L__BB0_3:
mov.u32 %r139, 512;
mov.u32 %r137, %r2;
mov.u32 %r138, %r5;
$L__BB0_4:
mul.hi.s32 %r26, %r137, 1402438301;
shr.u32 %r27, %r26, 31;
shr.s32 %r28, %r26, 5;
add.s32 %r29, %r28, %r27;
mul.lo.s32 %r30, %r29, 98;
sub.s32 %r31, %r137, %r30;
add.s32 %r32, %r31, %r8;
mad.lo.s32 %r33, %r29, 784, %r32;
mul.wide.s32 %rd11, %r33, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.v2.u32 {%r34, %r35}, [%rd12];
add.s32 %r38, %r6, %r139;
for.size.13 i in range(0, 2, 1):
st.shared.v2.u32 [%r38+(-512+i*128)], {%r(34+i*14), %r(35+i*14)};
add.s32 %r(39+i*13), %r137, (32+i*32);
mul.hi.s32 %r(40+i*13), %r(39+i*13), 1402438301;
shr.u32 %r(41+i*13), %r(40+i*13), 31;
shr.s32 %r(42+i*13), %r(40+i*13), 5;
add.s32 %r(43+i*13), %r(42+i*13), %r(41+i*13);
mul.lo.s32 %r(44+i*13), %r(43+i*13), 98;
sub.s32 %r(45+i*13), %r(39+i*13), %r(44+i*13);
add.s32 %r(46+i*13), %r(45+i*13), %r8;
mad.lo.s32 %r(47+i*13), %r(43+i*13), 784, %r(46+i*13);
mul.wide.s32 %rd(13+i*2), %r(47+i*13), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.v2.u32 {%r(48+i*13), %r(49+i*13)}, [%rd(14+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.v2.u32 [%r38+(-256+i*128)], {%r(61+i*13), %r(62+i*13)};
add.s32 %r(65+i*13), %r137, (96+i*32);
mul.hi.s32 %r(66+i*13), %r(65+i*13), 1402438301;
shr.u32 %r(67+i*13), %r(66+i*13), 31;
shr.s32 %r(68+i*13), %r(66+i*13), 5;
add.s32 %r(69+i*13), %r(68+i*13), %r(67+i*13);
mul.lo.s32 %r(70+i*13), %r(69+i*13), 98;
sub.s32 %r(71+i*13), %r(65+i*13), %r(70+i*13);
add.s32 %r(72+i*13), %r(71+i*13), %r8;
mad.lo.s32 %r(73+i*13), %r(69+i*13), 784, %r(72+i*13);
mul.wide.s32 %rd(17+i*2), %r(73+i*13), 4;
add.s64 %rd(18+i*2), %rd1, %rd(17+i*2);
ld.global.nc.v2.u32 {%r(74+i*13), %r(75+i*13)}, [%rd(18+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.v2.u32 [%r38+(0+i*128)], {%r(87+i*13), %r(88+i*13)};
add.s32 %r(91+i*13), %r138, (-64+i*32);
mul.hi.s32 %r(92+i*13), %r(91+i*13), 1402438301;
shr.u32 %r(93+i*13), %r(92+i*13), 31;
shr.s32 %r(94+i*13), %r(92+i*13), 5;
add.s32 %r(95+i*13), %r(94+i*13), %r(93+i*13);
mul.lo.s32 %r(96+i*13), %r(95+i*13), 98;
sub.s32 %r(97+i*13), %r(91+i*13), %r(96+i*13);
add.s32 %r(98+i*13), %r(97+i*13), %r8;
mad.lo.s32 %r(99+i*13), %r(95+i*13), 784, %r(98+i*13);
mul.wide.s32 %rd(21+i*2), %r(99+i*13), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.v2.u32 {%r(100+i*13), %r(101+i*13)}, [%rd(22+i*2)+0];
st.shared.v2.u32 [%r38+256], {%r113, %r114};
mul.hi.s32 %r117, %r138, 1402438301;
shr.u32 %r118, %r117, 31;
shr.s32 %r119, %r117, 5;
add.s32 %r120, %r119, %r118;
mul.lo.s32 %r121, %r120, 98;
sub.s32 %r122, %r138, %r121;
add.s32 %r123, %r122, %r8;
mad.lo.s32 %r124, %r120, 784, %r123;
mul.wide.s32 %rd25, %r124, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.v2.u32 {%r125, %r126}, [%rd26];
st.shared.v2.u32 [%r38+384], {%r125, %r126};
add.s32 %r139, %r139, 1024;
setp.ne.s32 %p2, %r139, 50688;
for.size.1 i in range(0, 2, 1):
add.s32 %r(13+i*124), %r138, (256+i*-224);
mov.u32 %r138, %r13;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r140, 0;
bra.uni $L__BB0_6;
$L__BB0_7:
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(75+i*1), [%r(17+i*-1)+8];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(77+i*2), %f76, %f(75+i*3), %f(126+i*1);
ld.shared.f32 %f(78+i*2), [%r17+(6280+i*6272)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(81+i*2), %f76, %f(80+i*2), %f(128+i*1);
ld.shared.f32 %f(82+i*2), [%r17+(18824+i*6272)];
fma.rn.f32 %f91, %f76, %f90, %f133;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(92+i*1), [%r(17+i*-1)+12];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(126+i*1), %f93, %f(92+i*2), %f(77+i*2);
ld.shared.f32 %f(94+i*1), [%r17+(6284+i*6272)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(128+i*1), %f93, %f(95+i*1), %f(81+i*2);
ld.shared.f32 %f(96+i*1), [%r17+(18828+i*6272)];
fma.rn.f32 %f133, %f93, %f100, %f91;
add.s32 %r140, %r140, 2;
$L__BB0_6:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(130+i*1), %r140, (1+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(16+i*117), %r(22+i*108), %r(131+i*-127);
shl.b32 %r134, %r133, 2;
add.s32 %r17, %r23, %r134;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(49+i*1), [%r(17+i*-1)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(51+i*2), %f50, %f(49+i*3), %f(126+i*1);
ld.shared.f32 %f(52+i*2), [%r17+(6272+i*6272)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(55+i*2), %f50, %f(54+i*2), %f(128+i*1);
ld.shared.f32 %f(56+i*2), [%r17+(18816+i*6272)];
fma.rn.f32 %f65, %f50, %f64, %f133;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(66+i*1), [%r(17+i*-1)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(126+i*1), %f67, %f(66+i*2), %f(51+i*2);
ld.shared.f32 %f(68+i*1), [%r17+(6276+i*6272)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(128+i*1), %f67, %f(69+i*1), %f(55+i*2);
ld.shared.f32 %f(70+i*1), [%r17+(18820+i*6272)];
fma.rn.f32 %f133, %f67, %f74, %f65;
setp.eq.s32 %p3, %r140, 48;
@%p3 bra $L__BB0_8;
bra.uni $L__BB0_7;
$L__BB0_8:
add.s32 %r136, %r136, 1;
setp.lt.u32 %p4, %r136, 8;
@%p4 bra $L__BB0_1;
mul.wide.s32 %rd27, %r1, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f101, [%rd28];
add.f32 %f102, %f126, %f101;
mov.f32 %f103, 0f00000000;
max.f32 %f104, %f102, %f103;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd29+(0+i*64)], %f(104+i*3);
ld.global.nc.f32 %f(105+i*3), [%rd28+(64+i*64)];
add.f32 %f(106+i*3), %f(127+i*1), %f(105+i*3);
max.f32 %f(107+i*3), %f(106+i*3), %f103;
st.global.f32 [%rd29+448], %f125;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[8];
__shared__ float placeholder_d_shared[98];
__shared__ float placeholder_shared[12544];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 8; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x * 2;
int index1 = k_outer_outer * 98 + (int)threadIdx.x * 2;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 33] = placeholder[index1 + 33];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 65] = placeholder[index1 + 65];
if ((int)threadIdx.x < 1) {
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
}
if ((int)threadIdx.x < 1) {
placeholder_d_shared[index0 + 97] = placeholder[index1 + 97];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 392; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder1 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 2) / 98 * 784 + k_outer_outer * 98 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x * 2) % 98))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 49; ++k_outer_inner) {
int index2 = k_outer_inner * 2;
int index3 = (int)threadIdx.x * 98 + k_outer_inner * 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 1568]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 3136]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 4704]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 6272]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 7840]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 9408]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 10976]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 1]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 1569]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 3137]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 4705]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 6273]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 7841]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 9409]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2 + 1] * placeholder_shared[index3 + 10977]);
}
}
int index4 = (int)threadIdx.x;
T_relu[index4] = max(T_dense[0] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 16] = max(T_dense[1] + placeholder2[index4 + 16], 0.000000e+00f);
T_relu[index4 + 32] = max(T_dense[2] + placeholder2[index4 + 32], 0.000000e+00f);
T_relu[index4 + 48] = max(T_dense[3] + placeholder2[index4 + 48], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[4] + placeholder2[index4 + 64], 0.000000e+00f);
T_relu[index4 + 80] = max(T_dense[5] + placeholder2[index4 + 80], 0.000000e+00f);
T_relu[index4 + 96] = max(T_dense[6] + placeholder2[index4 + 96], 0.000000e+00f);
T_relu[index4 + 112] = max(T_dense[7] + placeholder2[index4 + 112], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpew94g9yo | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.local .align 16 .b8 __local_depot0[256];
.reg .b64 %SP;
.reg .b64 %SPL;
.reg .pred %p<5>;
.reg .f32 %f<613>;
.reg .b32 %r<123>;
.reg .b64 %rd<37>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[112];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[14336];
mov.u64 %SPL, __local_depot0;
ld.param.u64 %rd7, [default_function_kernel0_param_0];
ld.param.u64 %rd8, [default_function_kernel0_param_1];
ld.param.u64 %rd9, [default_function_kernel0_param_2];
ld.param.u64 %rd10, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd9;
add.u64 %rd3, %SPL, 0;
mov.f32 %f29, 0f00000000;
for.size.1 i in range(0, 16, 1):
st.local.v4.f32 [%rd3+(0+i*16)], {%f29, %f29, %f29, %f29};
cvta.to.global.u64 %rd5, %rd8;
cvta.to.global.u64 %rd6, %rd10;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 1792;
shl.b32 %r20, %r1, 2;
mov.u32 %r21, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r21, %r20;
add.s32 %r4, %r1, 14;
mov.u32 %r22, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r22, %r20;
mov.u32 %r117, 0;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r7, %r117, 28;
add.s32 %r24, %r7, %r1;
mul.wide.s32 %rd12, %r24, 4;
add.s64 %rd13, %rd1, %rd12;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(30+i*1), [%rd13+(0+i*8)];
st.shared.f32 [%r3+(0+i*8)], %f(30+i*1);
ld.global.nc.f32 %f34, [%rd13+32];
mov.u32 %r120, 32;
for.size.2 i in range(0, 9, 1):
st.shared.f32 [%r3+(32+i*8)], %f(34+i*1);
ld.global.nc.f32 %f(35+i*1), [%rd13+(40+i*8)];
st.shared.f32 [%r3+104], %f43;
mov.u32 %r118, %r1;
mov.u32 %r119, %r4;
$L__BB0_2:
mul.hi.s32 %r25, %r118, -1840700269;
add.s32 %r26, %r25, %r118;
shr.u32 %r27, %r26, 31;
shr.s32 %r28, %r26, 4;
add.s32 %r29, %r28, %r27;
mul.lo.s32 %r30, %r29, 28;
sub.s32 %r31, %r118, %r30;
add.s32 %r32, %r31, %r7;
mad.lo.s32 %r33, %r29, 784, %r32;
mul.wide.s32 %rd14, %r33, 4;
add.s64 %rd15, %rd5, %rd14;
ld.global.nc.f32 %f44, [%rd15];
add.s32 %r34, %r5, %r120;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r34+(-32+i*8)], %f(44+i*1);
add.s32 %r(35+i*10), %r118, (2+i*2);
mul.hi.s32 %r(36+i*10), %r(35+i*10), -1840700269;
add.s32 %r(37+i*10), %r(36+i*10), %r(35+i*10);
shr.u32 %r(38+i*10), %r(37+i*10), 31;
shr.s32 %r(39+i*10), %r(37+i*10), 4;
add.s32 %r(40+i*10), %r(39+i*10), %r(38+i*10);
mul.lo.s32 %r(41+i*10), %r(40+i*10), 28;
sub.s32 %r(42+i*10), %r(35+i*10), %r(41+i*10);
add.s32 %r(43+i*10), %r(42+i*10), %r7;
mad.lo.s32 %r(44+i*10), %r(40+i*10), 784, %r(43+i*10);
mul.wide.s32 %rd(16+i*2), %r(44+i*10), 4;
add.s64 %rd(17+i*2), %rd5, %rd(16+i*2);
ld.global.nc.f32 %f(45+i*1), [%rd(17+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r34+(0+i*8)], %f(48+i*1);
add.s32 %r(75+i*10), %r119, (-4+i*2);
mul.hi.s32 %r(76+i*10), %r(75+i*10), -1840700269;
add.s32 %r(77+i*10), %r(76+i*10), %r(75+i*10);
shr.u32 %r(78+i*10), %r(77+i*10), 31;
shr.s32 %r(79+i*10), %r(77+i*10), 4;
add.s32 %r(80+i*10), %r(79+i*10), %r(78+i*10);
mul.lo.s32 %r(81+i*10), %r(80+i*10), 28;
sub.s32 %r(82+i*10), %r(75+i*10), %r(81+i*10);
add.s32 %r(83+i*10), %r(82+i*10), %r7;
mad.lo.s32 %r(84+i*10), %r(80+i*10), 784, %r(83+i*10);
mul.wide.s32 %rd(24+i*2), %r(84+i*10), 4;
add.s64 %rd(25+i*2), %rd5, %rd(24+i*2);
ld.global.nc.f32 %f(49+i*1), [%rd(25+i*2)+0];
st.shared.f32 [%r34+16], %f50;
mul.hi.s32 %r95, %r119, -1840700269;
add.s32 %r96, %r95, %r119;
shr.u32 %r97, %r96, 31;
shr.s32 %r98, %r96, 4;
add.s32 %r99, %r98, %r97;
mul.lo.s32 %r100, %r99, 28;
sub.s32 %r101, %r119, %r100;
add.s32 %r102, %r101, %r7;
mad.lo.s32 %r103, %r99, 784, %r102;
mul.wide.s32 %rd28, %r103, 4;
add.s64 %rd29, %rd5, %rd28;
ld.global.nc.f32 %f51, [%rd29];
st.shared.f32 [%r34+24], %f51;
for.size.1 i in range(0, 2, 1):
add.s32 %r(119+i*-1), %r(119+i*-1), 16;
add.s32 %r120, %r120, 64;
setp.ne.s32 %p1, %r120, 14368;
@%p1 bra $L__BB0_2;
bar.sync 0;
ld.shared.f32 %f1, [_ZZ24default_function_kernel0E20placeholder_d_shared];
mov.u32 %r121, 0;
for.size.1 i in range(0, 13, 1):
ld.shared.f32 %f(2+i*1), [_ZZ24default_function_kernel0E20placeholder_d_shared+(4+i*4)];
$L__BB0_4:
mad.lo.s32 %r105, %r121, 112, %r2;
shl.b32 %r106, %r121, 2;
mul.wide.s32 %rd30, %r106, 4;
add.s64 %rd31, %rd3, %rd30;
ld.local.v4.f32 {%f52, %f53, %f54, %f55}, [%rd31];
shl.b32 %r107, %r105, 2;
add.s32 %r109, %r22, %r107;
for.size.2 i in range(0, 4, 1):
ld.shared.f32 %f(60+i*2), [%r109+(0+i*112)];
fma.rn.f32 %f(61+i*2), %f1, %f(60+i*2), %f(52+i*1);
for.size.3 i in range(0, 12, 1):
for.size.2 j in range(0, 4, 1):
ld.shared.f32 %f(68+i*8+j*2), [%r109+(4+i*4+j*112)];
fma.rn.f32 %f(69+i*8+j*2), %f(2+j*1), %f(68+i*8+j*2), %f(61+i*8+j*2);
for.size.1 i in range(0, 4, 1):
ld.shared.f32 %f(164+i*1), [%r109+(52+i*112)];
for.size.1 i in range(0, 4, 1):
fma.rn.f32 %f(168+i*1), %f14, %f(167+i*-1), %f(163+i*-2);
st.local.v4.f32 [%rd31], {%f171, %f170, %f169, %f168};
add.s32 %r121, %r121, 1;
setp.ne.s32 %p2, %r121, 16;
@%p2 bra $L__BB0_4;
ld.shared.f32 %f15, [_ZZ24default_function_kernel0E20placeholder_d_shared+56];
mov.u32 %r122, 0;
for.size.1 i in range(0, 13, 1):
ld.shared.f32 %f(16+i*1), [_ZZ24default_function_kernel0E20placeholder_d_shared+(60+i*4)];
$L__BB0_6:
mad.lo.s32 %r111, %r122, 112, %r2;
shl.b32 %r112, %r122, 2;
mul.wide.s32 %rd32, %r112, 4;
add.s64 %rd33, %rd3, %rd32;
ld.local.v4.f32 {%f172, %f173, %f174, %f175}, [%rd33];
shl.b32 %r113, %r111, 2;
add.s32 %r115, %r22, %r113;
for.size.2 i in range(0, 4, 1):
ld.shared.f32 %f(180+i*2), [%r115+(56+i*112)];
fma.rn.f32 %f(181+i*2), %f15, %f(180+i*2), %f(172+i*1);
for.size.3 i in range(0, 12, 1):
for.size.2 j in range(0, 4, 1):
ld.shared.f32 %f(188+i*8+j*2), [%r115+(60+i*4+j*112)];
fma.rn.f32 %f(189+i*8+j*2), %f(16+j*1), %f(188+i*8+j*2), %f(181+i*8+j*2);
for.size.1 i in range(0, 4, 1):
ld.shared.f32 %f(284+i*1), [%r115+(108+i*112)];
for.size.1 i in range(0, 4, 1):
fma.rn.f32 %f(288+i*1), %f28, %f(287+i*-1), %f(283+i*-2);
st.local.v4.f32 [%rd33], {%f291, %f290, %f289, %f288};
add.s32 %r122, %r122, 1;
setp.ne.s32 %p3, %r122, 16;
@%p3 bra $L__BB0_6;
add.s32 %r117, %r117, 1;
setp.lt.u32 %p4, %r117, 28;
@%p4 bra $L__BB0_1;
shl.b32 %r116, %r1, 6;
ld.local.v4.f32 {%f292, %f293, %f294, %f295}, [%rd3];
mul.wide.s32 %rd34, %r116, 4;
add.s64 %rd35, %rd6, %rd34;
ld.global.nc.f32 %f300, [%rd35];
add.f32 %f301, %f292, %f300;
mov.f32 %f302, 0f00000000;
max.f32 %f303, %f301, %f302;
add.s64 %rd36, %rd2, %rd34;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd36+(0+i*4)], %f(303+i*3);
ld.global.nc.f32 %f(304+i*3), [%rd35+(4+i*4)];
add.f32 %f(305+i*3), %f(293+i*1), %f(304+i*3);
max.f32 %f(306+i*3), %f(305+i*3), %f302;
for.size.17 i in range(0, 15, 1):
st.global.f32 [%rd36+(12+i*16)], %f(312+i*20);
ld.local.v4.f32 {%f(313+i*20), %f(314+i*20), %f(315+i*20), %f(316+i*20)}, [%rd3+(16+i*16)];
ld.global.nc.f32 %f(321+i*20), [%rd35+(16+i*16)];
add.f32 %f(322+i*20), %f(313+i*20), %f(321+i*20);
max.f32 %f(323+i*20), %f(322+i*20), %f302;
st.global.f32 [%rd36+(16+i*16)], %f(323+i*20);
ld.global.nc.f32 %f(324+i*20), [%rd35+(20+i*16)];
add.f32 %f(325+i*20), %f(314+i*20), %f(324+i*20);
max.f32 %f(326+i*20), %f(325+i*20), %f302;
st.global.f32 [%rd36+(20+i*16)], %f(326+i*20);
ld.global.nc.f32 %f(327+i*20), [%rd35+(24+i*16)];
add.f32 %f(328+i*20), %f(315+i*20), %f(327+i*20);
max.f32 %f(329+i*20), %f(328+i*20), %f302;
st.global.f32 [%rd36+(24+i*16)], %f(329+i*20);
ld.global.nc.f32 %f(330+i*20), [%rd35+(28+i*16)];
add.f32 %f(331+i*20), %f(316+i*20), %f(330+i*20);
max.f32 %f(332+i*20), %f(331+i*20), %f302;
st.global.f32 [%rd36+252], %f612;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[28];
__shared__ float placeholder_shared[3584];
for (int j_outer_inner_init = 0; j_outer_inner_init < 16; ++j_outer_inner_init) {
for (int j_inner_init = 0; j_inner_init < 4; ++j_inner_init) {
int index0 = j_outer_inner_init * 4 + j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
}
for (int k_outer_outer = 0; k_outer_outer < 28; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 14; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index2 = k_outer_outer * 28 + ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 1792; ++ax0_ax1_fused_outer_outer1) {
int index3 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
int index4 = (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) / 28 * 784 + k_outer_outer * 28 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) % 28;
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int j_outer_inner = 0; j_outer_inner < 16; ++j_outer_inner) {
for (int k_inner = 0; k_inner < 14; ++k_inner) {
for (int j_inner = 0; j_inner < 4; ++j_inner) {
int index5 = j_outer_inner * 4 + j_inner;
int index6 = k_outer_inner * 14 + k_inner;
int index7 = (int)threadIdx.x * 1792 + j_outer_inner * 112 + j_inner * 28 + k_outer_inner * 14 + k_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index8 = (int)threadIdx.x * 64 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpg3o_ab4f | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<85>;
.reg .b32 %r<114>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[100352];
ld.param.u64 %rd3, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd5, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 392;
add.s32 %r3, %r1, 224;
shl.b32 %r17, %r1, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r4, %r18, %r17;
mov.u32 %r109, 0;
mov.f32 %f77, 0f00000000;
cvta.to.global.u64 %rd7, %rd3;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(78+i*1), %f77;
$L__BB0_1:
mov.u32 %r19, %tid.x;
setp.gt.s32 %p1, %r19, 3;
bar.sync 0;
mul.lo.s32 %r6, %r109, 196;
add.s32 %r20, %r6, %r19;
mul.wide.s32 %rd8, %r20, 4;
add.s64 %rd2, %rd7, %rd8;
ld.global.nc.f32 %f17, [%rd2];
shl.b32 %r21, %r19, 2;
mov.u32 %r22, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r23, %r22, %r21;
for.size.2 i in range(0, 5, 1):
st.shared.f32 [%r23+(0+i*128)], %f(17+i*1);
ld.global.nc.f32 %f(18+i*1), [%rd2+(128+i*128)];
st.shared.f32 [%r23+640], %f22;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f23, [%rd2+768];
st.shared.f32 [%r23+768], %f23;
$L__BB0_3:
mov.u32 %r112, 512;
mov.u32 %r110, %r1;
mov.u32 %r111, %r3;
$L__BB0_4:
mul.hi.s32 %r29, %r110, 1402438301;
shr.u32 %r30, %r29, 31;
shr.s32 %r31, %r29, 6;
add.s32 %r32, %r31, %r30;
mul.lo.s32 %r33, %r32, 196;
sub.s32 %r34, %r110, %r33;
add.s32 %r35, %r34, %r6;
mad.lo.s32 %r36, %r32, 784, %r35;
mul.wide.s32 %rd9, %r36, 4;
add.s64 %rd10, %rd1, %rd9;
ld.global.nc.f32 %f24, [%rd10];
add.s32 %r37, %r4, %r112;
for.size.13 i in range(0, 4, 1):
st.shared.f32 [%r37+(-512+i*128)], %f(24+i*1);
add.s32 %r(38+i*9), %r110, (32+i*32);
mul.hi.s32 %r(39+i*9), %r(38+i*9), 1402438301;
shr.u32 %r(40+i*9), %r(39+i*9), 31;
shr.s32 %r(41+i*9), %r(39+i*9), 6;
add.s32 %r(42+i*9), %r(41+i*9), %r(40+i*9);
mul.lo.s32 %r(43+i*9), %r(42+i*9), 196;
sub.s32 %r(44+i*9), %r(38+i*9), %r(43+i*9);
add.s32 %r(45+i*9), %r(44+i*9), %r6;
mad.lo.s32 %r(46+i*9), %r(42+i*9), 784, %r(45+i*9);
mul.wide.s32 %rd(11+i*2), %r(46+i*9), 4;
add.s64 %rd(12+i*2), %rd1, %rd(11+i*2);
ld.global.nc.f32 %f(25+i*1), [%rd(12+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.f32 [%r37+(0+i*128)], %f(28+i*1);
add.s32 %r(74+i*9), %r111, (-64+i*32);
mul.hi.s32 %r(75+i*9), %r(74+i*9), 1402438301;
shr.u32 %r(76+i*9), %r(75+i*9), 31;
shr.s32 %r(77+i*9), %r(75+i*9), 6;
add.s32 %r(78+i*9), %r(77+i*9), %r(76+i*9);
mul.lo.s32 %r(79+i*9), %r(78+i*9), 196;
sub.s32 %r(80+i*9), %r(74+i*9), %r(79+i*9);
add.s32 %r(81+i*9), %r(80+i*9), %r6;
mad.lo.s32 %r(82+i*9), %r(78+i*9), 784, %r(81+i*9);
mul.wide.s32 %rd(19+i*2), %r(82+i*9), 4;
add.s64 %rd(20+i*2), %rd1, %rd(19+i*2);
ld.global.nc.f32 %f(29+i*1), [%rd(20+i*2)+0];
st.shared.f32 [%r37+256], %f30;
mul.hi.s32 %r92, %r111, 1402438301;
shr.u32 %r93, %r92, 31;
shr.s32 %r94, %r92, 6;
add.s32 %r95, %r94, %r93;
mul.lo.s32 %r96, %r95, 196;
sub.s32 %r97, %r111, %r96;
add.s32 %r98, %r97, %r6;
mad.lo.s32 %r99, %r95, 784, %r98;
mul.wide.s32 %rd23, %r99, 4;
add.s64 %rd24, %rd1, %rd23;
ld.global.nc.f32 %f31, [%rd24];
st.shared.f32 [%r37+384], %f31;
for.size.1 i in range(0, 2, 1):
add.s32 %r(111+i*-1), %r(111+i*-1), 256;
add.s32 %r112, %r112, 1024;
setp.ne.s32 %p2, %r112, 100864;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r113, 0;
$L__BB0_6:
add.s32 %r101, %r113, %r2;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(102+i*3), %r(101+i*12), 2;
add.s32 %r(104+i*3), %r(18+i*4), %r(102+i*3);
ld.shared.f32 %f(32+i*1), [%r(104+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(34+i*2), %f33, %f(32+i*3), %f(80+i*-1);
ld.shared.f32 %f(35+i*2), [%r104+(50176+i*-49392)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(38+i*2), %f33, %f(37+i*2), %f(78+i*-1);
ld.shared.f32 %f(39+i*2), [%r104+(50960+i*-50956)];
for.size.9 i in range(0, 2, 1):
ld.shared.f32 %f(42+i*9), [%r107+(4+i*4)];
fma.rn.f32 %f(43+i*9), %f(42+i*9), %f(41+i*9), %f(34+i*9);
ld.shared.f32 %f(44+i*9), [%r104+(50180+i*4)];
fma.rn.f32 %f(45+i*9), %f(42+i*9), %f(44+i*9), %f(36+i*9);
ld.shared.f32 %f(46+i*9), [%r104+(788+i*4)];
fma.rn.f32 %f(47+i*9), %f(42+i*9), %f(46+i*9), %f(38+i*9);
ld.shared.f32 %f(48+i*9), [%r104+(50964+i*4)];
fma.rn.f32 %f(49+i*9), %f(42+i*9), %f(48+i*9), %f(40+i*9);
ld.shared.f32 %f(50+i*9), [%r104+(8+i*4)];
ld.shared.f32 %f60, [%r107+12];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(80+i*-1), %f60, %f(59+i*2), %f(52+i*2);
ld.shared.f32 %f(61+i*1), [%r104+(50188+i*-49392)];
fma.rn.f32 %f78, %f60, %f62, %f56;
ld.shared.f32 %f63, [%r104+50972];
fma.rn.f32 %f77, %f60, %f63, %f58;
add.s32 %r113, %r113, 4;
setp.ne.s32 %p3, %r113, 196;
@%p3 bra $L__BB0_6;
add.s32 %r109, %r109, 1;
setp.lt.u32 %p4, %r109, 4;
@%p4 bra $L__BB0_1;
shl.b32 %r108, %r1, 1;
cvta.to.global.u64 %rd25, %rd5;
mul.wide.s32 %rd26, %r108, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f64, [%rd27];
add.f32 %f65, %f80, %f64;
mov.f32 %f66, 0f00000000;
max.f32 %f67, %f65, %f66;
cvta.to.global.u64 %rd28, %rd4;
add.s64 %rd29, %rd28, %rd26;
for.size.4 i in range(0, 2, 1):
st.global.f32 [%rd29+(0+i*256)], %f(67+i*3);
ld.global.nc.f32 %f(68+i*3), [%rd27+(256+i*-252)];
add.f32 %f(69+i*3), %f(79+i*-1), %f(68+i*3);
max.f32 %f(70+i*3), %f(69+i*3), %f66;
st.global.f32 [%rd29+4], %f73;
ld.global.nc.f32 %f74, [%rd27+260];
add.f32 %f75, %f77, %f74;
max.f32 %f76, %f75, %f66;
st.global.f32 [%rd29+260], %f76;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[25088];
T_dense[0] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 196 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
placeholder_d_shared[index0 + 128] = placeholder[index1 + 128];
placeholder_d_shared[index0 + 160] = placeholder[index1 + 160];
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 192] = placeholder[index1 + 192];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 784; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) % 196;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 392 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 12544]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 196]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 12740]);
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index6 = (int)threadIdx.x * 2 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 64] = max(T_dense[index7 + 2] + placeholder2[index6 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp9cnr_h9l | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<86>;
.reg .b32 %r<29>;
.reg .b64 %rd<21>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[401408];
ld.param.u64 %rd11, [default_function_kernel0_param_0];
ld.param.u64 %rd8, [default_function_kernel0_param_1];
ld.param.u64 %rd9, [default_function_kernel0_param_2];
ld.param.u64 %rd10, [default_function_kernel0_param_3];
mov.u32 %r1, %tid.x;
shl.b32 %r13, %r1, 2;
cvta.to.global.u64 %rd12, %rd11;
mul.wide.s32 %rd13, %r13, 4;
add.s64 %rd1, %rd12, %rd13;
ld.global.nc.f32 %f5, [%rd1];
shl.b32 %r14, %r1, 4;
mov.u32 %r15, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r15, %r14;
for.size.2 i in range(0, 3, 1):
st.shared.f32 [%r2+(0+i*4)], %f(5+i*1);
ld.global.nc.f32 %f(6+i*1), [%rd1+(4+i*4)];
for.size.6 i in range(0, 2, 1):
for.size.2 j in range(0, 2, 1):
st.shared.f32 [%r2+(12+i*1024+j*1012)], %f(8+i*4+j*1);
ld.global.nc.f32 %f(9+i*4+j*1), [%rd1+(1024+i*1024+j*4)];
for.size.2 j in range(0, 2, 1):
st.shared.f32 [%r2+(1028+i*1024+j*4)], %f(10+i*4+j*1);
ld.global.nc.f32 %f(11+i*4+j*1), [%rd1+(1032+i*1024+j*4)];
st.shared.f32 [%r2+2060], %f16;
setp.gt.s32 %p1, %r1, 3;
@%p1 bra $L__BB0_2;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(17+i*1), [%rd1+(3072+i*4)];
st.shared.f32 [%r2+(3072+i*4)], %f(17+i*1);
$L__BB0_2:
mov.u32 %r18, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r19, %r18, %r13;
add.s32 %r25, %r19, 2048;
cvt.s64.s32 %rd2, %r1;
cvta.to.global.u64 %rd14, %rd8;
mul.wide.s32 %rd15, %r1, 4;
add.s64 %rd16, %rd14, %rd15;
add.s64 %rd20, %rd16, 2048;
cvta.to.global.u64 %rd4, %rd10;
cvta.to.global.u64 %rd5, %rd9;
mov.u32 %r26, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(21+i*1), [%rd20+(-2048+i*256)];
st.shared.f32 [%r25+(-2048+i*256)], %f(21+i*1);
add.s32 %r25, %r25, 4096;
add.s64 %rd20, %rd20, 4096;
add.s32 %r26, %r26, 16;
setp.ne.s32 %p2, %r26, 1568;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r22, %r1, 3136, %r18;
add.s32 %r27, %r22, 200732;
mov.f32 %f84, 0f00000000;
mov.u32 %r28, 200704;
mov.f32 %f85, %f84;
$L__BB0_5:
add.s32 %r24, %r15, %r28;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(39+i*5), [%r27+(-200732+i*4)];
ld.shared.f32 %f(40+i*5), [%r24+(-200704+i*4)];
fma.rn.f32 %f(41+i*5), %f(40+i*5), %f(39+i*5), %f(84+i*-43);
ld.shared.f32 %f(42+i*5), [%r27+(-28+i*4)];
fma.rn.f32 %f(43+i*5), %f(40+i*5), %f(42+i*5), %f(85+i*-42);
for.size.5 i in range(0, 5, 1):
ld.shared.f32 %f(49+i*5), [%r27+(-200724+i*4)];
ld.shared.f32 %f(50+i*5), [%r24+(-200696+i*4)];
fma.rn.f32 %f(51+i*5), %f(50+i*5), %f(49+i*5), %f(46+i*5);
ld.shared.f32 %f(52+i*5), [%r27+(-20+i*4)];
fma.rn.f32 %f(53+i*5), %f(50+i*5), %f(52+i*5), %f(48+i*5);
ld.shared.f32 %f74, [%r27+-200704];
ld.shared.f32 %f75, [%r24+-200676];
fma.rn.f32 %f84, %f75, %f74, %f71;
ld.shared.f32 %f76, [%r27];
fma.rn.f32 %f85, %f75, %f76, %f73;
for.size.1 i in range(0, 2, 1):
add.s32 %r(27+i*1), %r(27+i*1), 32;
setp.ne.s32 %p3, %r28, 203840;
@%p3 bra $L__BB0_5;
shl.b64 %rd17, %rd2, 2;
add.s64 %rd18, %rd4, %rd17;
ld.global.nc.f32 %f77, [%rd18];
add.f32 %f78, %f84, %f77;
mov.f32 %f79, 0f00000000;
max.f32 %f80, %f78, %f79;
add.s64 %rd19, %rd5, %rd17;
st.global.f32 [%rd19], %f80;
ld.global.nc.f32 %f81, [%rd18+256];
add.f32 %f82, %f85, %f81;
max.f32 %f83, %f82, %f79;
st.global.f32 [%rd19+256], %f83;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[100352];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
int index0 = (int)threadIdx.x * 4;
placeholder_d_shared[index0] = placeholder[index0];
placeholder_d_shared[index0 + 1] = placeholder[index0 + 1];
placeholder_d_shared[index0 + 2] = placeholder[index0 + 2];
placeholder_d_shared[index0 + 3] = placeholder[index0 + 3];
placeholder_d_shared[index0 + 256] = placeholder[index0 + 256];
placeholder_d_shared[index0 + 257] = placeholder[index0 + 257];
placeholder_d_shared[index0 + 258] = placeholder[index0 + 258];
placeholder_d_shared[index0 + 259] = placeholder[index0 + 259];
placeholder_d_shared[index0 + 512] = placeholder[index0 + 512];
placeholder_d_shared[index0 + 513] = placeholder[index0 + 513];
placeholder_d_shared[index0 + 514] = placeholder[index0 + 514];
placeholder_d_shared[index0 + 515] = placeholder[index0 + 515];
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 768] = placeholder[index0 + 768];
}
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 769] = placeholder[index0 + 769];
}
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 770] = placeholder[index0 + 770];
}
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 771] = placeholder[index0 + 771];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1568; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 64 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index1];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 784; ++k_outer_inner) {
int index2 = k_outer_inner;
int index3 = (int)threadIdx.x * 784 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 50176]);
}
int index4 = (int)threadIdx.x;
T_relu[index4] = max(T_dense[0] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[1] + placeholder2[index4 + 64], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpt6tw5g1s | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<96>;
.reg .b32 %r<32>;
.reg .b64 %rd<20>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
mov.u32 %r1, %tid.x;
shl.b32 %r2, %r1, 1;
cvta.to.global.u64 %rd11, %rd10;
mul.wide.s32 %rd12, %r2, 4;
add.s64 %rd1, %rd11, %rd12;
ld.global.nc.f32 %f5, [%rd1];
shl.b32 %r15, %r1, 3;
mov.u32 %r30, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r30, %r15;
for.size.3 i in range(0, 11, 1):
for.size.2 j in range(0, 2, 1):
st.shared.f32 [%r3+(0+i*256+j*4)], %f(5+i*2+j*1);
ld.global.nc.f32 %f(6+i*2+j*1), [%rd1+(4+i*256+j*252)];
st.shared.f32 [%r3+2816], %f27;
ld.global.nc.f32 %f28, [%rd1+2820];
st.shared.f32 [%r3+2820], %f28;
setp.gt.s32 %p1, %r1, 7;
@%p1 bra $L__BB0_2;
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(29+i*1), [%rd1+(3072+i*4)];
st.shared.f32 [%r3+(3072+i*4)], %f(29+i*1);
$L__BB0_2:
mov.u32 %r4, %ctaid.x;
mad.lo.s32 %r18, %r4, 50176, %r1;
cvta.to.global.u64 %rd13, %rd7;
mul.wide.s32 %rd14, %r18, 4;
add.s64 %rd15, %rd13, %rd14;
add.s64 %rd19, %rd15, 1024;
shl.b32 %r19, %r1, 2;
mov.u32 %r20, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r21, %r20, %r19;
add.s32 %r28, %r21, 1024;
cvta.to.global.u64 %rd3, %rd9;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r29, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(31+i*1), [%rd19+(-1024+i*128)];
st.shared.f32 [%r28+(-1024+i*128)], %f(31+i*1);
add.s64 %rd19, %rd19, 2048;
for.size.1 i in range(0, 2, 1):
add.s32 %r(28+i*1), %r(28+i*1), (2048+i*-2032);
setp.ne.s32 %p2, %r29, 1568;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r10, %r1, 6272, %r20;
mov.f32 %f94, 0f00000000;
mov.u32 %r31, 3136;
mov.f32 %f95, %f94;
$L__BB0_5:
add.s32 %r25, %r10, %r31;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(49+i*3), [%r25+(-3136+i*4)];
ld.shared.f32 %f(50+i*3), [%r30+(0+i*4)];
fma.rn.f32 %f(51+i*3), %f(50+i*3), %f(49+i*3), %f(95+i*-44);
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(55+i*3), [%r25+(-3128+i*4)];
ld.shared.f32 %f(56+i*3), [%r30+(8+i*4)];
fma.rn.f32 %f(57+i*3), %f(56+i*3), %f(55+i*3), %f(54+i*3);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(61+i*2), [%r25+(0+i*4)];
fma.rn.f32 %f(62+i*2), %f(50+i*3), %f(61+i*2), %f(94+i*-32);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(65+i*2), [%r25+(8+i*4)];
fma.rn.f32 %f(66+i*2), %f(56+i*3), %f(65+i*2), %f(64+i*2);
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(69+i*3), [%r25+(-3120+i*4)];
ld.shared.f32 %f(70+i*3), [%r30+(16+i*4)];
fma.rn.f32 %f(71+i*3), %f(70+i*3), %f(69+i*3), %f(60+i*11);
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(75+i*3), [%r25+(-3112+i*4)];
ld.shared.f32 %f(76+i*3), [%r30+(24+i*4)];
fma.rn.f32 %f(77+i*18), %f(76+i*3), %f(75+i*3), %f(74+i*3);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(80+i*2), [%r25+(16+i*4)];
fma.rn.f32 %f(81+i*2), %f(70+i*3), %f(80+i*2), %f(68+i*13);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(84+i*2), [%r25+(24+i*4)];
fma.rn.f32 %f(85+i*9), %f(76+i*3), %f(84+i*2), %f(83+i*2);
for.size.1 i in range(0, 2, 1):
add.s32 %r(30+i*1), %r(30+i*1), 32;
setp.ne.s32 %p3, %r31, 6272;
@%p3 bra $L__BB0_5;
shl.b32 %r26, %r4, 6;
add.s32 %r27, %r26, %r2;
mul.wide.s32 %rd16, %r27, 4;
add.s64 %rd17, %rd3, %rd16;
ld.global.nc.f32 %f87, [%rd17];
add.f32 %f88, %f95, %f87;
mov.f32 %f89, 0f00000000;
max.f32 %f90, %f88, %f89;
add.s64 %rd18, %rd4, %rd16;
st.global.f32 [%rd18], %f90;
ld.global.nc.f32 %f91, [%rd17+4];
add.f32 %f92, %f94, %f91;
max.f32 %f93, %f92, %f89;
st.global.f32 [%rd18+4], %f93;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[50176];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
int index0 = (int)threadIdx.x * 2;
placeholder_d_shared[index0] = placeholder[index0];
placeholder_d_shared[index0 + 1] = placeholder[index0 + 1];
placeholder_d_shared[index0 + 64] = placeholder[index0 + 64];
placeholder_d_shared[index0 + 65] = placeholder[index0 + 65];
placeholder_d_shared[index0 + 128] = placeholder[index0 + 128];
placeholder_d_shared[index0 + 129] = placeholder[index0 + 129];
placeholder_d_shared[index0 + 192] = placeholder[index0 + 192];
placeholder_d_shared[index0 + 193] = placeholder[index0 + 193];
placeholder_d_shared[index0 + 256] = placeholder[index0 + 256];
placeholder_d_shared[index0 + 257] = placeholder[index0 + 257];
placeholder_d_shared[index0 + 320] = placeholder[index0 + 320];
placeholder_d_shared[index0 + 321] = placeholder[index0 + 321];
placeholder_d_shared[index0 + 384] = placeholder[index0 + 384];
placeholder_d_shared[index0 + 385] = placeholder[index0 + 385];
placeholder_d_shared[index0 + 448] = placeholder[index0 + 448];
placeholder_d_shared[index0 + 449] = placeholder[index0 + 449];
placeholder_d_shared[index0 + 512] = placeholder[index0 + 512];
placeholder_d_shared[index0 + 513] = placeholder[index0 + 513];
placeholder_d_shared[index0 + 576] = placeholder[index0 + 576];
placeholder_d_shared[index0 + 577] = placeholder[index0 + 577];
placeholder_d_shared[index0 + 640] = placeholder[index0 + 640];
placeholder_d_shared[index0 + 641] = placeholder[index0 + 641];
placeholder_d_shared[index0 + 704] = placeholder[index0 + 704];
placeholder_d_shared[index0 + 705] = placeholder[index0 + 705];
if ((int)threadIdx.x < 8) {
placeholder_d_shared[index0 + 768] = placeholder[index0 + 768];
}
if ((int)threadIdx.x < 8) {
placeholder_d_shared[index0 + 769] = placeholder[index0 + 769];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1568; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
int index2 = (int)blockIdx.x * 50176 + ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index2];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
int index3 = k_outer_inner * 4;
int index4 = (int)threadIdx.x * 1568 + k_outer_inner * 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[index3] * placeholder_shared[index4];
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 1] * placeholder_shared[index4 + 1]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 2] * placeholder_shared[index4 + 2]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index3 + 3] * placeholder_shared[index4 + 3]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 784]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index3 + 1] * placeholder_shared[index4 + 785]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index3 + 2] * placeholder_shared[index4 + 786]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index3 + 3] * placeholder_shared[index4 + 787]);
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index5 = (int)blockIdx.x * 64 + (int)threadIdx.x * 2 + ax1_inner;
int index6 = ax1_inner;
T_relu[index5] = max(T_dense[index6] + placeholder2[index5], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp_c7__9bw | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<7>;
.reg .f32 %f<1318>;
.reg .b32 %r<60>;
.reg .b64 %rd<17>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[64];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[8192];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd3, %rd5;
mov.u32 %r54, 0;
mov.f32 %f1062, 0f00000000;
for.size.1 i in range(0, 127, 1):
mov.f32 %f(1063+i*1), %f1062;
$L__BB0_1:
mov.u32 %r55, 0;
bar.sync 0;
shl.b32 %r2, %r54, 4;
mul.wide.s32 %rd9, %r2, 4;
add.s64 %rd10, %rd3, %rd9;
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(513+i*1), [%rd10+(0+i*4)];
st.shared.f32 [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)], %f(513+i*1);
$L__BB0_2:
shl.b32 %r45, %r54, 4;
shr.u32 %r16, %r55, 4;
mad.lo.s32 %r17, %r16, 784, %r45;
mul.wide.s32 %rd11, %r17, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f529, [%rd12];
shl.b32 %r18, %r55, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r20, %r19, %r18;
for.size.2 i in range(0, 15, 1):
st.shared.f32 [%r20+(0+i*4)], %f(529+i*1);
ld.global.nc.f32 %f(530+i*1), [%rd12+(4+i*4)];
st.shared.f32 [%r20+60], %f544;
add.s32 %r55, %r55, 16;
setp.ne.s32 %p1, %r55, 2048;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r56, 0;
$L__BB0_4:
mov.u32 %r46, _ZZ24default_function_kernel0E18placeholder_shared;
shl.b32 %r22, %r56, 2;
add.s32 %r24, %r46, %r22;
ld.shared.f32 %f545, [%r24];
mov.u32 %r25, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r26, %r25, %r22;
ld.shared.f32 %f546, [%r26];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(1189+i*-1), %f546, %f(545+i*2), %f(1189+i*-1);
ld.shared.f32 %f(547+i*1), [%r24+(64+i*64)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(1187+i*-1), %f546, %f(548+i*1), %f(1187+i*-1);
ld.shared.f32 %f(549+i*1), [%r24+(192+i*64)];
fma.rn.f32 %f1158, %f546, %f577, %f1158;
add.s32 %r56, %r56, 1;
setp.ne.s32 %p2, %r56, 16;
@%p2 bra $L__BB0_4;
mov.u32 %r57, 0;
$L__BB0_6:
mov.u32 %r51, _ZZ24default_function_kernel0E20placeholder_d_shared;
mov.u32 %r47, _ZZ24default_function_kernel0E18placeholder_shared;
shl.b32 %r28, %r57, 2;
add.s32 %r30, %r47, %r28;
ld.shared.f32 %f578, [%r30+2048];
add.s32 %r32, %r51, %r28;
ld.shared.f32 %f579, [%r32];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(1157+i*-1), %f579, %f(578+i*2), %f(1157+i*-1);
ld.shared.f32 %f(580+i*1), [%r30+(2112+i*64)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(1155+i*-1), %f579, %f(581+i*1), %f(1155+i*-1);
ld.shared.f32 %f(582+i*1), [%r30+(2240+i*64)];
fma.rn.f32 %f1126, %f579, %f610, %f1126;
add.s32 %r57, %r57, 1;
setp.ne.s32 %p3, %r57, 16;
@%p3 bra $L__BB0_6;
mov.u32 %r58, 0;
$L__BB0_8:
mov.u32 %r52, _ZZ24default_function_kernel0E20placeholder_d_shared;
mov.u32 %r48, _ZZ24default_function_kernel0E18placeholder_shared;
shl.b32 %r34, %r58, 2;
add.s32 %r36, %r48, %r34;
ld.shared.f32 %f611, [%r36+4096];
add.s32 %r38, %r52, %r34;
ld.shared.f32 %f612, [%r38];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(1125+i*-1), %f612, %f(611+i*2), %f(1125+i*-1);
ld.shared.f32 %f(613+i*1), [%r36+(4160+i*64)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(1123+i*-1), %f612, %f(614+i*1), %f(1123+i*-1);
ld.shared.f32 %f(615+i*1), [%r36+(4288+i*64)];
fma.rn.f32 %f1094, %f612, %f643, %f1094;
add.s32 %r58, %r58, 1;
setp.ne.s32 %p4, %r58, 16;
@%p4 bra $L__BB0_8;
mov.u32 %r59, 0;
$L__BB0_10:
mov.u32 %r53, _ZZ24default_function_kernel0E20placeholder_d_shared;
mov.u32 %r49, _ZZ24default_function_kernel0E18placeholder_shared;
shl.b32 %r40, %r59, 2;
add.s32 %r42, %r49, %r40;
ld.shared.f32 %f644, [%r42+6144];
add.s32 %r44, %r53, %r40;
ld.shared.f32 %f645, [%r44];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(1093+i*-1), %f645, %f(644+i*2), %f(1093+i*-1);
ld.shared.f32 %f(646+i*1), [%r42+(6208+i*64)];
for.size.2 i in range(0, 29, 1):
fma.rn.f32 %f(1091+i*-1), %f645, %f(647+i*1), %f(1091+i*-1);
ld.shared.f32 %f(648+i*1), [%r42+(6336+i*64)];
fma.rn.f32 %f1062, %f645, %f676, %f1062;
add.s32 %r59, %r59, 1;
setp.ne.s32 %p5, %r59, 16;
@%p5 bra $L__BB0_10;
add.s32 %r54, %r54, 1;
setp.lt.u32 %p6, %r54, 49;
@%p6 bra $L__BB0_1;
ld.param.u64 %rd16, [default_function_kernel0_param_2];
cvta.to.global.u64 %rd15, %rd16;
ld.param.u64 %rd14, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd14;
ld.global.nc.f32 %f677, [%rd13];
add.f32 %f678, %f1189, %f677;
mov.f32 %f679, 0f00000000;
for.size.4 i in range(0, 2, 1):
max.f32 %f(680+i*3), %f(678+i*4), %f679;
st.global.f32 [%rd15+(0+i*4)], %f(680+i*3);
ld.global.nc.f32 %f(681+i*3), [%rd13+(4+i*4)];
add.f32 %f(682+i*3), %f(1188+i*-1), %f(681+i*3);
for.size.4 i in range(0, 125, 1):
max.f32 %f(686+i*3), %f(685+i*3), %f679;
st.global.f32 [%rd15+(8+i*4)], %f(686+i*3);
ld.global.nc.f32 %f(687+i*3), [%rd13+(12+i*4)];
add.f32 %f(688+i*3), %f(1186+i*-1), %f(687+i*3);
max.f32 %f1061, %f1060, %f679;
st.global.f32 [%rd15+508], %f1061;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[128];
__shared__ float placeholder_d_shared[16];
__shared__ float placeholder_shared[2048];
for (int j_outer_inner_init = 0; j_outer_inner_init < 4; ++j_outer_inner_init) {
int index0 = j_outer_inner_init * 32;
T_dense[index0] = 0.000000e+00f;
T_dense[index0 + 1] = 0.000000e+00f;
T_dense[index0 + 2] = 0.000000e+00f;
T_dense[index0 + 3] = 0.000000e+00f;
T_dense[index0 + 4] = 0.000000e+00f;
T_dense[index0 + 5] = 0.000000e+00f;
T_dense[index0 + 6] = 0.000000e+00f;
T_dense[index0 + 7] = 0.000000e+00f;
T_dense[index0 + 8] = 0.000000e+00f;
T_dense[index0 + 9] = 0.000000e+00f;
T_dense[index0 + 10] = 0.000000e+00f;
T_dense[index0 + 11] = 0.000000e+00f;
T_dense[index0 + 12] = 0.000000e+00f;
T_dense[index0 + 13] = 0.000000e+00f;
T_dense[index0 + 14] = 0.000000e+00f;
T_dense[index0 + 15] = 0.000000e+00f;
T_dense[index0 + 16] = 0.000000e+00f;
T_dense[index0 + 17] = 0.000000e+00f;
T_dense[index0 + 18] = 0.000000e+00f;
T_dense[index0 + 19] = 0.000000e+00f;
T_dense[index0 + 20] = 0.000000e+00f;
T_dense[index0 + 21] = 0.000000e+00f;
T_dense[index0 + 22] = 0.000000e+00f;
T_dense[index0 + 23] = 0.000000e+00f;
T_dense[index0 + 24] = 0.000000e+00f;
T_dense[index0 + 25] = 0.000000e+00f;
T_dense[index0 + 26] = 0.000000e+00f;
T_dense[index0 + 27] = 0.000000e+00f;
T_dense[index0 + 28] = 0.000000e+00f;
T_dense[index0 + 29] = 0.000000e+00f;
T_dense[index0 + 30] = 0.000000e+00f;
T_dense[index0 + 31] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 49; ++k_outer_outer) {
__syncthreads();
int index1 = k_outer_outer * 16;
placeholder_d_shared[0] = placeholder[index1];
placeholder_d_shared[1] = placeholder[index1 + 1];
placeholder_d_shared[2] = placeholder[index1 + 2];
placeholder_d_shared[3] = placeholder[index1 + 3];
placeholder_d_shared[4] = placeholder[index1 + 4];
placeholder_d_shared[5] = placeholder[index1 + 5];
placeholder_d_shared[6] = placeholder[index1 + 6];
placeholder_d_shared[7] = placeholder[index1 + 7];
placeholder_d_shared[8] = placeholder[index1 + 8];
placeholder_d_shared[9] = placeholder[index1 + 9];
placeholder_d_shared[10] = placeholder[index1 + 10];
placeholder_d_shared[11] = placeholder[index1 + 11];
placeholder_d_shared[12] = placeholder[index1 + 12];
placeholder_d_shared[13] = placeholder[index1 + 13];
placeholder_d_shared[14] = placeholder[index1 + 14];
placeholder_d_shared[15] = placeholder[index1 + 15];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 2048; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer;
int index3 = (ax0_ax1_fused_outer_outer >> 4) * 784 + k_outer_outer * 16 + (ax0_ax1_fused_outer_outer & 15);
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int j_outer_inner = 0; j_outer_inner < 4; ++j_outer_inner) {
for (int k_inner = 0; k_inner < 16; ++k_inner) {
int index4 = j_outer_inner * 32;
int index5 = k_inner;
int index6 = j_outer_inner * 512 + k_inner;
T_dense[index4] = T_dense[index4] + placeholder_d_shared[index5] * placeholder_shared[index6];
T_dense[index4 + 1] = T_dense[index4 + 1] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 16]);
T_dense[index4 + 2] = T_dense[index4 + 2] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 32]);
T_dense[index4 + 3] = T_dense[index4 + 3] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 48]);
T_dense[index4 + 4] = T_dense[index4 + 4] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 64]);
T_dense[index4 + 5] = T_dense[index4 + 5] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 80]);
T_dense[index4 + 6] = T_dense[index4 + 6] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 96]);
T_dense[index4 + 7] = T_dense[index4 + 7] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 112]);
T_dense[index4 + 8] = T_dense[index4 + 8] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 128]);
T_dense[index4 + 9] = T_dense[index4 + 9] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 144]);
T_dense[index4 + 10] = T_dense[index4 + 10] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 160]);
T_dense[index4 + 11] = T_dense[index4 + 11] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 176]);
T_dense[index4 + 12] = T_dense[index4 + 12] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 192]);
T_dense[index4 + 13] = T_dense[index4 + 13] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 208]);
T_dense[index4 + 14] = T_dense[index4 + 14] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 224]);
T_dense[index4 + 15] = T_dense[index4 + 15] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 240]);
T_dense[index4 + 16] = T_dense[index4 + 16] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 256]);
T_dense[index4 + 17] = T_dense[index4 + 17] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 272]);
T_dense[index4 + 18] = T_dense[index4 + 18] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 288]);
T_dense[index4 + 19] = T_dense[index4 + 19] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 304]);
T_dense[index4 + 20] = T_dense[index4 + 20] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 320]);
T_dense[index4 + 21] = T_dense[index4 + 21] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 336]);
T_dense[index4 + 22] = T_dense[index4 + 22] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 352]);
T_dense[index4 + 23] = T_dense[index4 + 23] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 368]);
T_dense[index4 + 24] = T_dense[index4 + 24] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 384]);
T_dense[index4 + 25] = T_dense[index4 + 25] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 400]);
T_dense[index4 + 26] = T_dense[index4 + 26] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 416]);
T_dense[index4 + 27] = T_dense[index4 + 27] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 432]);
T_dense[index4 + 28] = T_dense[index4 + 28] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 448]);
T_dense[index4 + 29] = T_dense[index4 + 29] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 464]);
T_dense[index4 + 30] = T_dense[index4 + 30] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 480]);
T_dense[index4 + 31] = T_dense[index4 + 31] + (placeholder_d_shared[index5] * placeholder_shared[index6 + 496]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 128; ++ax1_inner) {
int index7 = ax1_inner;
T_relu[index7] = max(T_dense[index7] + placeholder2[index7], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpe9xc83a8 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .b16 %rs<73>;
.reg .f32 %f<771>;
.reg .b32 %r<213>;
.reg .b64 %rd<32>;
// demoted variable
.shared .align 8 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[112];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[7168];
ld.param.u64 %rd1, [default_function_kernel0_param_0];
ld.param.u64 %rd2, [default_function_kernel0_param_1];
mov.u32 %r209, 0;
mov.f32 %f707, 0f00000000;
cvta.to.global.u64 %rd5, %rd1;
mov.u32 %r70, %ctaid.x;
cvta.to.global.u64 %rd8, %rd2;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(708+i*1), %f707;
$L__BB0_1:
mov.u32 %r210, 0;
bar.sync 0;
mul.lo.s32 %r13, %r209, 28;
mul.wide.s32 %rd6, %r13, 4;
add.s64 %rd7, %rd5, %rd6;
for.size.2 i in range(0, 14, 1):
ld.global.nc.v2.u32 {%r(14+i*4), %r(15+i*4)}, [%rd7+(0+i*8)];
st.shared.v2.u32 [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*8)], {%r(14+i*4), %r(15+i*4)};
$L__BB0_2:
mul.lo.s32 %r204, %r209, 28;
mad.lo.s32 %r203, %r70, 50176, %r204;
shl.b32 %r71, %r210, 4;
mov.u32 %r72, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r73, %r72, %r71;
for.size.20 i in range(0, 2, 1):
cvt.u16.u32 %rs(1+i*9), %r(210+i*-122);
mul.wide.u16 %r(74+i*15), %rs(1+i*9), 9363;
shr.u32 %r(75+i*15), %r(74+i*15), 16;
cvt.u16.u32 %rs(2+i*9), %r(75+i*15);
sub.s16 %rs(3+i*9), %rs(1+i*9), %rs(2+i*9);
shr.u16 %rs(4+i*9), %rs(3+i*9), 1;
add.s16 %rs(5+i*9), %rs(4+i*9), %rs(2+i*9);
shr.u16 %rs(6+i*9), %rs(5+i*9), 2;
cvt.u32.u16 %r(76+i*15), %rs(6+i*9);
mul.lo.s16 %rs(7+i*9), %rs(6+i*9), 7;
sub.s16 %rs(8+i*9), %rs(1+i*9), %rs(7+i*9);
shl.b16 %rs(9+i*9), %rs(8+i*9), 2;
cvt.u32.u16 %r(77+i*15), %rs(9+i*9);
mad.lo.s32 %r(78+i*15), %r(76+i*15), 784, %r203;
add.s32 %r(79+i*15), %r(78+i*15), %r(77+i*15);
mul.wide.s32 %rd(9+i*2), %r(79+i*15), 4;
add.s64 %rd(10+i*2), %rd8, %rd(9+i*2);
ld.global.nc.v4.u32 {%r(80+i*15), %r(81+i*15), %r(82+i*15), %r(83+i*15)}, [%rd(10+i*2)+0];
st.shared.v4.u32 [%r73+(0+i*16)], {%r(80+i*15), %r(81+i*15), %r(82+i*15), %r(83+i*15)};
add.s32 %r(88+i*15), %r210, (1+i*1);
for.size.20 i in range(0, 5, 1):
cvt.u16.u32 %rs(19+i*9), %r(103+i*15);
mul.wide.u16 %r(104+i*15), %rs(19+i*9), 9363;
shr.u32 %r(105+i*15), %r(104+i*15), 16;
cvt.u16.u32 %rs(20+i*9), %r(105+i*15);
sub.s16 %rs(21+i*9), %rs(19+i*9), %rs(20+i*9);
shr.u16 %rs(22+i*9), %rs(21+i*9), 1;
add.s16 %rs(23+i*9), %rs(22+i*9), %rs(20+i*9);
shr.u16 %rs(24+i*9), %rs(23+i*9), 2;
cvt.u32.u16 %r(106+i*15), %rs(24+i*9);
mul.lo.s16 %rs(25+i*9), %rs(24+i*9), 7;
sub.s16 %rs(26+i*9), %rs(19+i*9), %rs(25+i*9);
shl.b16 %rs(27+i*9), %rs(26+i*9), 2;
cvt.u32.u16 %r(107+i*15), %rs(27+i*9);
mad.lo.s32 %r(108+i*15), %r(106+i*15), 784, %r203;
add.s32 %r(109+i*15), %r(108+i*15), %r(107+i*15);
mul.wide.s32 %rd(13+i*2), %r(109+i*15), 4;
add.s64 %rd(14+i*2), %rd8, %rd(13+i*2);
ld.global.nc.v4.u32 {%r(110+i*15), %r(111+i*15), %r(112+i*15), %r(113+i*15)}, [%rd(14+i*2)+0];
st.shared.v4.u32 [%r73+(32+i*16)], {%r(110+i*15), %r(111+i*15), %r(112+i*15), %r(113+i*15)};
add.s32 %r(118+i*15), %r210, (3+i*1);
cvt.u16.u32 %rs64, %r178;
mul.wide.u16 %r179, %rs64, 9363;
shr.u32 %r180, %r179, 16;
cvt.u16.u32 %rs65, %r180;
sub.s16 %rs66, %rs64, %rs65;
shr.u16 %rs67, %rs66, 1;
add.s16 %rs68, %rs67, %rs65;
shr.u16 %rs69, %rs68, 2;
cvt.u32.u16 %r181, %rs69;
mul.lo.s16 %rs70, %rs69, 7;
sub.s16 %rs71, %rs64, %rs70;
shl.b16 %rs72, %rs71, 2;
cvt.u32.u16 %r182, %rs72;
mad.lo.s32 %r183, %r181, 784, %r203;
add.s32 %r184, %r183, %r182;
mul.wide.s32 %rd23, %r184, 4;
add.s64 %rd24, %rd8, %rd23;
ld.global.nc.v4.u32 {%r185, %r186, %r187, %r188}, [%rd24];
st.shared.v4.u32 [%r73+112], {%r185, %r186, %r187, %r188};
add.s32 %r210, %r210, 8;
setp.ne.s32 %p1, %r210, 448;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r211, 0;
$L__BB0_4:
mov.u32 %r212, 0;
$L__BB0_5:
mul.lo.s32 %r207, %r211, 14;
mov.u32 %r205, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r195, %r212, %r207;
shl.b32 %r196, %r195, 2;
mov.u32 %r197, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(198+i*2), %r(197+i*8), %r196;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(321+i*1), [%r(200+i*-2)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(770+i*-1), %f322, %f(321+i*2), %f(770+i*-1);
ld.shared.f32 %f(323+i*1), [%r200+(112+i*112)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(768+i*-1), %f322, %f(324+i*1), %f(768+i*-1);
ld.shared.f32 %f(325+i*1), [%r200+(336+i*112)];
fma.rn.f32 %f707, %f322, %f385, %f707;
add.s32 %r212, %r212, 1;
setp.ne.s32 %p2, %r212, 14;
@%p2 bra $L__BB0_5;
add.s32 %r211, %r211, 1;
setp.lt.u32 %p3, %r211, 2;
@%p3 bra $L__BB0_4;
add.s32 %r209, %r209, 1;
setp.lt.u32 %p4, %r209, 28;
@%p4 bra $L__BB0_1;
ld.param.u64 %rd31, [default_function_kernel0_param_2];
ld.param.u64 %rd30, [default_function_kernel0_param_3];
shl.b32 %r202, %r70, 6;
cvta.to.global.u64 %rd25, %rd30;
mul.wide.s32 %rd26, %r202, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f386, [%rd27];
add.f32 %f387, %f770, %f386;
mov.f32 %f388, 0f00000000;
max.f32 %f389, %f387, %f388;
cvta.to.global.u64 %rd28, %rd31;
add.s64 %rd29, %rd28, %rd26;
for.size.4 i in range(0, 63, 1):
st.global.f32 [%rd29+(0+i*4)], %f(389+i*3);
ld.global.nc.f32 %f(390+i*3), [%rd27+(4+i*4)];
add.f32 %f(391+i*3), %f(769+i*-1), %f(390+i*3);
max.f32 %f(392+i*3), %f(391+i*3), %f388;
st.global.f32 [%rd29+252], %f578;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[28];
__shared__ float placeholder_shared[1792];
for (int j_inner_init = 0; j_inner_init < 64; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 28; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 14; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_d_shared + ax0_ax1_fused_outer_outer * 2))[0] = ((float2*)(placeholder + k_outer_outer * 28 + ax0_ax1_fused_outer_outer * 2))[0];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 448; ++ax0_ax1_fused_outer_outer1) {
((float4*)(placeholder_shared + ax0_ax1_fused_outer_outer1 * 4))[0] = ((float4*)(placeholder1 + (int)blockIdx.x * 50176 + ax0_ax1_fused_outer_outer1 / 7 * 784 + k_outer_outer * 28 + ax0_ax1_fused_outer_outer1 % 7 * 4))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 14; ++k_inner) {
for (int j_inner = 0; j_inner < 64; ++j_inner) {
int index1 = j_inner;
int index2 = k_outer_inner * 14 + k_inner;
int index3 = j_inner * 28 + k_outer_inner * 14 + k_inner;
T_dense[index1] = T_dense[index1] + placeholder_d_shared[index2] * placeholder_shared[index3];
}
}
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index4 = (int)blockIdx.x * 64 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp59vz2zqk | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<106>;
.reg .b32 %r<16>;
.reg .b64 %rd<23>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[28];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[3584];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd1, %rd11;
mov.u32 %r1, %tid.x;
shl.b32 %r8, %r1, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r9, %r8;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(10+i*1), %r1, (6272+i*-6216);
shl.b32 %r12, %r11, 2;
mov.u32 %r13, _ZZ24default_function_kernel0E18placeholder_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(3+i*1), %r13, %r(12+i*-1);
cvta.to.global.u64 %rd13, %rd9;
for.size.2 i in range(0, 2, 1):
mul.wide.s32 %rd(14+i*1), %r(1+i*9), 4;
add.s64 %rd(22+i*-6), %rd(13+i*-1), %rd(14+i*1);
add.s64 %rd21, %rd16, 12544;
mov.u32 %r15, 0;
mov.f32 %f104, 0f00000000;
mov.f32 %f105, %f104;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 6;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f7, [%rd22];
st.shared.f32 [%r2], %f7;
$L__BB0_3:
setp.gt.s32 %p2, %r1, 15;
@%p2 bra $L__BB0_5;
for.size.3 i in range(0, 8, 1):
for.size.2 j in range(0, 7, 1):
ld.global.nc.f32 %f(8+i*7+j*1), [%rd21+(-12544+i*3136+j*4)];
st.shared.f32 [%r3+(0+i*28+j*4)], %f(8+i*7+j*1);
$L__BB0_5:
bar.sync 0;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(64+i*3), [%r4+(0+i*4)];
ld.shared.f32 %f(65+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(66+i*3), %f(65+i*3), %f(64+i*3), %f(105+i*-39);
for.size.3 i in range(0, 4, 1):
ld.shared.f32 %f(70+i*3), [%r4+(8+i*4)];
ld.shared.f32 %f(71+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(72+i*3), %f(71+i*3), %f(70+i*3), %f(69+i*3);
ld.shared.f32 %f82, [%r4+24];
ld.shared.f32 %f83, [_ZZ24default_function_kernel0E20placeholder_d_shared+24];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(105+i*-20), %f(83+i*-18), %f(82+i*2), %f(81+i*23);
ld.shared.f32 %f(84+i*2), [%r4+(28+i*4)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(87+i*2), %f(68+i*3), %f(86+i*2), %f(85+i*2);
ld.shared.f32 %f(88+i*2), [%r4+(36+i*4)];
fma.rn.f32 %f104, %f83, %f96, %f95;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(22+i*-1), %rd(22+i*-1), 28;
add.s32 %r15, %r15, 1;
setp.ne.s32 %p3, %r15, 112;
@%p3 bra $L__BB0_1;
cvta.to.global.u64 %rd17, %rd8;
shl.b32 %r14, %r1, 1;
mul.wide.s32 %rd18, %r14, 4;
add.s64 %rd19, %rd17, %rd18;
ld.global.nc.f32 %f97, [%rd19];
add.f32 %f98, %f105, %f97;
mov.f32 %f99, 0f00000000;
max.f32 %f100, %f98, %f99;
add.s64 %rd20, %rd1, %rd18;
st.global.f32 [%rd20], %f100;
ld.global.nc.f32 %f101, [%rd19+4];
add.f32 %f102, %f104, %f101;
max.f32 %f103, %f102, %f99;
st.global.f32 [%rd20+4], %f103;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[7];
__shared__ float placeholder_shared[896];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 112; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 7) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 7 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x;
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
placeholder_shared[index0 * 56] = placeholder1[index2];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index3 = (int)threadIdx.x * 56 + 1;
placeholder_shared[index3] = placeholder1[index2 + 1];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index4 = (int)threadIdx.x * 56 + 2;
placeholder_shared[index4] = placeholder1[index2 + 2];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index5 = (int)threadIdx.x * 56 + 3;
placeholder_shared[index5] = placeholder1[index2 + 3];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index6 = (int)threadIdx.x * 56 + 4;
placeholder_shared[index6] = placeholder1[index2 + 4];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index7 = (int)threadIdx.x * 56 + 5;
placeholder_shared[index7] = placeholder1[index2 + 5];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index8 = (int)threadIdx.x * 56 + 6;
placeholder_shared[index8] = placeholder1[index2 + 6];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index9 = (int)threadIdx.x * 56 + 7;
placeholder_shared[index9] = placeholder1[index2 + 784];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index10 = (int)threadIdx.x * 56 + 8;
placeholder_shared[index10] = placeholder1[index2 + 785];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index11 = (int)threadIdx.x * 56 + 9;
placeholder_shared[index11] = placeholder1[index2 + 786];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index12 = (int)threadIdx.x * 56 + 10;
placeholder_shared[index12] = placeholder1[index2 + 787];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index13 = (int)threadIdx.x * 56 + 11;
placeholder_shared[index13] = placeholder1[index2 + 788];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index14 = (int)threadIdx.x * 56 + 12;
placeholder_shared[index14] = placeholder1[index2 + 789];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index15 = (int)threadIdx.x * 56 + 13;
placeholder_shared[index15] = placeholder1[index2 + 790];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index16 = (int)threadIdx.x * 56 + 14;
placeholder_shared[index16] = placeholder1[index2 + 1568];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index17 = (int)threadIdx.x * 56 + 15;
placeholder_shared[index17] = placeholder1[index2 + 1569];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index18 = (int)threadIdx.x * 56 + 16;
placeholder_shared[index18] = placeholder1[index2 + 1570];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index19 = (int)threadIdx.x * 56 + 17;
placeholder_shared[index19] = placeholder1[index2 + 1571];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index20 = (int)threadIdx.x * 56 + 18;
placeholder_shared[index20] = placeholder1[index2 + 1572];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index21 = (int)threadIdx.x * 56 + 19;
placeholder_shared[index21] = placeholder1[index2 + 1573];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index22 = (int)threadIdx.x * 56 + 20;
placeholder_shared[index22] = placeholder1[index2 + 1574];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index23 = (int)threadIdx.x * 56 + 21;
placeholder_shared[index23] = placeholder1[index2 + 2352];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index24 = (int)threadIdx.x * 56 + 22;
placeholder_shared[index24] = placeholder1[index2 + 2353];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index25 = (int)threadIdx.x * 56 + 23;
placeholder_shared[index25] = placeholder1[index2 + 2354];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index26 = (int)threadIdx.x * 56 + 24;
placeholder_shared[index26] = placeholder1[index2 + 2355];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index27 = (int)threadIdx.x * 56 + 25;
placeholder_shared[index27] = placeholder1[index2 + 2356];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index28 = (int)threadIdx.x * 56 + 26;
placeholder_shared[index28] = placeholder1[index2 + 2357];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index29 = (int)threadIdx.x * 56 + 27;
placeholder_shared[index29] = placeholder1[index2 + 2358];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index30 = (int)threadIdx.x * 56 + 28;
placeholder_shared[index30] = placeholder1[index2 + 3136];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index31 = (int)threadIdx.x * 56 + 29;
placeholder_shared[index31] = placeholder1[index2 + 3137];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index32 = (int)threadIdx.x * 56 + 30;
placeholder_shared[index32] = placeholder1[index2 + 3138];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index33 = (int)threadIdx.x * 56 + 31;
placeholder_shared[index33] = placeholder1[index2 + 3139];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index34 = (int)threadIdx.x * 56 + 32;
placeholder_shared[index34] = placeholder1[index2 + 3140];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index35 = (int)threadIdx.x * 56 + 33;
placeholder_shared[index35] = placeholder1[index2 + 3141];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index36 = (int)threadIdx.x * 56 + 34;
placeholder_shared[index36] = placeholder1[index2 + 3142];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index37 = (int)threadIdx.x * 56 + 35;
placeholder_shared[index37] = placeholder1[index2 + 3920];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index38 = (int)threadIdx.x * 56 + 36;
placeholder_shared[index38] = placeholder1[index2 + 3921];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index39 = (int)threadIdx.x * 56 + 37;
placeholder_shared[index39] = placeholder1[index2 + 3922];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index40 = (int)threadIdx.x * 56 + 38;
placeholder_shared[index40] = placeholder1[index2 + 3923];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index41 = (int)threadIdx.x * 56 + 39;
placeholder_shared[index41] = placeholder1[index2 + 3924];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index42 = (int)threadIdx.x * 56 + 40;
placeholder_shared[index42] = placeholder1[index2 + 3925];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index43 = (int)threadIdx.x * 56 + 41;
placeholder_shared[index43] = placeholder1[index2 + 3926];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index44 = (int)threadIdx.x * 56 + 42;
placeholder_shared[index44] = placeholder1[index2 + 4704];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index45 = (int)threadIdx.x * 56 + 43;
placeholder_shared[index45] = placeholder1[index2 + 4705];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index46 = (int)threadIdx.x * 56 + 44;
placeholder_shared[index46] = placeholder1[index2 + 4706];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index47 = (int)threadIdx.x * 56 + 45;
placeholder_shared[index47] = placeholder1[index2 + 4707];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index48 = (int)threadIdx.x * 56 + 46;
placeholder_shared[index48] = placeholder1[index2 + 4708];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index49 = (int)threadIdx.x * 56 + 47;
placeholder_shared[index49] = placeholder1[index2 + 4709];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index50 = (int)threadIdx.x * 56 + 48;
placeholder_shared[index50] = placeholder1[index2 + 4710];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index51 = (int)threadIdx.x * 56 + 49;
placeholder_shared[index51] = placeholder1[index2 + 5488];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index52 = (int)threadIdx.x * 56 + 50;
placeholder_shared[index52] = placeholder1[index2 + 5489];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index53 = (int)threadIdx.x * 56 + 51;
placeholder_shared[index53] = placeholder1[index2 + 5490];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index54 = (int)threadIdx.x * 56 + 52;
placeholder_shared[index54] = placeholder1[index2 + 5491];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index55 = (int)threadIdx.x * 56 + 53;
placeholder_shared[index55] = placeholder1[index2 + 5492];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index56 = (int)threadIdx.x * 56 + 54;
placeholder_shared[index56] = placeholder1[index2 + 5493];
}
if ((int)threadIdx.x < 16) {
int index2 = (int)threadIdx.x * 6272 + k_outer_outer * 7;
int index57 = (int)threadIdx.x * 56 + 55;
placeholder_shared[index57] = placeholder1[index2 + 5494];
}
__syncthreads();
int index0 = (int)threadIdx.x;
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 14];
int index58 = (int)threadIdx.x * 14 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index58];
int index59 = (int)threadIdx.x * 14 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index59];
int index60 = (int)threadIdx.x * 14 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index60];
int index61 = (int)threadIdx.x * 14 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index61];
int index62 = (int)threadIdx.x * 14 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index62];
int index63 = (int)threadIdx.x * 14 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index63];
int index64 = (int)threadIdx.x * 14 + 7;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index64];
int index65 = (int)threadIdx.x * 14 + 8;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index65];
int index66 = (int)threadIdx.x * 14 + 9;
T_dense[1] = T_dense[1] + placeholder_d_shared[2] * placeholder_shared[index66];
int index67 = (int)threadIdx.x * 14 + 10;
T_dense[1] = T_dense[1] + placeholder_d_shared[3] * placeholder_shared[index67];
int index68 = (int)threadIdx.x * 14 + 11;
T_dense[1] = T_dense[1] + placeholder_d_shared[4] * placeholder_shared[index68];
int index69 = (int)threadIdx.x * 14 + 12;
T_dense[1] = T_dense[1] + placeholder_d_shared[5] * placeholder_shared[index69];
int index70 = (int)threadIdx.x * 14 + 13;
T_dense[1] = T_dense[1] + placeholder_d_shared[6] * placeholder_shared[index70];
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index71 = (int)threadIdx.x * 2 + ax1_inner;
int index72 = ax1_inner;
T_relu[index71] = max(T_dense[index72] + placeholder2[index71], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpu8xj972e | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<68>;
.reg .b32 %r<120>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[224];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[14336];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd5, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
mov.u32 %r1, %tid.x;
shl.b32 %r21, %r1, 2;
mov.u32 %r22, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r22, %r21;
mov.u32 %r3, %ctaid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(4+i*1), %r(3+i*-2), (50176+i*-50120);
add.s32 %r6, %r1, 224;
mov.u32 %r23, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r7, %r23, %r21;
mov.f32 %f64, 0f00000000;
mov.u32 %r115, 0;
mov.f32 %f65, %f64;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r9, %r115, 56;
add.s32 %r24, %r9, %r1;
mul.wide.s32 %rd8, %r24, 4;
add.s64 %rd3, %rd2, %rd8;
ld.global.nc.f32 %f9, [%rd3];
st.shared.f32 [%r2], %f9;
setp.gt.s32 %p1, %r1, 23;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f10, [%rd3+128];
st.shared.f32 [%r2+128], %f10;
$L__BB0_3:
add.s32 %r10, %r9, %r4;
mov.u32 %r118, 512;
mov.u32 %r116, %r1;
mov.u32 %r117, %r6;
$L__BB0_4:
mul.hi.s32 %r26, %r116, -1840700269;
add.s32 %r27, %r26, %r116;
shr.u32 %r28, %r27, 31;
shr.s32 %r29, %r27, 5;
add.s32 %r30, %r29, %r28;
mul.lo.s32 %r31, %r30, 56;
sub.s32 %r32, %r116, %r31;
add.s32 %r33, %r10, %r32;
mad.lo.s32 %r34, %r30, 784, %r33;
mul.wide.s32 %rd9, %r34, 4;
add.s64 %rd10, %rd1, %rd9;
ld.global.nc.f32 %f11, [%rd10];
add.s32 %r35, %r7, %r118;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r35+(-512+i*128)], %f(11+i*1);
add.s32 %r(36+i*10), %r116, (32+i*32);
mul.hi.s32 %r(37+i*10), %r(36+i*10), -1840700269;
add.s32 %r(38+i*10), %r(37+i*10), %r(36+i*10);
shr.u32 %r(39+i*10), %r(38+i*10), 31;
shr.s32 %r(40+i*10), %r(38+i*10), 5;
add.s32 %r(41+i*10), %r(40+i*10), %r(39+i*10);
mul.lo.s32 %r(42+i*10), %r(41+i*10), 56;
sub.s32 %r(43+i*10), %r(36+i*10), %r(42+i*10);
add.s32 %r(44+i*10), %r10, %r(43+i*10);
mad.lo.s32 %r(45+i*10), %r(41+i*10), 784, %r(44+i*10);
mul.wide.s32 %rd(11+i*2), %r(45+i*10), 4;
add.s64 %rd(12+i*2), %rd1, %rd(11+i*2);
ld.global.nc.f32 %f(12+i*1), [%rd(12+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r35+(0+i*128)], %f(15+i*1);
add.s32 %r(76+i*10), %r117, (-64+i*32);
mul.hi.s32 %r(77+i*10), %r(76+i*10), -1840700269;
add.s32 %r(78+i*10), %r(77+i*10), %r(76+i*10);
shr.u32 %r(79+i*10), %r(78+i*10), 31;
shr.s32 %r(80+i*10), %r(78+i*10), 5;
add.s32 %r(81+i*10), %r(80+i*10), %r(79+i*10);
mul.lo.s32 %r(82+i*10), %r(81+i*10), 56;
sub.s32 %r(83+i*10), %r(76+i*10), %r(82+i*10);
add.s32 %r(84+i*10), %r10, %r(83+i*10);
mad.lo.s32 %r(85+i*10), %r(81+i*10), 784, %r(84+i*10);
mul.wide.s32 %rd(19+i*2), %r(85+i*10), 4;
add.s64 %rd(20+i*2), %rd1, %rd(19+i*2);
ld.global.nc.f32 %f(16+i*1), [%rd(20+i*2)+0];
st.shared.f32 [%r35+256], %f17;
mul.hi.s32 %r96, %r117, -1840700269;
add.s32 %r97, %r96, %r117;
shr.u32 %r98, %r97, 31;
shr.s32 %r99, %r97, 5;
add.s32 %r100, %r99, %r98;
mul.lo.s32 %r101, %r100, 56;
sub.s32 %r102, %r117, %r101;
add.s32 %r103, %r10, %r102;
mad.lo.s32 %r104, %r100, 784, %r103;
mul.wide.s32 %rd23, %r104, 4;
add.s64 %rd24, %rd1, %rd23;
ld.global.nc.f32 %f18, [%rd24];
st.shared.f32 [%r35+384], %f18;
for.size.1 i in range(0, 2, 1):
add.s32 %r(117+i*-1), %r(117+i*-1), 256;
add.s32 %r118, %r118, 1024;
setp.ne.s32 %p2, %r118, 14848;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r119, 0;
$L__BB0_6:
add.s32 %r106, %r119, %r5;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(107+i*3), %r(106+i*13), 2;
add.s32 %r(109+i*3), %r(23+i*-1), %r(107+i*3);
ld.shared.f32 %f(19+i*1), [%r(109+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(21+i*2), %f20, %f(19+i*3), %f(64+i*1);
ld.shared.f32 %f(22+i*2), [%r109+(7168+i*-7164)];
for.size.5 i in range(0, 6, 1):
ld.shared.f32 %f(25+i*5), [%r112+(4+i*4)];
fma.rn.f32 %f(26+i*5), %f(25+i*5), %f(24+i*5), %f(21+i*5);
ld.shared.f32 %f(27+i*5), [%r109+(7172+i*4)];
fma.rn.f32 %f(28+i*5), %f(25+i*5), %f(27+i*5), %f(23+i*5);
ld.shared.f32 %f(29+i*5), [%r109+(8+i*4)];
ld.shared.f32 %f55, [%r112+28];
fma.rn.f32 %f64, %f55, %f54, %f51;
ld.shared.f32 %f56, [%r109+7196];
fma.rn.f32 %f65, %f55, %f56, %f53;
add.s32 %r119, %r119, 8;
setp.ne.s32 %p3, %r119, 56;
@%p3 bra $L__BB0_6;
add.s32 %r115, %r115, 1;
setp.lt.u32 %p4, %r115, 14;
@%p4 bra $L__BB0_1;
shl.b32 %r113, %r3, 6;
add.s32 %r114, %r113, %r1;
cvta.to.global.u64 %rd25, %rd5;
mul.wide.s32 %rd26, %r114, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f57, [%rd27];
add.f32 %f58, %f64, %f57;
mov.f32 %f59, 0f00000000;
max.f32 %f60, %f58, %f59;
cvta.to.global.u64 %rd28, %rd4;
add.s64 %rd29, %rd28, %rd26;
st.global.f32 [%rd29], %f60;
ld.global.nc.f32 %f61, [%rd27+128];
add.f32 %f62, %f65, %f61;
max.f32 %f63, %f62, %f59;
st.global.f32 [%rd29+128], %f63;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[56];
__shared__ float placeholder_shared[3584];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 14; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 56 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 112; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 50176 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) / 56 * 784 + k_outer_outer * 56 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) % 56;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 56; ++k_outer_inner) {
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 56 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1792]);
}
}
int index6 = (int)blockIdx.x * 64 + (int)threadIdx.x;
T_relu[index6] = max(T_dense[0] + placeholder2[index6], 0.000000e+00f);
T_relu[index6 + 32] = max(T_dense[1] + placeholder2[index6 + 32], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpxg628h6h | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<96>;
.reg .b32 %r<85>;
.reg .b64 %rd<26>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[2048];
ld.param.u64 %rd11, [default_function_kernel0_param_0];
ld.param.u64 %rd12, [default_function_kernel0_param_1];
ld.param.u64 %rd13, [default_function_kernel0_param_2];
ld.param.u64 %rd10, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd14, %rd12;
cvta.to.global.u64 %rd1, %rd13;
mov.u32 %r1, %tid.x;
shl.b32 %r10, %r1, 2;
mov.u32 %r11, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r11, %r10;
shl.b32 %r12, %r1, 5;
mov.u32 %r13, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r13, %r12;
shl.b32 %r14, %r1, 6;
add.s32 %r4, %r13, %r14;
cvta.to.global.u64 %rd15, %rd11;
mul.wide.s32 %rd16, %r1, 4;
add.s64 %rd17, %rd15, %rd16;
add.s64 %rd25, %rd17, 32;
mov.u32 %r5, %ctaid.x;
mul.lo.s32 %r15, %r1, 784;
mad.lo.s32 %r16, %r5, 50176, %r15;
mul.wide.s32 %rd18, %r16, 4;
add.s64 %rd19, %rd14, %rd18;
add.s64 %rd24, %rd19, 100400;
mov.u32 %r84, 0;
mov.f32 %f94, 0f00000000;
mov.f32 %f95, %f94;
$L__BB0_1:
bar.sync 0;
add.s64 %rd6, %rd25, -32;
setp.gt.s32 %p1, %r1, 7;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f9, [%rd6];
st.shared.f32 [%r2], %f9;
$L__BB0_3:
add.s64 %rd7, %rd24, -100400;
for.size.4 i in range(0, 2, 1):
for.size.1 j in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(17+i*16+j*8), %r(18+i*16+j*8), %r(19+i*16+j*8), %r(20+i*16+j*8)}, [%rd24+(-100400+i*100352+j*16)];
for.size.1 j in range(0, 2, 1):
st.shared.v4.u32 [%r3+(0+i*1024+j*16)], {%r(17+i*16+j*8), %r(18+i*16+j*8), %r(19+i*16+j*8), %r(20+i*16+j*8)};
bar.sync 0;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(10+i*3), [%r4+(0+i*4)];
ld.shared.f32 %f(11+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(12+i*3), %f(11+i*3), %f(10+i*3), %f(95+i*-83);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(16+i*2), [%r4+(32+i*4)];
fma.rn.f32 %f(17+i*2), %f(11+i*3), %f(16+i*2), %f(94+i*-77);
for.size.7 i in range(0, 2, 1):
for.size.3 j in range(0, 2, 1):
ld.shared.f32 %f(20+i*10+j*3), [%r4+(8+i*8+j*4)];
ld.shared.f32 %f(21+i*10+j*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*8+j*4)];
fma.rn.f32 %f(22+i*10+j*3), %f(21+i*10+j*3), %f(20+i*10+j*3), %f(15+i*10+j*7);
for.size.2 j in range(0, 2, 1):
ld.shared.f32 %f(26+i*10+j*2), [%r4+(40+i*8+j*4)];
fma.rn.f32 %f(27+i*10+j*2), %f(21+i*10+j*3), %f(26+i*10+j*2), %f(19+i*10+j*8);
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(40+i*3), [%r4+(24+i*4)];
ld.shared.f32 %f(41+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(24+i*4)];
fma.rn.f32 %f(42+i*-39), %f(41+i*3), %f(40+i*3), %f(35+i*7);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(45+i*2), [%r4+(56+i*4)];
fma.rn.f32 %f(46+i*-42), %f(41+i*3), %f(45+i*2), %f(39+i*7);
bar.sync 0;
@%p1 bra $L__BB0_5;
ld.global.nc.f32 %f48, [%rd6+32];
st.shared.f32 [%r2], %f48;
$L__BB0_5:
for.size.4 i in range(0, 2, 1):
for.size.1 j in range(0, 2, 1):
ld.global.nc.v4.u32 {%r(49+i*16+j*8), %r(50+i*16+j*8), %r(51+i*16+j*8), %r(52+i*16+j*8)}, [%rd7+(32+i*100352+j*16)];
for.size.1 j in range(0, 2, 1):
st.shared.v4.u32 [%r3+(0+i*1024+j*16)], {%r(49+i*16+j*8), %r(50+i*16+j*8), %r(51+i*16+j*8), %r(52+i*16+j*8)};
bar.sync 0;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(49+i*3), [%r4+(0+i*4)];
ld.shared.f32 %f(50+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(51+i*3), %f(50+i*3), %f(49+i*3), %f(3+i*48);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(55+i*2), [%r4+(32+i*4)];
fma.rn.f32 %f(56+i*2), %f(50+i*3), %f(55+i*2), %f(4+i*52);
for.size.7 i in range(0, 2, 1):
for.size.3 j in range(0, 2, 1):
ld.shared.f32 %f(59+i*10+j*3), [%r4+(8+i*8+j*4)];
ld.shared.f32 %f(60+i*10+j*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*8+j*4)];
fma.rn.f32 %f(61+i*10+j*3), %f(60+i*10+j*3), %f(59+i*10+j*3), %f(54+i*10+j*7);
for.size.2 j in range(0, 2, 1):
ld.shared.f32 %f(65+i*10+j*2), [%r4+(40+i*8+j*4)];
fma.rn.f32 %f(66+i*10+j*2), %f(60+i*10+j*3), %f(65+i*10+j*2), %f(58+i*10+j*8);
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(79+i*3), [%r4+(24+i*4)];
ld.shared.f32 %f(80+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(24+i*4)];
fma.rn.f32 %f(81+i*14), %f(80+i*3), %f(79+i*3), %f(74+i*7);
for.size.2 i in range(0, 2, 1):
ld.shared.f32 %f(84+i*2), [%r4+(56+i*4)];
fma.rn.f32 %f(85+i*9), %f(80+i*3), %f(84+i*2), %f(78+i*7);
for.size.1 i in range(0, 2, 1):
add.s64 %rd(25+i*-1), %rd(25+i*-1), 64;
add.s32 %r84, %r84, 2;
setp.ne.s32 %p3, %r84, 98;
@%p3 bra $L__BB0_1;
cvta.to.global.u64 %rd20, %rd10;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(81+i*1), %r(1+i*4), (1+i*5);
add.s32 %r83, %r82, %r81;
mul.wide.s32 %rd21, %r83, 4;
add.s64 %rd22, %rd20, %rd21;
ld.global.nc.f32 %f87, [%rd22];
add.f32 %f88, %f95, %f87;
mov.f32 %f89, 0f00000000;
max.f32 %f90, %f88, %f89;
add.s64 %rd23, %rd1, %rd21;
st.global.f32 [%rd23], %f90;
ld.global.nc.f32 %f91, [%rd22+4];
add.f32 %f92, %f94, %f91;
max.f32 %f93, %f92, %f89;
st.global.f32 [%rd23+4], %f93;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[512];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 8) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 8 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
ulonglong4*(placeholder_shared + (int)threadIdx.x * 8)[0] = ulonglong4*(placeholder1 + (int)blockIdx.x * 50176 + (int)threadIdx.x * 784 + k_outer_outer * 8)[0];
ulonglong4*(placeholder_shared + (int)threadIdx.x * 8 + 256)[0] = ulonglong4*(placeholder1 + (int)blockIdx.x * 50176 + (int)threadIdx.x * 784 + k_outer_outer * 8 + 25088)[0];
__syncthreads();
int index0 = (int)threadIdx.x;
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 16];
int index2 = (int)threadIdx.x * 16 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index2];
int index3 = (int)threadIdx.x * 16 + 8;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index3];
int index4 = (int)threadIdx.x * 16 + 9;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index4];
int index5 = (int)threadIdx.x * 16 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index5];
int index6 = (int)threadIdx.x * 16 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index6];
int index7 = (int)threadIdx.x * 16 + 10;
T_dense[1] = T_dense[1] + placeholder_d_shared[2] * placeholder_shared[index7];
int index8 = (int)threadIdx.x * 16 + 11;
T_dense[1] = T_dense[1] + placeholder_d_shared[3] * placeholder_shared[index8];
int index9 = (int)threadIdx.x * 16 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index9];
int index10 = (int)threadIdx.x * 16 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index10];
int index11 = (int)threadIdx.x * 16 + 12;
T_dense[1] = T_dense[1] + placeholder_d_shared[4] * placeholder_shared[index11];
int index12 = (int)threadIdx.x * 16 + 13;
T_dense[1] = T_dense[1] + placeholder_d_shared[5] * placeholder_shared[index12];
int index13 = (int)threadIdx.x * 16 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index13];
int index14 = (int)threadIdx.x * 16 + 7;
T_dense[0] = T_dense[0] + placeholder_d_shared[7] * placeholder_shared[index14];
int index15 = (int)threadIdx.x * 16 + 14;
T_dense[1] = T_dense[1] + placeholder_d_shared[6] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 16 + 15;
T_dense[1] = T_dense[1] + placeholder_d_shared[7] * placeholder_shared[index16];
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index17 = (int)blockIdx.x * 64 + (int)threadIdx.x * 2 + ax1_inner;
int index18 = ax1_inner;
T_relu[index17] = max(T_dense[index18] + placeholder2[index17], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpsduw_uz7 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<156>;
.reg .b32 %r<110>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[100352];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(2+i*17), %r1, (1+i*2);
mov.u32 %r20, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r20, %r19;
mul.lo.s32 %r4, %r1, 1568;
add.s32 %r5, %r1, 112;
shl.b32 %r21, %r1, 2;
mov.u32 %r22, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r6, %r22, %r21;
mov.u32 %r105, 0;
mov.f32 %f140, 0f00000000;
for.size.1 i in range(0, 7, 1):
mov.f32 %f(141+i*1), %f140;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r8, %r105, 196;
add.s32 %r23, %r8, %r2;
mul.wide.s32 %rd10, %r23, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.3 i in range(0, 6, 1):
for.size.2 j in range(0, 2, 1):
ld.global.nc.f32 %f(33+i*2+j*1), [%rd5+(0+i*128+j*4)];
st.shared.f32 [%r3+(0+i*128+j*4)], %f(33+i*2+j*1);
setp.gt.s32 %p1, %r1, 1;
@%p1 bra $L__BB0_3;
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(45+i*1), [%rd5+(768+i*4)];
st.shared.f32 [%r3+(768+i*4)], %f(45+i*1);
$L__BB0_3:
mov.u32 %r108, 256;
mov.u32 %r106, %r1;
mov.u32 %r107, %r5;
$L__BB0_4:
mul.hi.s32 %r25, %r106, 1402438301;
shr.u32 %r26, %r25, 31;
shr.s32 %r27, %r25, 6;
add.s32 %r28, %r27, %r26;
mul.lo.s32 %r29, %r28, 196;
sub.s32 %r30, %r106, %r29;
add.s32 %r31, %r30, %r8;
mad.lo.s32 %r32, %r28, 784, %r31;
mul.wide.s32 %rd11, %r32, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f47, [%rd12];
add.s32 %r33, %r6, %r108;
for.size.13 i in range(0, 4, 1):
st.shared.f32 [%r33+(-256+i*64)], %f(47+i*1);
add.s32 %r(34+i*9), %r106, (16+i*16);
mul.hi.s32 %r(35+i*9), %r(34+i*9), 1402438301;
shr.u32 %r(36+i*9), %r(35+i*9), 31;
shr.s32 %r(37+i*9), %r(35+i*9), 6;
add.s32 %r(38+i*9), %r(37+i*9), %r(36+i*9);
mul.lo.s32 %r(39+i*9), %r(38+i*9), 196;
sub.s32 %r(40+i*9), %r(34+i*9), %r(39+i*9);
add.s32 %r(41+i*9), %r(40+i*9), %r8;
mad.lo.s32 %r(42+i*9), %r(38+i*9), 784, %r(41+i*9);
mul.wide.s32 %rd(13+i*2), %r(42+i*9), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(48+i*1), [%rd(14+i*2)+0];
for.size.13 i in range(0, 2, 1):
st.shared.f32 [%r33+(0+i*64)], %f(51+i*1);
add.s32 %r(70+i*9), %r107, (-32+i*16);
mul.hi.s32 %r(71+i*9), %r(70+i*9), 1402438301;
shr.u32 %r(72+i*9), %r(71+i*9), 31;
shr.s32 %r(73+i*9), %r(71+i*9), 6;
add.s32 %r(74+i*9), %r(73+i*9), %r(72+i*9);
mul.lo.s32 %r(75+i*9), %r(74+i*9), 196;
sub.s32 %r(76+i*9), %r(70+i*9), %r(75+i*9);
add.s32 %r(77+i*9), %r(76+i*9), %r8;
mad.lo.s32 %r(78+i*9), %r(74+i*9), 784, %r(77+i*9);
mul.wide.s32 %rd(21+i*2), %r(78+i*9), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(52+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r33+128], %f53;
mul.hi.s32 %r88, %r107, 1402438301;
shr.u32 %r89, %r88, 31;
shr.s32 %r90, %r88, 6;
add.s32 %r91, %r90, %r89;
mul.lo.s32 %r92, %r91, 196;
sub.s32 %r93, %r107, %r92;
add.s32 %r94, %r93, %r8;
mad.lo.s32 %r95, %r91, 784, %r94;
mul.wide.s32 %rd25, %r95, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f54, [%rd26];
st.shared.f32 [%r33+192], %f54;
add.s32 %r108, %r108, 512;
setp.ne.s32 %p2, %r108, 100608;
for.size.1 i in range(0, 2, 1):
add.s32 %r(13+i*93), %r107, (128+i*-112);
mov.u32 %r107, %r13;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r109, 0;
$L__BB0_6:
add.s32 %r97, %r109, %r4;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(98+i*3), %r(97+i*12), 2;
add.s32 %r(100+i*3), %r(22+i*-2), %r(98+i*3);
ld.shared.f32 %f(55+i*1), [%r(100+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(57+i*2), %f56, %f(55+i*3), %f(147+i*-1);
ld.shared.f32 %f(58+i*2), [%r100+(784+i*784)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(61+i*2), %f56, %f(60+i*2), %f(145+i*-1);
ld.shared.f32 %f(62+i*2), [%r100+(2352+i*784)];
for.size.17 i in range(0, 2, 1):
fma.rn.f32 %f(71+i*17), %f(56+i*17), %f(70+i*17), %f(140+i*-69);
ld.shared.f32 %f(72+i*17), [%r100+(4+i*4)];
ld.shared.f32 %f(73+i*17), [%r103+(4+i*4)];
fma.rn.f32 %f(74+i*17), %f(73+i*17), %f(72+i*17), %f(57+i*17);
ld.shared.f32 %f(75+i*17), [%r100+(788+i*4)];
fma.rn.f32 %f(76+i*17), %f(73+i*17), %f(75+i*17), %f(59+i*17);
ld.shared.f32 %f(77+i*17), [%r100+(1572+i*4)];
fma.rn.f32 %f(78+i*17), %f(73+i*17), %f(77+i*17), %f(61+i*17);
ld.shared.f32 %f(79+i*17), [%r100+(2356+i*4)];
fma.rn.f32 %f(80+i*17), %f(73+i*17), %f(79+i*17), %f(63+i*17);
ld.shared.f32 %f(81+i*17), [%r100+(3140+i*4)];
fma.rn.f32 %f(82+i*17), %f(73+i*17), %f(81+i*17), %f(65+i*17);
ld.shared.f32 %f(83+i*17), [%r100+(3924+i*4)];
fma.rn.f32 %f(84+i*17), %f(73+i*17), %f(83+i*17), %f(67+i*17);
ld.shared.f32 %f(85+i*17), [%r100+(4708+i*4)];
fma.rn.f32 %f(86+i*17), %f(73+i*17), %f(85+i*17), %f(69+i*17);
ld.shared.f32 %f(87+i*17), [%r100+(5492+i*4)];
fma.rn.f32 %f105, %f90, %f104, %f88;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(106+i*1), [%r(100+i*3)+12];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(147+i*-1), %f107, %f(106+i*2), %f(91+i*2);
ld.shared.f32 %f(108+i*1), [%r100+(796+i*784)];
for.size.2 i in range(0, 5, 1):
fma.rn.f32 %f(145+i*-1), %f107, %f(109+i*1), %f(95+i*2);
ld.shared.f32 %f(110+i*1), [%r100+(2364+i*784)];
fma.rn.f32 %f140, %f107, %f114, %f105;
add.s32 %r109, %r109, 4;
setp.ne.s32 %p3, %r109, 196;
@%p3 bra $L__BB0_6;
add.s32 %r105, %r105, 1;
setp.lt.u32 %p4, %r105, 4;
@%p4 bra $L__BB0_1;
mul.wide.s32 %rd27, %r19, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f115, [%rd28];
add.f32 %f116, %f147, %f115;
mov.f32 %f117, 0f00000000;
max.f32 %f118, %f116, %f117;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 7, 1):
st.global.f32 [%rd29+(0+i*4)], %f(118+i*3);
ld.global.nc.f32 %f(119+i*3), [%rd28+(4+i*4)];
add.f32 %f(120+i*3), %f(146+i*-1), %f(119+i*3);
max.f32 %f(121+i*3), %f(120+i*3), %f117;
st.global.f32 [%rd29+28], %f139;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[8];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[25088];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x * 2;
int index1 = k_outer_outer * 196 + (int)threadIdx.x * 2;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 1] = placeholder[index1 + 1];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 33] = placeholder[index1 + 33];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 65] = placeholder[index1 + 65];
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
placeholder_d_shared[index0 + 97] = placeholder[index1 + 97];
placeholder_d_shared[index0 + 128] = placeholder[index1 + 128];
placeholder_d_shared[index0 + 129] = placeholder[index1 + 129];
placeholder_d_shared[index0 + 160] = placeholder[index1 + 160];
placeholder_d_shared[index0 + 161] = placeholder[index1 + 161];
if ((int)threadIdx.x < 2) {
placeholder_d_shared[index0 + 192] = placeholder[index1 + 192];
}
if ((int)threadIdx.x < 2) {
placeholder_d_shared[index0 + 193] = placeholder[index1 + 193];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1568; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 16 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 16 + (int)threadIdx.x) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer * 16 + (int)threadIdx.x) % 196;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 1568 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 196]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 392]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 588]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 980]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1176]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1372]);
}
}
for (int ax1_inner = 0; ax1_inner < 8; ++ax1_inner) {
int index6 = (int)threadIdx.x * 8 + ax1_inner;
int index7 = ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp_huawiqr | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<71>;
.reg .b32 %r<22>;
.reg .b64 %rd<23>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[2048];
ld.param.u64 %rd9, [default_function_kernel0_param_0];
ld.param.u64 %rd10, [default_function_kernel0_param_1];
ld.param.u64 %rd11, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd10;
cvta.to.global.u64 %rd1, %rd11;
mov.u32 %r1, %tid.x;
shl.b32 %r9, %r1, 2;
mov.u32 %r10, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r10, %r9;
shr.s32 %r11, %r1, 3;
and.b32 %r12, %r1, 7;
mov.u32 %r13, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r13, %r9;
shl.b32 %r14, %r1, 6;
add.s32 %r4, %r13, %r14;
cvta.to.global.u64 %rd13, %rd9;
mul.wide.s32 %rd14, %r1, 4;
add.s64 %rd22, %rd13, %rd14;
mov.u32 %r5, %ctaid.x;
mul.lo.s32 %r15, %r11, 784;
mad.lo.s32 %r16, %r5, 50176, %r15;
or.b32 %r17, %r16, %r12;
mul.wide.s32 %rd15, %r17, 4;
add.s64 %rd16, %rd12, %rd15;
add.s64 %rd21, %rd16, 100352;
mov.u32 %r21, 0;
mov.f32 %f69, 0f00000000;
mov.f32 %f70, %f69;
$L__BB0_1:
bar.sync 0;
setp.gt.s32 %p1, %r1, 7;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f7, [%rd22];
st.shared.f32 [%r2], %f7;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(8+i*1), [%rd21+(-100352+i*12544)];
st.shared.f32 [%r3+(0+i*128)], %f(8+i*1);
bar.sync 0;
for.size.5 i in range(0, 2, 1):
ld.shared.f32 %f(24+i*5), [%r4+(0+i*4)];
ld.shared.f32 %f(25+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(26+i*5), %f(25+i*5), %f(24+i*5), %f(70+i*-44);
ld.shared.f32 %f(27+i*5), [%r4+(32+i*4)];
fma.rn.f32 %f(28+i*5), %f(25+i*5), %f(27+i*5), %f(69+i*-41);
for.size.5 i in range(0, 5, 1):
ld.shared.f32 %f(34+i*5), [%r4+(8+i*4)];
ld.shared.f32 %f(35+i*5), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(36+i*5), %f(35+i*5), %f(34+i*5), %f(31+i*5);
ld.shared.f32 %f(37+i*5), [%r4+(40+i*4)];
fma.rn.f32 %f(38+i*5), %f(35+i*5), %f(37+i*5), %f(33+i*5);
ld.shared.f32 %f59, [%r4+28];
ld.shared.f32 %f60, [_ZZ24default_function_kernel0E20placeholder_d_shared+28];
fma.rn.f32 %f70, %f60, %f59, %f56;
ld.shared.f32 %f61, [%r4+60];
fma.rn.f32 %f69, %f60, %f61, %f58;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(22+i*-1), %rd(22+i*-1), 32;
add.s32 %r21, %r21, 1;
setp.ne.s32 %p2, %r21, 98;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd17, %rd8;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(18+i*1), %r(1+i*4), (1+i*5);
add.s32 %r20, %r19, %r18;
mul.wide.s32 %rd18, %r20, 4;
add.s64 %rd19, %rd17, %rd18;
ld.global.nc.f32 %f62, [%rd19];
add.f32 %f63, %f70, %f62;
mov.f32 %f64, 0f00000000;
max.f32 %f65, %f63, %f64;
add.s64 %rd20, %rd1, %rd18;
st.global.f32 [%rd20], %f65;
ld.global.nc.f32 %f66, [%rd19+4];
add.f32 %f67, %f69, %f66;
max.f32 %f68, %f67, %f64;
st.global.f32 [%rd20+4], %f68;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[512];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 8) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 8 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index2 = (int)blockIdx.x * 50176 + ((int)threadIdx.x >> 3) * 784 + k_outer_outer * 8 + ((int)threadIdx.x & 7);
placeholder_shared[index0] = placeholder1[index2];
placeholder_shared[index0 + 32] = placeholder1[index2 + 3136];
placeholder_shared[index0 + 64] = placeholder1[index2 + 6272];
placeholder_shared[index0 + 96] = placeholder1[index2 + 9408];
placeholder_shared[index0 + 128] = placeholder1[index2 + 12544];
placeholder_shared[index0 + 160] = placeholder1[index2 + 15680];
placeholder_shared[index0 + 192] = placeholder1[index2 + 18816];
placeholder_shared[index0 + 224] = placeholder1[index2 + 21952];
placeholder_shared[index0 + 256] = placeholder1[index2 + 25088];
placeholder_shared[index0 + 288] = placeholder1[index2 + 28224];
placeholder_shared[index0 + 320] = placeholder1[index2 + 31360];
placeholder_shared[index0 + 352] = placeholder1[index2 + 34496];
placeholder_shared[index0 + 384] = placeholder1[index2 + 37632];
placeholder_shared[index0 + 416] = placeholder1[index2 + 40768];
placeholder_shared[index0 + 448] = placeholder1[index2 + 43904];
placeholder_shared[index0 + 480] = placeholder1[index2 + 47040];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 16];
int index3 = (int)threadIdx.x * 16 + 8;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index3];
int index4 = (int)threadIdx.x * 16 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index4];
int index5 = (int)threadIdx.x * 16 + 9;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index5];
int index6 = (int)threadIdx.x * 16 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index6];
int index7 = (int)threadIdx.x * 16 + 10;
T_dense[1] = T_dense[1] + placeholder_d_shared[2] * placeholder_shared[index7];
int index8 = (int)threadIdx.x * 16 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index8];
int index9 = (int)threadIdx.x * 16 + 11;
T_dense[1] = T_dense[1] + placeholder_d_shared[3] * placeholder_shared[index9];
int index10 = (int)threadIdx.x * 16 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index10];
int index11 = (int)threadIdx.x * 16 + 12;
T_dense[1] = T_dense[1] + placeholder_d_shared[4] * placeholder_shared[index11];
int index12 = (int)threadIdx.x * 16 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index12];
int index13 = (int)threadIdx.x * 16 + 13;
T_dense[1] = T_dense[1] + placeholder_d_shared[5] * placeholder_shared[index13];
int index14 = (int)threadIdx.x * 16 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index14];
int index15 = (int)threadIdx.x * 16 + 14;
T_dense[1] = T_dense[1] + placeholder_d_shared[6] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 16 + 7;
T_dense[0] = T_dense[0] + placeholder_d_shared[7] * placeholder_shared[index16];
int index17 = (int)threadIdx.x * 16 + 15;
T_dense[1] = T_dense[1] + placeholder_d_shared[7] * placeholder_shared[index17];
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index18 = (int)blockIdx.x * 64 + (int)threadIdx.x * 2 + ax1_inner;
int index19 = ax1_inner;
T_relu[index18] = max(T_dense[index19] + placeholder2[index18], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpi9mjttrd | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<531>;
.reg .b32 %r<53>;
.reg .b64 %rd<28>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[401408];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd11, [default_function_kernel0_param_1];
cvta.to.global.u64 %rd14, %rd10;
cvta.to.global.u64 %rd2, %rd11;
mov.u32 %r1, %tid.x;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(16+i*1), %r1, (1+i*2);
mov.u32 %r18, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r19, %r18, %r17;
add.s32 %r48, %r19, 32;
mul.wide.s32 %rd15, %r16, 4;
add.s64 %rd16, %rd14, %rd15;
add.s64 %rd26, %rd16, 32;
mov.u32 %r49, 0;
$L__BB0_1:
for.size.2 i in range(0, 4, 1):
ld.global.nc.v2.u32 {%r(20+i*4), %r(21+i*4)}, [%rd26+(-32+i*16)];
st.shared.v2.u32 [%r48+(-32+i*16)], {%r(20+i*4), %r(21+i*4)};
add.s32 %r48, %r48, 64;
add.s64 %rd26, %rd26, 64;
add.s32 %r49, %r49, 4;
setp.ne.s32 %p1, %r49, 196;
@%p1 bra $L__BB0_1;
shl.b32 %r37, %r1, 2;
mov.u32 %r38, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r39, %r38, %r37;
add.s32 %r50, %r39, 64;
mul.wide.s32 %rd17, %r1, 4;
add.s64 %rd18, %rd2, %rd17;
add.s64 %rd27, %rd18, 64;
mov.u32 %r51, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(129+i*1), [%rd27+(-64+i*8)];
st.shared.f32 [%r50+(-64+i*8)], %f(129+i*1);
add.s32 %r50, %r50, 128;
add.s64 %rd27, %rd27, 128;
add.s32 %r51, %r51, 16;
setp.ne.s32 %p2, %r51, 50176;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r12, %r1, 200704, %r38;
mov.f32 %f467, 0f00000000;
mov.u32 %r52, 100352;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(468+i*1), %f467;
$L__BB0_5:
mov.u32 %r46, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.1 i in range(0, 2, 1):
add.s32 %r(43+i*1), %r(46+i*-34), %r52;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(209+i*1), [%r(44+i*-1)+-100352];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(530+i*-1), %f210, %f(209+i*2), %f(530+i*-1);
ld.shared.f32 %f(211+i*1), [%r44+(-97216+i*3136)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(528+i*-1), %f210, %f(212+i*1), %f(528+i*-1);
ld.shared.f32 %f(213+i*1), [%r44+(-90944+i*3136)];
fma.rn.f32 %f467, %f210, %f273, %f467;
add.s32 %r52, %r52, 4;
setp.ne.s32 %p3, %r52, 103488;
@%p3 bra $L__BB0_5;
ld.param.u64 %rd25, [default_function_kernel0_param_2];
cvta.to.global.u64 %rd24, %rd25;
ld.param.u64 %rd23, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd22, %rd23;
mov.u32 %r47, %tid.x;
shl.b32 %r45, %r47, 6;
mul.wide.s32 %rd19, %r45, 4;
add.s64 %rd20, %rd22, %rd19;
ld.global.nc.f32 %f274, [%rd20];
add.f32 %f275, %f530, %f274;
mov.f32 %f276, 0f00000000;
max.f32 %f277, %f275, %f276;
add.s64 %rd21, %rd24, %rd19;
for.size.4 i in range(0, 63, 1):
st.global.f32 [%rd21+(0+i*4)], %f(277+i*3);
ld.global.nc.f32 %f(278+i*3), [%rd20+(4+i*4)];
add.f32 %f(279+i*3), %f(529+i*-1), %f(278+i*3);
max.f32 %f(280+i*3), %f(279+i*3), %f276;
st.global.f32 [%rd21+252], %f466;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[100352];
for (int j_outer_inner_init = 0; j_outer_inner_init < 4; ++j_outer_inner_init) {
int index0 = j_outer_inner_init * 16;
T_dense[index0] = 0.000000e+00f;
T_dense[index0 + 1] = 0.000000e+00f;
T_dense[index0 + 2] = 0.000000e+00f;
T_dense[index0 + 3] = 0.000000e+00f;
T_dense[index0 + 4] = 0.000000e+00f;
T_dense[index0 + 5] = 0.000000e+00f;
T_dense[index0 + 6] = 0.000000e+00f;
T_dense[index0 + 7] = 0.000000e+00f;
T_dense[index0 + 8] = 0.000000e+00f;
T_dense[index0 + 9] = 0.000000e+00f;
T_dense[index0 + 10] = 0.000000e+00f;
T_dense[index0 + 11] = 0.000000e+00f;
T_dense[index0 + 12] = 0.000000e+00f;
T_dense[index0 + 13] = 0.000000e+00f;
T_dense[index0 + 14] = 0.000000e+00f;
T_dense[index0 + 15] = 0.000000e+00f;
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 196; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_d_shared + ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder + ax0_ax1_fused_outer_outer * 4 + (int)threadIdx.x * 2))[0];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 50176; ++ax0_ax1_fused_outer_outer1) {
int index1 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index1];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 784; ++k_outer_inner) {
for (int j_outer_inner = 0; j_outer_inner < 4; ++j_outer_inner) {
int index2 = j_outer_inner * 16;
int index3 = k_outer_inner;
int index4 = (int)threadIdx.x * 50176 + j_outer_inner * 12544 + k_outer_inner;
T_dense[index2] = T_dense[index2] + placeholder_d_shared[index3] * placeholder_shared[index4];
T_dense[index2 + 1] = T_dense[index2 + 1] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 784]);
T_dense[index2 + 2] = T_dense[index2 + 2] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 1568]);
T_dense[index2 + 3] = T_dense[index2 + 3] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 2352]);
T_dense[index2 + 4] = T_dense[index2 + 4] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 3136]);
T_dense[index2 + 5] = T_dense[index2 + 5] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 3920]);
T_dense[index2 + 6] = T_dense[index2 + 6] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 4704]);
T_dense[index2 + 7] = T_dense[index2 + 7] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 5488]);
T_dense[index2 + 8] = T_dense[index2 + 8] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 6272]);
T_dense[index2 + 9] = T_dense[index2 + 9] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 7056]);
T_dense[index2 + 10] = T_dense[index2 + 10] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 7840]);
T_dense[index2 + 11] = T_dense[index2 + 11] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 8624]);
T_dense[index2 + 12] = T_dense[index2 + 12] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 9408]);
T_dense[index2 + 13] = T_dense[index2 + 13] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 10192]);
T_dense[index2 + 14] = T_dense[index2 + 14] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 10976]);
T_dense[index2 + 15] = T_dense[index2 + 15] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 11760]);
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index5 = (int)threadIdx.x * 64 + ax1_inner;
int index6 = ax1_inner;
T_relu[index5] = max(T_dense[index6] + placeholder2[index5], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp1pnzsprn | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<325>;
.reg .b32 %r<182>;
.reg .b64 %rd<34>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[50176];
ld.param.u64 %rd8, [default_function_kernel0_param_0];
ld.param.u64 %rd9, [default_function_kernel0_param_1];
ld.param.u64 %rd10, [default_function_kernel0_param_2];
ld.param.u64 %rd11, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd12, %rd8;
cvta.to.global.u64 %rd1, %rd9;
cvta.to.global.u64 %rd2, %rd10;
cvta.to.global.u64 %rd3, %rd11;
mov.u32 %r1, %tid.x;
shl.b32 %r2, %r1, 2;
mov.u32 %r3, %ctaid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(4+i*1), %r(3+i*-2), (50176+i*-49784);
add.s64 %rd4, %rd12, 8;
mov.u32 %r21, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r22, %r21, %r2;
add.s32 %r6, %r22, 8;
mov.u32 %r20, 0;
mov.f32 %f261, 0f00000000;
for.size.1 i in range(0, 31, 1):
mov.f32 %f(262+i*1), %f261;
mov.u32 %r177, %r20;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r9, %r177, 196;
add.s32 %r24, %r1, %r9;
mul.wide.s32 %rd13, %r24, 4;
add.s64 %rd33, %rd4, %rd13;
mov.u32 %r178, %r6;
mov.u32 %r179, %r20;
$L__BB0_2:
for.size.2 i in range(0, 2, 1):
ld.global.nc.f32 %f(129+i*1), [%rd33+(-8+i*8)];
st.shared.f32 [%r178+(-8+i*8)], %f(129+i*1);
add.s64 %rd33, %rd33, 16;
for.size.1 i in range(0, 2, 1):
add.s32 %r(178+i*1), %r(178+i*1), (16+i*-14);
setp.ne.s32 %p1, %r179, 98;
@%p1 bra $L__BB0_2;
add.s32 %r14, %r9, %r4;
mov.u32 %r180, 0;
$L__BB0_4:
shl.b32 %r26, %r180, 3;
add.s32 %r27, %r26, %r2;
shl.b32 %r28, %r27, 2;
mov.u32 %r29, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r30, %r29, %r28;
for.size.13 i in range(0, 2, 1):
mul.hi.s32 %r(31+i*17), %r(27+i*20), 1402438301;
shr.u32 %r(32+i*17), %r(31+i*17), 31;
shr.s32 %r(33+i*17), %r(31+i*17), 6;
add.s32 %r(34+i*17), %r(33+i*17), %r(32+i*17);
mul.lo.s32 %r(35+i*17), %r(34+i*17), 196;
sub.s32 %r(36+i*17), %r(27+i*20), %r(35+i*17);
add.s32 %r(37+i*17), %r14, %r(36+i*17);
mad.lo.s32 %r(38+i*17), %r(34+i*17), 784, %r(37+i*17);
mul.wide.s32 %rd(14+i*2), %r(38+i*17), 4;
add.s64 %rd(15+i*2), %rd1, %rd(14+i*2);
ld.global.nc.v4.u32 {%r(39+i*17), %r(40+i*17), %r(41+i*17), %r(42+i*17)}, [%rd(15+i*2)+0];
st.shared.v4.u32 [%r30+(0+i*32)], {%r(39+i*17), %r(40+i*17), %r(41+i*17), %r(42+i*17)};
add.s32 %r(47+i*17), %r27, (8+i*8);
for.size.13 i in range(0, 5, 1):
mul.hi.s32 %r(65+i*17), %r(64+i*17), 1402438301;
shr.u32 %r(66+i*17), %r(65+i*17), 31;
shr.s32 %r(67+i*17), %r(65+i*17), 6;
add.s32 %r(68+i*17), %r(67+i*17), %r(66+i*17);
mul.lo.s32 %r(69+i*17), %r(68+i*17), 196;
sub.s32 %r(70+i*17), %r(64+i*17), %r(69+i*17);
add.s32 %r(71+i*17), %r14, %r(70+i*17);
mad.lo.s32 %r(72+i*17), %r(68+i*17), 784, %r(71+i*17);
mul.wide.s32 %rd(18+i*2), %r(72+i*17), 4;
add.s64 %rd(19+i*2), %rd1, %rd(18+i*2);
ld.global.nc.v4.u32 {%r(73+i*17), %r(74+i*17), %r(75+i*17), %r(76+i*17)}, [%rd(19+i*2)+0];
st.shared.v4.u32 [%r30+(64+i*32)], {%r(73+i*17), %r(74+i*17), %r(75+i*17), %r(76+i*17)};
add.s32 %r(81+i*17), %r27, (24+i*8);
mul.hi.s32 %r150, %r149, 1402438301;
shr.u32 %r151, %r150, 31;
shr.s32 %r152, %r150, 6;
add.s32 %r153, %r152, %r151;
mul.lo.s32 %r154, %r153, 196;
sub.s32 %r155, %r149, %r154;
add.s32 %r156, %r14, %r155;
mad.lo.s32 %r157, %r153, 784, %r156;
mul.wide.s32 %rd28, %r157, 4;
add.s64 %rd29, %rd1, %rd28;
ld.global.nc.v4.u32 {%r158, %r159, %r160, %r161}, [%rd29];
st.shared.v4.u32 [%r30+224], {%r158, %r159, %r160, %r161};
add.s32 %r180, %r180, 8;
setp.ne.s32 %p2, %r180, 1568;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r181, 0;
$L__BB0_6:
add.s32 %r167, %r181, %r5;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(168+i*3), %r(167+i*14), 2;
add.s32 %r(170+i*3), %r(29+i*-8), %r(168+i*3);
ld.shared.f32 %f(131+i*1), [%r(170+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(292+i*-2), %f132, %f(131+i*2), %f(292+i*-2);
ld.shared.f32 %f(133+i*1), [%r170+(3136+i*3136)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(288+i*-2), %f132, %f(134+i*1), %f(288+i*-2);
ld.shared.f32 %f(135+i*1), [%r170+(9408+i*3136)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(262+i*29), %f132, %f(147+i*1), %f(262+i*29);
ld.shared.f32 %f(148+i*1), [%r170+(784+i*3136)];
for.size.2 i in range(0, 14, 1):
fma.rn.f32 %f(289+i*-2), %f132, %f(149+i*1), %f(289+i*-2);
ld.shared.f32 %f(150+i*1), [%r170+(7056+i*3136)];
fma.rn.f32 %f261, %f132, %f163, %f261;
add.s32 %r181, %r181, 1;
setp.ne.s32 %p3, %r181, 196;
@%p3 bra $L__BB0_6;
add.s32 %r177, %r177, 1;
setp.lt.u32 %p4, %r177, 4;
@%p4 bra $L__BB0_1;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(174+i*1), %r(3+i*-2), (6+i*-5);
add.s32 %r176, %r175, %r174;
mul.wide.s32 %rd30, %r176, 4;
add.s64 %rd31, %rd3, %rd30;
ld.global.nc.f32 %f164, [%rd31];
add.f32 %f165, %f292, %f164;
mov.f32 %f166, 0f00000000;
max.f32 %f167, %f165, %f166;
add.s64 %rd32, %rd2, %rd30;
for.size.4 i in range(0, 15, 1):
st.global.f32 [%rd32+(0+i*16)], %f(167+i*3);
ld.global.nc.f32 %f(168+i*3), [%rd31+(16+i*16)];
add.f32 %f(169+i*3), %f(290+i*-2), %f(168+i*3);
max.f32 %f(170+i*3), %f(169+i*3), %f166;
for.size.4 i in range(0, 2, 1):
st.global.f32 [%rd32+(240+i*-236)], %f(212+i*3);
ld.global.nc.f32 %f(213+i*3), [%rd31+(4+i*16)];
add.f32 %f(214+i*3), %f(291+i*-2), %f(213+i*3);
max.f32 %f(215+i*3), %f(214+i*3), %f166;
for.size.4 i in range(0, 14, 1):
st.global.f32 [%rd32+(20+i*16)], %f(218+i*3);
ld.global.nc.f32 %f(219+i*3), [%rd31+(36+i*16)];
add.f32 %f(220+i*3), %f(287+i*-2), %f(219+i*3);
max.f32 %f(221+i*3), %f(220+i*3), %f166;
st.global.f32 [%rd32+244], %f260;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[32];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[12544];
for (int j_outer_inner_init = 0; j_outer_inner_init < 2; ++j_outer_inner_init) {
int index0 = j_outer_inner_init;
T_dense[index0] = 0.000000e+00f;
T_dense[index0 + 2] = 0.000000e+00f;
T_dense[index0 + 4] = 0.000000e+00f;
T_dense[index0 + 6] = 0.000000e+00f;
T_dense[index0 + 8] = 0.000000e+00f;
T_dense[index0 + 10] = 0.000000e+00f;
T_dense[index0 + 12] = 0.000000e+00f;
T_dense[index0 + 14] = 0.000000e+00f;
T_dense[index0 + 16] = 0.000000e+00f;
T_dense[index0 + 18] = 0.000000e+00f;
T_dense[index0 + 20] = 0.000000e+00f;
T_dense[index0 + 22] = 0.000000e+00f;
T_dense[index0 + 24] = 0.000000e+00f;
T_dense[index0 + 26] = 0.000000e+00f;
T_dense[index0 + 28] = 0.000000e+00f;
T_dense[index0 + 30] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index2 = k_outer_outer * 196 + ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 1568; ++ax0_ax1_fused_outer_outer1) {
((float4*)(placeholder_shared + ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 4))[0] = ((float4*)(placeholder1 + (int)blockIdx.x * 50176 + (ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 4) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x * 4) % 196))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
for (int j_outer_inner = 0; j_outer_inner < 2; ++j_outer_inner) {
int index3 = j_outer_inner;
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 392 + j_outer_inner * 196 + k_outer_inner;
T_dense[index3] = T_dense[index3] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[index3 + 2] = T_dense[index3 + 2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 784]);
T_dense[index3 + 4] = T_dense[index3 + 4] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 1568]);
T_dense[index3 + 6] = T_dense[index3 + 6] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 2352]);
T_dense[index3 + 8] = T_dense[index3 + 8] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3136]);
T_dense[index3 + 10] = T_dense[index3 + 10] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3920]);
T_dense[index3 + 12] = T_dense[index3 + 12] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 4704]);
T_dense[index3 + 14] = T_dense[index3 + 14] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 5488]);
T_dense[index3 + 16] = T_dense[index3 + 16] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 6272]);
T_dense[index3 + 18] = T_dense[index3 + 18] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7056]);
T_dense[index3 + 20] = T_dense[index3 + 20] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7840]);
T_dense[index3 + 22] = T_dense[index3 + 22] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 8624]);
T_dense[index3 + 24] = T_dense[index3 + 24] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 9408]);
T_dense[index3 + 26] = T_dense[index3 + 26] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10192]);
T_dense[index3 + 28] = T_dense[index3 + 28] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10976]);
T_dense[index3 + 30] = T_dense[index3 + 30] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 11760]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
for (int vthread_s = 0; vthread_s < 16; ++vthread_s) {
int index6 = (int)blockIdx.x * 64 + vthread_s * 4 + (int)threadIdx.x * 2 + ax1_inner;
int index7 = vthread_s * 2 + ax1_inner;
T_relu[index6] = max(T_dense[index7] + placeholder2[index6], 0.000000e+00f);
}
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpgofxsm61 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<51>;
.reg .b32 %r<80>;
.reg .b64 %rd<22>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[784];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[6272];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
shl.b32 %r24, %r1, 2;
mov.u32 %r25, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r25, %r24;
mov.u32 %r3, %ctaid.x;
for.size.1 i in range(0, 2, 1):
mul.lo.s32 %r(4+i*1), %r(3+i*-2), (6272+i*-6076);
add.s32 %r6, %r1, 24;
mov.u32 %r26, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r27, %r26, %r24;
add.s32 %r7, %r27, 64;
mov.f32 %f49, 0f00000000;
mov.u32 %r74, 0;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r9, %r74, 196;
add.s32 %r28, %r9, %r1;
mul.wide.s32 %rd10, %r28, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.2 i in range(0, 24, 1):
ld.global.nc.f32 %f(5+i*1), [%rd5+(0+i*32)];
st.shared.f32 [%r2+(0+i*32)], %f(5+i*1);
setp.gt.s32 %p1, %r1, 3;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f29, [%rd5+768];
st.shared.f32 [%r2+768], %f29;
$L__BB0_3:
add.s32 %r11, %r9, %r4;
mov.u32 %r78, 64;
mov.u32 %r75, %r1;
mov.u32 %r76, %r7;
mov.u32 %r77, %r6;
$L__BB0_4:
for.size.13 i in range(0, 2, 1):
mul.hi.s32 %r(30+i*9), %r(75+i*-37), 1402438301;
shr.u32 %r(31+i*9), %r(30+i*9), 31;
shr.s32 %r(32+i*9), %r(30+i*9), 6;
add.s32 %r(33+i*9), %r(32+i*9), %r(31+i*9);
mul.lo.s32 %r(34+i*9), %r(33+i*9), 196;
sub.s32 %r(35+i*9), %r(75+i*-37), %r(34+i*9);
add.s32 %r(36+i*9), %r11, %r(35+i*9);
mad.lo.s32 %r(37+i*9), %r(33+i*9), 784, %r(36+i*9);
mul.wide.s32 %rd(11+i*2), %r(37+i*9), 4;
add.s64 %rd(12+i*2), %rd1, %rd(11+i*2);
ld.global.nc.f32 %f(30+i*1), [%rd(12+i*2)+0];
st.shared.f32 [%r76+(-64+i*32)], %f(30+i*1);
add.s32 %r(38+i*9), %r77, (-16+i*8);
for.size.12 i in range(0, 2, 1):
mul.hi.s32 %r(48+i*8), %r(47+i*30), 1402438301;
shr.u32 %r(49+i*8), %r(48+i*8), 31;
shr.s32 %r(50+i*8), %r(48+i*8), 6;
add.s32 %r(51+i*8), %r(50+i*8), %r(49+i*8);
mul.lo.s32 %r(52+i*8), %r(51+i*8), 196;
sub.s32 %r(53+i*8), %r(47+i*30), %r(52+i*8);
add.s32 %r(54+i*8), %r11, %r(53+i*8);
mad.lo.s32 %r(55+i*8), %r(51+i*8), 784, %r(54+i*8);
mul.wide.s32 %rd(15+i*2), %r(55+i*8), 4;
add.s64 %rd(16+i*2), %rd1, %rd(15+i*2);
ld.global.nc.f32 %f(32+i*1), [%rd(16+i*2)+0];
st.shared.f32 [%r76+(0+i*32)], %f(32+i*1);
for.size.1 i in range(0, 2, 1):
add.s32 %r(77+i*-1), %r(77+i*-1), (32+i*96);
for.size.1 i in range(0, 2, 1):
add.s32 %r(75+i*3), %r(75+i*3), (32+i*96);
setp.ne.s32 %p2, %r78, 6336;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r79, 0;
$L__BB0_6:
add.s32 %r65, %r79, %r5;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(66+i*3), %r(65+i*14), 2;
add.s32 %r(68+i*3), %r(26+i*-1), %r(66+i*3);
ld.shared.f32 %f(34+i*1), [%r(68+i*3)+0];
for.size.3 i in range(0, 2, 1):
fma.rn.f32 %f(36+i*3), %f(35+i*3), %f(34+i*3), %f(49+i*-13);
ld.shared.f32 %f(37+i*3), [%r68+(4+i*4)];
ld.shared.f32 %f(38+i*3), [%r71+(4+i*4)];
fma.rn.f32 %f42, %f41, %f40, %f39;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(43+i*1), [%r(68+i*3)+12];
fma.rn.f32 %f49, %f44, %f43, %f42;
add.s32 %r79, %r79, 4;
setp.ne.s32 %p3, %r79, 196;
@%p3 bra $L__BB0_6;
add.s32 %r74, %r74, 1;
setp.lt.u32 %p4, %r74, 4;
@%p4 bra $L__BB0_1;
shl.b32 %r72, %r3, 3;
add.s32 %r73, %r72, %r1;
mul.wide.s32 %rd19, %r73, 4;
add.s64 %rd20, %rd4, %rd19;
ld.global.nc.f32 %f45, [%rd20];
add.f32 %f46, %f49, %f45;
mov.f32 %f47, 0f00000000;
max.f32 %f48, %f46, %f47;
add.s64 %rd21, %rd3, %rd19;
st.global.f32 [%rd21], %f48;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[196];
__shared__ float placeholder_shared[1568];
T_dense[0] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 4; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 196 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 8] = placeholder[index1 + 8];
placeholder_d_shared[index0 + 16] = placeholder[index1 + 16];
placeholder_d_shared[index0 + 24] = placeholder[index1 + 24];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 40] = placeholder[index1 + 40];
placeholder_d_shared[index0 + 48] = placeholder[index1 + 48];
placeholder_d_shared[index0 + 56] = placeholder[index1 + 56];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
placeholder_d_shared[index0 + 72] = placeholder[index1 + 72];
placeholder_d_shared[index0 + 80] = placeholder[index1 + 80];
placeholder_d_shared[index0 + 88] = placeholder[index1 + 88];
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
placeholder_d_shared[index0 + 104] = placeholder[index1 + 104];
placeholder_d_shared[index0 + 112] = placeholder[index1 + 112];
placeholder_d_shared[index0 + 120] = placeholder[index1 + 120];
placeholder_d_shared[index0 + 128] = placeholder[index1 + 128];
placeholder_d_shared[index0 + 136] = placeholder[index1 + 136];
placeholder_d_shared[index0 + 144] = placeholder[index1 + 144];
placeholder_d_shared[index0 + 152] = placeholder[index1 + 152];
placeholder_d_shared[index0 + 160] = placeholder[index1 + 160];
placeholder_d_shared[index0 + 168] = placeholder[index1 + 168];
placeholder_d_shared[index0 + 176] = placeholder[index1 + 176];
placeholder_d_shared[index0 + 184] = placeholder[index1 + 184];
if ((int)threadIdx.x < 4) {
placeholder_d_shared[index0 + 192] = placeholder[index1 + 192];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 196; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
int index3 = (int)blockIdx.x * 6272 + (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x) / 196 * 784 + k_outer_outer * 196 + (ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x) % 196;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 196; ++k_outer_inner) {
int index4 = k_outer_inner;
int index5 = (int)threadIdx.x * 196 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
}
}
int index6 = (int)blockIdx.x * 8 + (int)threadIdx.x;
T_relu[index6] = max(T_dense[0] + placeholder2[index6], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmprm677kjd | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<55>;
.reg .b32 %r<56>;
.reg .b64 %rd<18>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[392];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[50176];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd7, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd5;
cvta.to.global.u64 %rd3, %rd7;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r1, %tid.x;
shl.b32 %r18, %r1, 2;
mov.u32 %r19, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r19, %r18;
mul.lo.s32 %r3, %r1, 98;
add.s32 %r4, %r1, 128;
mov.u32 %r20, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r20, %r18;
mov.f32 %f53, 0f00000000;
mov.u32 %r51, 0;
$L__BB0_1:
bar.sync 0;
setp.lt.s32 %p1, %r1, 98;
@%p1 bra $L__BB0_2;
bra.uni $L__BB0_3;
$L__BB0_2:
mad.lo.s32 %r21, %r51, 98, %r1;
mul.wide.s32 %rd9, %r21, 4;
add.s64 %rd10, %rd2, %rd9;
ld.global.nc.f32 %f5, [%rd10];
st.shared.f32 [%r2], %f5;
$L__BB0_3:
mul.lo.s32 %r7, %r51, 98;
mov.u32 %r54, 512;
mov.u32 %r52, 0;
mov.u32 %r53, %r4;
$L__BB0_4:
add.s32 %r24, %r53, -128;
mul.hi.s32 %r25, %r24, 1402438301;
shr.u32 %r26, %r25, 31;
shr.s32 %r27, %r25, 5;
add.s32 %r28, %r27, %r26;
mul.lo.s32 %r29, %r28, 98;
sub.s32 %r30, %r24, %r29;
add.s32 %r31, %r30, %r7;
mad.lo.s32 %r32, %r28, 784, %r31;
mul.wide.s32 %rd11, %r32, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f6, [%rd12];
add.s32 %r33, %r5, %r52;
st.shared.f32 [%r33], %f6;
mul.hi.s32 %r34, %r53, 1402438301;
shr.u32 %r35, %r34, 31;
shr.s32 %r36, %r34, 5;
add.s32 %r37, %r36, %r35;
mul.lo.s32 %r38, %r37, 98;
sub.s32 %r39, %r53, %r38;
add.s32 %r40, %r39, %r7;
mad.lo.s32 %r41, %r37, 784, %r40;
mul.wide.s32 %rd13, %r41, 4;
add.s64 %rd14, %rd1, %rd13;
ld.global.nc.f32 %f7, [%rd14];
st.shared.f32 [%r33+512], %f7;
for.size.1 i in range(0, 2, 1):
add.s32 %r(53+i*-1), %r(53+i*-1), (256+i*768);
add.s32 %r54, %r54, 1024;
setp.ne.s32 %p2, %r54, 50688;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r55, 0;
$L__BB0_6:
mul.lo.s32 %r43, %r55, 14;
shl.b32 %r44, %r43, 2;
for.size.1 i in range(0, 2, 1):
add.s32 %r(46+i*1), %r(19+i*24), %r(44+i*-41);
shl.b32 %r48, %r47, 2;
add.s32 %r50, %r20, %r48;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(8+i*1), [%r(50+i*-4)+0];
for.size.3 i in range(0, 2, 1):
fma.rn.f32 %f(10+i*3), %f(9+i*3), %f(8+i*3), %f(53+i*-43);
ld.shared.f32 %f(11+i*3), [%r50+(4+i*4)];
ld.shared.f32 %f(12+i*3), [%r46+(4+i*4)];
for.size.3 i in range(0, 11, 1):
fma.rn.f32 %f(16+i*3), %f(15+i*3), %f(14+i*3), %f(13+i*3);
ld.shared.f32 %f(17+i*3), [%r50+(12+i*4)];
ld.shared.f32 %f(18+i*3), [%r46+(12+i*4)];
fma.rn.f32 %f53, %f48, %f47, %f46;
add.s32 %r55, %r55, 1;
setp.ne.s32 %p3, %r55, 7;
@%p3 bra $L__BB0_6;
add.s32 %r51, %r51, 1;
setp.lt.u32 %p4, %r51, 8;
@%p4 bra $L__BB0_1;
mul.wide.s32 %rd15, %r1, 4;
add.s64 %rd16, %rd4, %rd15;
ld.global.nc.f32 %f49, [%rd16];
add.f32 %f50, %f53, %f49;
mov.f32 %f51, 0f00000000;
max.f32 %f52, %f50, %f51;
add.s64 %rd17, %rd3, %rd15;
st.global.f32 [%rd17], %f52;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[98];
__shared__ float placeholder_shared[12544];
T_dense[0] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 8; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 98) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 98 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 98; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 128 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 128 + (int)threadIdx.x) / 98 * 784 + k_outer_outer * 98 + (ax0_ax1_fused_outer_outer * 128 + (int)threadIdx.x) % 98;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 7; ++k_outer_inner) {
int index4 = k_outer_inner * 14;
int index5 = (int)threadIdx.x * 98 + k_outer_inner * 14;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 1] * placeholder_shared[index5 + 1]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 2] * placeholder_shared[index5 + 2]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 3] * placeholder_shared[index5 + 3]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 4] * placeholder_shared[index5 + 4]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 5] * placeholder_shared[index5 + 5]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 6] * placeholder_shared[index5 + 6]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 7] * placeholder_shared[index5 + 7]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 8] * placeholder_shared[index5 + 8]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 9] * placeholder_shared[index5 + 9]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 10] * placeholder_shared[index5 + 10]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 11] * placeholder_shared[index5 + 11]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 12] * placeholder_shared[index5 + 12]);
T_dense[0] = T_dense[0] + (placeholder_d_shared[index4 + 13] * placeholder_shared[index5 + 13]);
}
}
int index0 = (int)threadIdx.x;
T_relu[index0] = max(T_dense[0] + placeholder2[index0], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpt4o3s6ci | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<793>;
.reg .b32 %r<127>;
.reg .b64 %rd<32>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[112];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[14336];
ld.param.u64 %rd1, [default_function_kernel0_param_0];
ld.param.u64 %rd2, [default_function_kernel0_param_1];
mov.u32 %r13, %tid.x;
mul.lo.s32 %r1, %r13, 1792;
mov.u32 %r123, 0;
mov.f32 %f729, 0f00000000;
cvta.to.global.u64 %rd5, %rd1;
shl.b32 %r17, %r13, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r19, %r18, %r17;
cvta.to.global.u64 %rd8, %rd2;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(730+i*1), %f729;
$L__BB0_1:
mov.u32 %r124, 0;
bar.sync 0;
mul.lo.s32 %r3, %r123, 28;
add.s32 %r16, %r3, %r13;
mul.wide.s32 %rd6, %r16, 4;
add.s64 %rd7, %rd5, %rd6;
for.size.2 i in range(0, 14, 1):
ld.global.nc.f32 %f(321+i*1), [%rd7+(0+i*8)];
st.shared.f32 [%r19+(0+i*8)], %f(321+i*1);
$L__BB0_2:
mul.lo.s32 %r117, %r123, 28;
shl.b32 %r21, %r124, 1;
add.s32 %r22, %r21, %r13;
mul.hi.s32 %r23, %r22, -1840700269;
add.s32 %r24, %r23, %r22;
shr.u32 %r25, %r24, 31;
shr.s32 %r26, %r24, 4;
add.s32 %r27, %r26, %r25;
mul.lo.s32 %r28, %r27, 28;
sub.s32 %r29, %r22, %r28;
add.s32 %r30, %r29, %r117;
mad.lo.s32 %r31, %r27, 784, %r30;
mul.wide.s32 %rd9, %r31, 4;
add.s64 %rd10, %rd8, %rd9;
ld.global.nc.f32 %f335, [%rd10];
shl.b32 %r32, %r22, 2;
mov.u32 %r33, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r34, %r33, %r32;
for.size.14 i in range(0, 7, 1):
st.shared.f32 [%r34+(0+i*8)], %f(335+i*1);
add.s32 %r(35+i*10), %r22, (2+i*2);
mul.hi.s32 %r(36+i*10), %r(35+i*10), -1840700269;
add.s32 %r(37+i*10), %r(36+i*10), %r(35+i*10);
shr.u32 %r(38+i*10), %r(37+i*10), 31;
shr.s32 %r(39+i*10), %r(37+i*10), 4;
add.s32 %r(40+i*10), %r(39+i*10), %r(38+i*10);
mul.lo.s32 %r(41+i*10), %r(40+i*10), 28;
sub.s32 %r(42+i*10), %r(35+i*10), %r(41+i*10);
add.s32 %r(43+i*10), %r(42+i*10), %r117;
mad.lo.s32 %r(44+i*10), %r(40+i*10), 784, %r(43+i*10);
mul.wide.s32 %rd(11+i*2), %r(44+i*10), 4;
add.s64 %rd(12+i*2), %rd8, %rd(11+i*2);
ld.global.nc.f32 %f(336+i*1), [%rd(12+i*2)+0];
st.shared.f32 [%r34+56], %f342;
add.s32 %r124, %r124, 8;
setp.ne.s32 %p1, %r124, 1792;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r125, 0;
$L__BB0_4:
mov.u32 %r126, 0;
$L__BB0_5:
mul.lo.s32 %r121, %r125, 14;
mov.u32 %r119, _ZZ24default_function_kernel0E18placeholder_shared;
mov.u32 %r118, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.3 i in range(0, 2, 1):
add.s32 %r(107+i*4), %r(126+i*-19), %r(121+i*-120);
shl.b32 %r(108+i*4), %r(107+i*4), 2;
add.s32 %r(110+i*4), %r(118+i*1), %r(108+i*4);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(343+i*1), [%r(114+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(792+i*-1), %f344, %f(343+i*2), %f(792+i*-1);
ld.shared.f32 %f(345+i*1), [%r114+(112+i*112)];
for.size.2 i in range(0, 61, 1):
fma.rn.f32 %f(790+i*-1), %f344, %f(346+i*1), %f(790+i*-1);
ld.shared.f32 %f(347+i*1), [%r114+(336+i*112)];
fma.rn.f32 %f729, %f344, %f407, %f729;
add.s32 %r126, %r126, 1;
setp.ne.s32 %p2, %r126, 14;
@%p2 bra $L__BB0_5;
add.s32 %r125, %r125, 1;
setp.lt.u32 %p3, %r125, 2;
@%p3 bra $L__BB0_4;
add.s32 %r123, %r123, 1;
setp.lt.u32 %p4, %r123, 28;
@%p4 bra $L__BB0_1;
ld.param.u64 %rd31, [default_function_kernel0_param_2];
ld.param.u64 %rd30, [default_function_kernel0_param_3];
shl.b32 %r116, %r13, 6;
cvta.to.global.u64 %rd25, %rd30;
mul.wide.s32 %rd26, %r116, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f408, [%rd27];
add.f32 %f409, %f792, %f408;
mov.f32 %f410, 0f00000000;
max.f32 %f411, %f409, %f410;
cvta.to.global.u64 %rd28, %rd31;
add.s64 %rd29, %rd28, %rd26;
for.size.4 i in range(0, 63, 1):
st.global.f32 [%rd29+(0+i*4)], %f(411+i*3);
ld.global.nc.f32 %f(412+i*3), [%rd27+(4+i*4)];
add.f32 %f(413+i*3), %f(791+i*-1), %f(412+i*3);
max.f32 %f(414+i*3), %f(413+i*3), %f410;
st.global.f32 [%rd29+252], %f600;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[28];
__shared__ float placeholder_shared[3584];
for (int j_inner_init = 0; j_inner_init < 64; ++j_inner_init) {
int index0 = j_inner_init;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 28; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 14; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index2 = k_outer_outer * 28 + ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 1792; ++ax0_ax1_fused_outer_outer1) {
int index3 = ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x;
int index4 = (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) / 28 * 784 + k_outer_outer * 28 + (ax0_ax1_fused_outer_outer1 * 2 + (int)threadIdx.x) % 28;
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 14; ++k_inner) {
for (int j_inner = 0; j_inner < 64; ++j_inner) {
int index5 = j_inner;
int index6 = k_outer_inner * 14 + k_inner;
int index7 = (int)threadIdx.x * 1792 + j_inner * 28 + k_outer_inner * 14 + k_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
}
for (int ax1_inner = 0; ax1_inner < 64; ++ax1_inner) {
int index8 = (int)threadIdx.x * 64 + ax1_inner;
int index9 = ax1_inner;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpvbbf60js | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<36>;
.reg .b32 %r<97>;
.reg .b64 %rd<22>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[392];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[50176];
ld.param.u64 %rd2, [default_function_kernel0_param_0];
ld.param.u64 %rd5, [default_function_kernel0_param_1];
ld.param.u64 %rd3, [default_function_kernel0_param_2];
ld.param.u64 %rd4, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd5;
mov.u32 %r22, %tid.x;
shl.b32 %r1, %r22, 1;
mul.lo.s32 %r2, %r22, 98;
add.s32 %r3, %r1, 256;
shl.b32 %r23, %r22, 3;
mov.u32 %r24, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r4, %r24, %r23;
mov.f32 %f34, 0f00000000;
mov.u32 %r91, 0;
cvta.to.global.u64 %rd6, %rd2;
$L__BB0_1:
setp.gt.s32 %p1, %r22, 48;
bar.sync 0;
mul.lo.s32 %r6, %r91, 98;
@%p1 bra $L__BB0_3;
add.s32 %r26, %r6, %r1;
mul.wide.s32 %rd7, %r26, 4;
add.s64 %rd8, %rd6, %rd7;
ld.global.nc.f32 %f6, [%rd8];
mov.u32 %r29, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r30, %r29, %r23;
st.shared.f32 [%r30], %f6;
ld.global.nc.f32 %f7, [%rd8+4];
st.shared.f32 [%r30+4], %f7;
$L__BB0_3:
mov.u32 %r94, 0;
mov.u32 %r93, 2048;
mov.u32 %r92, %r1;
mov.u32 %r95, %r3;
bra.uni $L__BB0_4;
$L__BB0_5:
for.size.13 i in range(0, 2, 1):
mul.hi.s32 %r(45+i*13), %r(95+i*-38), 1402438301;
shr.u32 %r(46+i*13), %r(45+i*13), 31;
shr.s32 %r(47+i*13), %r(45+i*13), 5;
add.s32 %r(48+i*13), %r(47+i*13), %r(46+i*13);
mul.lo.s32 %r(49+i*13), %r(48+i*13), 98;
sub.s32 %r(50+i*13), %r(95+i*-38), %r(49+i*13);
add.s32 %r(51+i*13), %r(50+i*13), %r6;
mad.lo.s32 %r(52+i*13), %r(48+i*13), 784, %r(51+i*13);
mul.wide.s32 %rd(11+i*2), %r(52+i*13), 4;
add.s64 %rd(12+i*2), %rd1, %rd(11+i*2);
ld.global.nc.v2.u32 {%r(53+i*13), %r(54+i*13)}, [%rd(12+i*2)+0];
st.shared.v2.u32 [%r11+(1024+i*1024)], {%r(53+i*13), %r(54+i*13)};
add.s32 %r(57+i*13), %r95, (256+i*256);
mul.hi.s32 %r71, %r70, 1402438301;
shr.u32 %r72, %r71, 31;
shr.s32 %r73, %r71, 5;
add.s32 %r74, %r73, %r72;
mul.lo.s32 %r75, %r74, 98;
sub.s32 %r76, %r70, %r75;
add.s32 %r77, %r76, %r6;
mad.lo.s32 %r78, %r74, 784, %r77;
mul.wide.s32 %rd15, %r78, 4;
add.s64 %rd16, %rd1, %rd15;
ld.global.nc.v2.u32 {%r79, %r80}, [%rd16];
st.shared.v2.u32 [%r11+3072], {%r79, %r80};
for.size.1 i in range(0, 2, 1):
add.s32 %r(95+i*-1), %r(95+i*-1), (1024+i*3072);
for.size.1 i in range(0, 2, 1):
add.s32 %r(93+i*-1), %r(93+i*-1), (4096+i*-3072);
$L__BB0_4:
mul.hi.s32 %r33, %r92, 1402438301;
shr.u32 %r34, %r33, 31;
shr.s32 %r35, %r33, 5;
add.s32 %r36, %r35, %r34;
mul.lo.s32 %r37, %r36, 98;
sub.s32 %r38, %r92, %r37;
add.s32 %r39, %r38, %r6;
mad.lo.s32 %r40, %r36, 784, %r39;
mul.wide.s32 %rd9, %r40, 4;
add.s64 %rd10, %rd1, %rd9;
ld.global.nc.v2.u32 {%r41, %r42}, [%rd10];
add.s32 %r11, %r4, %r94;
st.shared.v2.u32 [%r11], {%r41, %r42};
setp.eq.s32 %p2, %r93, 51200;
@%p2 bra $L__BB0_6;
bra.uni $L__BB0_5;
$L__BB0_6:
bar.sync 0;
mov.u32 %r96, 0;
bra.uni $L__BB0_7;
$L__BB0_8:
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(13+i*1), [%r(18+i*-1)+8];
for.size.3 i in range(0, 2, 1):
fma.rn.f32 %f(15+i*3), %f(14+i*3), %f(13+i*3), %f(34+i*-19);
ld.shared.f32 %f(16+i*3), [%r18+(12+i*4)];
ld.shared.f32 %f(17+i*3), [%r17+(12+i*4)];
for.size.3 i in range(0, 3, 1):
fma.rn.f32 %f(21+i*3), %f(20+i*3), %f(19+i*3), %f(18+i*3);
ld.shared.f32 %f(22+i*3), [%r18+(20+i*4)];
ld.shared.f32 %f(23+i*3), [%r17+(20+i*4)];
fma.rn.f32 %f34, %f29, %f28, %f27;
add.s32 %r96, %r96, 4;
$L__BB0_7:
shl.b32 %r84, %r96, 1;
add.s32 %r85, %r84, %r2;
shl.b32 %r86, %r96, 3;
mov.u32 %r87, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r17, %r87, %r86;
shl.b32 %r88, %r85, 2;
add.s32 %r18, %r24, %r88;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(8+i*1), [%r(18+i*-1)+0];
fma.rn.f32 %f10, %f9, %f8, %f34;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(11+i*1), [%r(18+i*-1)+4];
fma.rn.f32 %f34, %f12, %f11, %f10;
setp.eq.s32 %p3, %r96, 48;
@%p3 bra $L__BB0_9;
bra.uni $L__BB0_8;
$L__BB0_9:
add.s32 %r91, %r91, 1;
setp.lt.u32 %p4, %r91, 8;
@%p4 bra $L__BB0_1;
cvta.to.global.u64 %rd17, %rd4;
mul.wide.s32 %rd18, %r22, 4;
add.s64 %rd19, %rd17, %rd18;
ld.global.nc.f32 %f30, [%rd19];
add.f32 %f31, %f34, %f30;
mov.f32 %f32, 0f00000000;
max.f32 %f33, %f31, %f32;
cvta.to.global.u64 %rd20, %rd3;
add.s64 %rd21, %rd20, %rd18;
st.global.f32 [%rd21], %f33;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[98];
__shared__ float placeholder_shared[12544];
T_dense[0] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 8; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_inner_s = 0; ax0_ax1_fused_inner_s < 2; ++ax0_ax1_fused_inner_s) {
if ((int)threadIdx.x * 2 + ax0_ax1_fused_inner_s < 98) {
if ((int)threadIdx.x < 49) {
int index0 = (int)threadIdx.x * 2 + ax0_ax1_fused_inner_s;
int index1 = k_outer_outer * 98 + (int)threadIdx.x * 2 + ax0_ax1_fused_inner_s;
placeholder_d_shared[index0] = placeholder[index1];
}
}
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 49; ++ax0_ax1_fused_outer_outer) {
((float2*)(placeholder_shared + ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder1 + (ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 2) / 98 * 784 + k_outer_outer * 98 + (ax0_ax1_fused_outer_outer * 256 + (int)threadIdx.x * 2) % 98))[0];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 49; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 2; ++k_inner) {
int index2 = k_outer_inner * 2 + k_inner;
int index3 = (int)threadIdx.x * 98 + k_outer_inner * 2 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
}
}
}
int index4 = (int)threadIdx.x;
T_relu[index4] = max(T_dense[0] + placeholder2[index4], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpv9u8xkm5 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<10>;
.reg .f32 %f<48>;
.reg .b32 %r<16>;
.reg .b64 %rd<27>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[4];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[128];
ld.param.u64 %rd12, [default_function_kernel0_param_0];
ld.param.u64 %rd13, [default_function_kernel0_param_1];
ld.param.u64 %rd14, [default_function_kernel0_param_2];
ld.param.u64 %rd11, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd15, %rd13;
cvta.to.global.u64 %rd1, %rd14;
mov.u32 %r1, %tid.x;
shl.b32 %r8, %r1, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r9, %r8;
mov.u32 %r10, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r3, %r10, %r8;
mov.u32 %r4, %ctaid.x;
mul.lo.s32 %r11, %r1, 784;
mad.lo.s32 %r12, %r4, 25088, %r11;
mul.wide.s32 %rd16, %r12, 4;
add.s64 %rd17, %rd15, %rd16;
add.s64 %rd26, %rd17, 16;
cvta.to.global.u64 %rd18, %rd12;
mul.wide.s32 %rd19, %r11, 4;
add.s64 %rd20, %rd18, %rd19;
add.s64 %rd25, %rd20, 16;
mov.f32 %f47, 0f00000000;
mov.u32 %r15, 0;
$L__BB0_1:
bar.sync 0;
add.s64 %rd6, %rd25, -16;
setp.gt.s32 %p1, %r1, 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f11, [%rd6];
st.shared.f32 [%r2], %f11;
$L__BB0_3:
add.s64 %rd7, %rd26, -16;
ld.global.nc.f32 %f12, [%rd26+-16];
st.shared.f32 [%r3], %f12;
bar.sync 0;
ld.shared.f32 %f13, [%r3];
ld.shared.f32 %f14, [_ZZ24default_function_kernel0E20placeholder_d_shared];
fma.rn.f32 %f2, %f14, %f13, %f47;
bar.sync 0;
@%p1 bra $L__BB0_5;
ld.global.nc.f32 %f15, [%rd6+4];
st.shared.f32 [%r2], %f15;
$L__BB0_5:
ld.global.nc.f32 %f16, [%rd7+4];
st.shared.f32 [%r3], %f16;
bar.sync 0;
ld.shared.f32 %f17, [%r3];
ld.shared.f32 %f18, [_ZZ24default_function_kernel0E20placeholder_d_shared];
fma.rn.f32 %f3, %f18, %f17, %f2;
bar.sync 0;
@%p1 bra $L__BB0_7;
ld.global.nc.f32 %f19, [%rd6+8];
st.shared.f32 [%r2], %f19;
$L__BB0_7:
ld.global.nc.f32 %f20, [%rd7+8];
st.shared.f32 [%r3], %f20;
bar.sync 0;
ld.shared.f32 %f21, [%r3];
ld.shared.f32 %f22, [_ZZ24default_function_kernel0E20placeholder_d_shared];
fma.rn.f32 %f4, %f22, %f21, %f3;
bar.sync 0;
@%p1 bra $L__BB0_9;
ld.global.nc.f32 %f23, [%rd6+12];
st.shared.f32 [%r2], %f23;
$L__BB0_9:
ld.global.nc.f32 %f24, [%rd7+12];
st.shared.f32 [%r3], %f24;
bar.sync 0;
ld.shared.f32 %f25, [%r3];
ld.shared.f32 %f26, [_ZZ24default_function_kernel0E20placeholder_d_shared];
fma.rn.f32 %f5, %f26, %f25, %f4;
bar.sync 0;
@%p1 bra $L__BB0_11;
ld.global.nc.f32 %f27, [%rd6+16];
st.shared.f32 [%r2], %f27;
$L__BB0_11:
ld.global.nc.f32 %f28, [%rd7+16];
st.shared.f32 [%r3], %f28;
bar.sync 0;
ld.shared.f32 %f29, [%r3];
ld.shared.f32 %f30, [_ZZ24default_function_kernel0E20placeholder_d_shared];
fma.rn.f32 %f6, %f30, %f29, %f5;
bar.sync 0;
@%p1 bra $L__BB0_13;
ld.global.nc.f32 %f31, [%rd6+20];
st.shared.f32 [%r2], %f31;
$L__BB0_13:
ld.global.nc.f32 %f32, [%rd7+20];
st.shared.f32 [%r3], %f32;
bar.sync 0;
ld.shared.f32 %f33, [%r3];
ld.shared.f32 %f34, [_ZZ24default_function_kernel0E20placeholder_d_shared];
fma.rn.f32 %f7, %f34, %f33, %f6;
bar.sync 0;
@%p1 bra $L__BB0_15;
ld.global.nc.f32 %f35, [%rd6+24];
st.shared.f32 [%r2], %f35;
$L__BB0_15:
ld.global.nc.f32 %f36, [%rd7+24];
st.shared.f32 [%r3], %f36;
bar.sync 0;
ld.shared.f32 %f37, [%r3];
ld.shared.f32 %f38, [_ZZ24default_function_kernel0E20placeholder_d_shared];
fma.rn.f32 %f8, %f38, %f37, %f7;
bar.sync 0;
@%p1 bra $L__BB0_17;
ld.global.nc.f32 %f39, [%rd6+28];
st.shared.f32 [%r2], %f39;
$L__BB0_17:
ld.global.nc.f32 %f40, [%rd7+28];
st.shared.f32 [%r3], %f40;
bar.sync 0;
ld.shared.f32 %f41, [%r3];
ld.shared.f32 %f42, [_ZZ24default_function_kernel0E20placeholder_d_shared];
fma.rn.f32 %f47, %f42, %f41, %f8;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(26+i*-1), %rd(26+i*-1), 32;
add.s32 %r15, %r15, 8;
setp.ne.s32 %p9, %r15, 784;
@%p9 bra $L__BB0_1;
cvta.to.global.u64 %rd21, %rd11;
shl.b32 %r13, %r4, 5;
add.s32 %r14, %r13, %r1;
mul.wide.s32 %rd22, %r14, 4;
add.s64 %rd23, %rd21, %rd22;
ld.global.nc.f32 %f43, [%rd23];
add.f32 %f44, %f47, %f43;
mov.f32 %f45, 0f00000000;
max.f32 %f46, %f44, %f45;
add.s64 %rd24, %rd1, %rd22;
st.global.f32 [%rd24], %f46;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[1];
__shared__ float placeholder_shared[32];
T_dense[0] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 784; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 1) {
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 784 + k_outer_outer;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index2 = (int)blockIdx.x * 25088 + (int)threadIdx.x * 784 + k_outer_outer;
placeholder_shared[index0] = placeholder1[index2];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0];
}
int index3 = (int)blockIdx.x * 32 + (int)threadIdx.x;
T_relu[index3] = max(T_dense[0] + placeholder2[index3], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp1o25zkr1 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<235>;
.reg .b32 %r<22>;
.reg .b64 %rd<16>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[64];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[8192];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd7, [default_function_kernel0_param_2];
ld.param.u64 %rd4, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd6;
cvta.to.global.u64 %rd2, %rd5;
cvta.to.global.u64 %rd3, %rd7;
mov.u32 %r1, %tid.x;
shl.b32 %r9, %r1, 2;
mov.u32 %r10, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r10, %r9;
shr.s32 %r11, %r1, 4;
mul.lo.s32 %r12, %r11, 784;
and.b32 %r13, %r1, 15;
or.b32 %r3, %r12, %r13;
mov.u32 %r14, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r4, %r14, %r9;
shl.b32 %r15, %r1, 8;
add.s32 %r5, %r14, %r15;
mov.u32 %r21, 0;
mov.f32 %f231, 0f00000000;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(232+i*1), %f231;
$L__BB0_1:
bar.sync 0;
setp.lt.s32 %p1, %r1, 16;
@%p1 bra $L__BB0_2;
bra.uni $L__BB0_3;
$L__BB0_2:
shl.b32 %r16, %r21, 4;
add.s32 %r17, %r16, %r1;
mul.wide.s32 %rd8, %r17, 4;
add.s64 %rd9, %rd2, %rd8;
ld.global.nc.f32 %f13, [%rd9];
st.shared.f32 [%r2], %f13;
$L__BB0_3:
shl.b32 %r18, %r21, 4;
add.s32 %r19, %r3, %r18;
mul.wide.s32 %rd10, %r19, 4;
add.s64 %rd11, %rd1, %rd10;
for.size.2 i in range(0, 64, 1):
ld.global.nc.f32 %f(14+i*1), [%rd11+(0+i*6272)];
st.shared.f32 [%r4+(0+i*128)], %f(14+i*1);
bar.sync 0;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(78+i*3), [%r5+(0+i*4)];
ld.shared.f32 %f(79+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(80+i*3), %f(79+i*3), %f(78+i*3), %f(234+i*-154);
for.size.3 i in range(0, 13, 1):
ld.shared.f32 %f(84+i*3), [%r5+(8+i*4)];
ld.shared.f32 %f(85+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(86+i*3), %f(85+i*3), %f(84+i*3), %f(83+i*3);
ld.shared.f32 %f123, [%r5+60];
ld.shared.f32 %f124, [_ZZ24default_function_kernel0E20placeholder_d_shared+60];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(234+i*-108), %f(124+i*-45), %f(123+i*2), %f(122+i*111);
ld.shared.f32 %f(125+i*2), [%r5+(64+i*4)];
for.size.2 i in range(0, 14, 1):
fma.rn.f32 %f(128+i*2), %f(82+i*3), %f(127+i*2), %f(126+i*2);
ld.shared.f32 %f(129+i*2), [%r5+(72+i*4)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(233+i*-76), %f(124+i*-45), %f(155+i*1), %f(154+i*78);
ld.shared.f32 %f(156+i*2), [%r5+(128+i*4)];
for.size.2 i in range(0, 14, 1):
fma.rn.f32 %f(159+i*2), %f(82+i*3), %f(158+i*2), %f(157+i*2);
ld.shared.f32 %f(160+i*2), [%r5+(136+i*4)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(232+i*-44), %f(124+i*-45), %f(186+i*1), %f(185+i*46);
ld.shared.f32 %f(187+i*2), [%r5+(192+i*4)];
for.size.2 i in range(0, 14, 1):
fma.rn.f32 %f(190+i*2), %f(82+i*3), %f(189+i*2), %f(188+i*2);
ld.shared.f32 %f(191+i*2), [%r5+(200+i*4)];
fma.rn.f32 %f231, %f124, %f217, %f216;
add.s32 %r21, %r21, 1;
setp.ne.s32 %p2, %r21, 49;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd12, %rd4;
mul.wide.s32 %rd13, %r9, 4;
add.s64 %rd14, %rd12, %rd13;
ld.global.nc.f32 %f218, [%rd14];
add.f32 %f219, %f234, %f218;
mov.f32 %f220, 0f00000000;
max.f32 %f221, %f219, %f220;
add.s64 %rd15, %rd3, %rd13;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd15+(0+i*4)], %f(221+i*3);
ld.global.nc.f32 %f(222+i*3), [%rd14+(4+i*4)];
add.f32 %f(223+i*3), %f(233+i*-1), %f(222+i*3);
max.f32 %f(224+i*3), %f(223+i*3), %f220;
st.global.f32 [%rd15+12], %f230;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[16];
__shared__ float placeholder_shared[2048];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 49; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 16 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index2 = ((int)threadIdx.x >> 4) * 784 + k_outer_outer * 16 + ((int)threadIdx.x & 15);
placeholder_shared[index0] = placeholder1[index2];
placeholder_shared[index0 + 32] = placeholder1[index2 + 1568];
placeholder_shared[index0 + 64] = placeholder1[index2 + 3136];
placeholder_shared[index0 + 96] = placeholder1[index2 + 4704];
placeholder_shared[index0 + 128] = placeholder1[index2 + 6272];
placeholder_shared[index0 + 160] = placeholder1[index2 + 7840];
placeholder_shared[index0 + 192] = placeholder1[index2 + 9408];
placeholder_shared[index0 + 224] = placeholder1[index2 + 10976];
placeholder_shared[index0 + 256] = placeholder1[index2 + 12544];
placeholder_shared[index0 + 288] = placeholder1[index2 + 14112];
placeholder_shared[index0 + 320] = placeholder1[index2 + 15680];
placeholder_shared[index0 + 352] = placeholder1[index2 + 17248];
placeholder_shared[index0 + 384] = placeholder1[index2 + 18816];
placeholder_shared[index0 + 416] = placeholder1[index2 + 20384];
placeholder_shared[index0 + 448] = placeholder1[index2 + 21952];
placeholder_shared[index0 + 480] = placeholder1[index2 + 23520];
placeholder_shared[index0 + 512] = placeholder1[index2 + 25088];
placeholder_shared[index0 + 544] = placeholder1[index2 + 26656];
placeholder_shared[index0 + 576] = placeholder1[index2 + 28224];
placeholder_shared[index0 + 608] = placeholder1[index2 + 29792];
placeholder_shared[index0 + 640] = placeholder1[index2 + 31360];
placeholder_shared[index0 + 672] = placeholder1[index2 + 32928];
placeholder_shared[index0 + 704] = placeholder1[index2 + 34496];
placeholder_shared[index0 + 736] = placeholder1[index2 + 36064];
placeholder_shared[index0 + 768] = placeholder1[index2 + 37632];
placeholder_shared[index0 + 800] = placeholder1[index2 + 39200];
placeholder_shared[index0 + 832] = placeholder1[index2 + 40768];
placeholder_shared[index0 + 864] = placeholder1[index2 + 42336];
placeholder_shared[index0 + 896] = placeholder1[index2 + 43904];
placeholder_shared[index0 + 928] = placeholder1[index2 + 45472];
placeholder_shared[index0 + 960] = placeholder1[index2 + 47040];
placeholder_shared[index0 + 992] = placeholder1[index2 + 48608];
placeholder_shared[index0 + 1024] = placeholder1[index2 + 50176];
placeholder_shared[index0 + 1056] = placeholder1[index2 + 51744];
placeholder_shared[index0 + 1088] = placeholder1[index2 + 53312];
placeholder_shared[index0 + 1120] = placeholder1[index2 + 54880];
placeholder_shared[index0 + 1152] = placeholder1[index2 + 56448];
placeholder_shared[index0 + 1184] = placeholder1[index2 + 58016];
placeholder_shared[index0 + 1216] = placeholder1[index2 + 59584];
placeholder_shared[index0 + 1248] = placeholder1[index2 + 61152];
placeholder_shared[index0 + 1280] = placeholder1[index2 + 62720];
placeholder_shared[index0 + 1312] = placeholder1[index2 + 64288];
placeholder_shared[index0 + 1344] = placeholder1[index2 + 65856];
placeholder_shared[index0 + 1376] = placeholder1[index2 + 67424];
placeholder_shared[index0 + 1408] = placeholder1[index2 + 68992];
placeholder_shared[index0 + 1440] = placeholder1[index2 + 70560];
placeholder_shared[index0 + 1472] = placeholder1[index2 + 72128];
placeholder_shared[index0 + 1504] = placeholder1[index2 + 73696];
placeholder_shared[index0 + 1536] = placeholder1[index2 + 75264];
placeholder_shared[index0 + 1568] = placeholder1[index2 + 76832];
placeholder_shared[index0 + 1600] = placeholder1[index2 + 78400];
placeholder_shared[index0 + 1632] = placeholder1[index2 + 79968];
placeholder_shared[index0 + 1664] = placeholder1[index2 + 81536];
placeholder_shared[index0 + 1696] = placeholder1[index2 + 83104];
placeholder_shared[index0 + 1728] = placeholder1[index2 + 84672];
placeholder_shared[index0 + 1760] = placeholder1[index2 + 86240];
placeholder_shared[index0 + 1792] = placeholder1[index2 + 87808];
placeholder_shared[index0 + 1824] = placeholder1[index2 + 89376];
placeholder_shared[index0 + 1856] = placeholder1[index2 + 90944];
placeholder_shared[index0 + 1888] = placeholder1[index2 + 92512];
placeholder_shared[index0 + 1920] = placeholder1[index2 + 94080];
placeholder_shared[index0 + 1952] = placeholder1[index2 + 95648];
placeholder_shared[index0 + 1984] = placeholder1[index2 + 97216];
placeholder_shared[index0 + 2016] = placeholder1[index2 + 98784];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 64];
int index3 = (int)threadIdx.x * 64 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index3];
int index4 = (int)threadIdx.x * 64 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index4];
int index5 = (int)threadIdx.x * 64 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index5];
int index6 = (int)threadIdx.x * 64 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index6];
int index7 = (int)threadIdx.x * 64 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index7];
int index8 = (int)threadIdx.x * 64 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index8];
int index9 = (int)threadIdx.x * 64 + 7;
T_dense[0] = T_dense[0] + placeholder_d_shared[7] * placeholder_shared[index9];
int index10 = (int)threadIdx.x * 64 + 8;
T_dense[0] = T_dense[0] + placeholder_d_shared[8] * placeholder_shared[index10];
int index11 = (int)threadIdx.x * 64 + 9;
T_dense[0] = T_dense[0] + placeholder_d_shared[9] * placeholder_shared[index11];
int index12 = (int)threadIdx.x * 64 + 10;
T_dense[0] = T_dense[0] + placeholder_d_shared[10] * placeholder_shared[index12];
int index13 = (int)threadIdx.x * 64 + 11;
T_dense[0] = T_dense[0] + placeholder_d_shared[11] * placeholder_shared[index13];
int index14 = (int)threadIdx.x * 64 + 12;
T_dense[0] = T_dense[0] + placeholder_d_shared[12] * placeholder_shared[index14];
int index15 = (int)threadIdx.x * 64 + 13;
T_dense[0] = T_dense[0] + placeholder_d_shared[13] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 64 + 14;
T_dense[0] = T_dense[0] + placeholder_d_shared[14] * placeholder_shared[index16];
int index17 = (int)threadIdx.x * 64 + 15;
T_dense[0] = T_dense[0] + placeholder_d_shared[15] * placeholder_shared[index17];
int index18 = (int)threadIdx.x * 64 + 16;
T_dense[1] = T_dense[1] + placeholder_d_shared[0] * placeholder_shared[index18];
int index19 = (int)threadIdx.x * 64 + 17;
T_dense[1] = T_dense[1] + placeholder_d_shared[1] * placeholder_shared[index19];
int index20 = (int)threadIdx.x * 64 + 18;
T_dense[1] = T_dense[1] + placeholder_d_shared[2] * placeholder_shared[index20];
int index21 = (int)threadIdx.x * 64 + 19;
T_dense[1] = T_dense[1] + placeholder_d_shared[3] * placeholder_shared[index21];
int index22 = (int)threadIdx.x * 64 + 20;
T_dense[1] = T_dense[1] + placeholder_d_shared[4] * placeholder_shared[index22];
int index23 = (int)threadIdx.x * 64 + 21;
T_dense[1] = T_dense[1] + placeholder_d_shared[5] * placeholder_shared[index23];
int index24 = (int)threadIdx.x * 64 + 22;
T_dense[1] = T_dense[1] + placeholder_d_shared[6] * placeholder_shared[index24];
int index25 = (int)threadIdx.x * 64 + 23;
T_dense[1] = T_dense[1] + placeholder_d_shared[7] * placeholder_shared[index25];
int index26 = (int)threadIdx.x * 64 + 24;
T_dense[1] = T_dense[1] + placeholder_d_shared[8] * placeholder_shared[index26];
int index27 = (int)threadIdx.x * 64 + 25;
T_dense[1] = T_dense[1] + placeholder_d_shared[9] * placeholder_shared[index27];
int index28 = (int)threadIdx.x * 64 + 26;
T_dense[1] = T_dense[1] + placeholder_d_shared[10] * placeholder_shared[index28];
int index29 = (int)threadIdx.x * 64 + 27;
T_dense[1] = T_dense[1] + placeholder_d_shared[11] * placeholder_shared[index29];
int index30 = (int)threadIdx.x * 64 + 28;
T_dense[1] = T_dense[1] + placeholder_d_shared[12] * placeholder_shared[index30];
int index31 = (int)threadIdx.x * 64 + 29;
T_dense[1] = T_dense[1] + placeholder_d_shared[13] * placeholder_shared[index31];
int index32 = (int)threadIdx.x * 64 + 30;
T_dense[1] = T_dense[1] + placeholder_d_shared[14] * placeholder_shared[index32];
int index33 = (int)threadIdx.x * 64 + 31;
T_dense[1] = T_dense[1] + placeholder_d_shared[15] * placeholder_shared[index33];
int index34 = (int)threadIdx.x * 64 + 32;
T_dense[2] = T_dense[2] + placeholder_d_shared[0] * placeholder_shared[index34];
int index35 = (int)threadIdx.x * 64 + 33;
T_dense[2] = T_dense[2] + placeholder_d_shared[1] * placeholder_shared[index35];
int index36 = (int)threadIdx.x * 64 + 34;
T_dense[2] = T_dense[2] + placeholder_d_shared[2] * placeholder_shared[index36];
int index37 = (int)threadIdx.x * 64 + 35;
T_dense[2] = T_dense[2] + placeholder_d_shared[3] * placeholder_shared[index37];
int index38 = (int)threadIdx.x * 64 + 36;
T_dense[2] = T_dense[2] + placeholder_d_shared[4] * placeholder_shared[index38];
int index39 = (int)threadIdx.x * 64 + 37;
T_dense[2] = T_dense[2] + placeholder_d_shared[5] * placeholder_shared[index39];
int index40 = (int)threadIdx.x * 64 + 38;
T_dense[2] = T_dense[2] + placeholder_d_shared[6] * placeholder_shared[index40];
int index41 = (int)threadIdx.x * 64 + 39;
T_dense[2] = T_dense[2] + placeholder_d_shared[7] * placeholder_shared[index41];
int index42 = (int)threadIdx.x * 64 + 40;
T_dense[2] = T_dense[2] + placeholder_d_shared[8] * placeholder_shared[index42];
int index43 = (int)threadIdx.x * 64 + 41;
T_dense[2] = T_dense[2] + placeholder_d_shared[9] * placeholder_shared[index43];
int index44 = (int)threadIdx.x * 64 + 42;
T_dense[2] = T_dense[2] + placeholder_d_shared[10] * placeholder_shared[index44];
int index45 = (int)threadIdx.x * 64 + 43;
T_dense[2] = T_dense[2] + placeholder_d_shared[11] * placeholder_shared[index45];
int index46 = (int)threadIdx.x * 64 + 44;
T_dense[2] = T_dense[2] + placeholder_d_shared[12] * placeholder_shared[index46];
int index47 = (int)threadIdx.x * 64 + 45;
T_dense[2] = T_dense[2] + placeholder_d_shared[13] * placeholder_shared[index47];
int index48 = (int)threadIdx.x * 64 + 46;
T_dense[2] = T_dense[2] + placeholder_d_shared[14] * placeholder_shared[index48];
int index49 = (int)threadIdx.x * 64 + 47;
T_dense[2] = T_dense[2] + placeholder_d_shared[15] * placeholder_shared[index49];
int index50 = (int)threadIdx.x * 64 + 48;
T_dense[3] = T_dense[3] + placeholder_d_shared[0] * placeholder_shared[index50];
int index51 = (int)threadIdx.x * 64 + 49;
T_dense[3] = T_dense[3] + placeholder_d_shared[1] * placeholder_shared[index51];
int index52 = (int)threadIdx.x * 64 + 50;
T_dense[3] = T_dense[3] + placeholder_d_shared[2] * placeholder_shared[index52];
int index53 = (int)threadIdx.x * 64 + 51;
T_dense[3] = T_dense[3] + placeholder_d_shared[3] * placeholder_shared[index53];
int index54 = (int)threadIdx.x * 64 + 52;
T_dense[3] = T_dense[3] + placeholder_d_shared[4] * placeholder_shared[index54];
int index55 = (int)threadIdx.x * 64 + 53;
T_dense[3] = T_dense[3] + placeholder_d_shared[5] * placeholder_shared[index55];
int index56 = (int)threadIdx.x * 64 + 54;
T_dense[3] = T_dense[3] + placeholder_d_shared[6] * placeholder_shared[index56];
int index57 = (int)threadIdx.x * 64 + 55;
T_dense[3] = T_dense[3] + placeholder_d_shared[7] * placeholder_shared[index57];
int index58 = (int)threadIdx.x * 64 + 56;
T_dense[3] = T_dense[3] + placeholder_d_shared[8] * placeholder_shared[index58];
int index59 = (int)threadIdx.x * 64 + 57;
T_dense[3] = T_dense[3] + placeholder_d_shared[9] * placeholder_shared[index59];
int index60 = (int)threadIdx.x * 64 + 58;
T_dense[3] = T_dense[3] + placeholder_d_shared[10] * placeholder_shared[index60];
int index61 = (int)threadIdx.x * 64 + 59;
T_dense[3] = T_dense[3] + placeholder_d_shared[11] * placeholder_shared[index61];
int index62 = (int)threadIdx.x * 64 + 60;
T_dense[3] = T_dense[3] + placeholder_d_shared[12] * placeholder_shared[index62];
int index63 = (int)threadIdx.x * 64 + 61;
T_dense[3] = T_dense[3] + placeholder_d_shared[13] * placeholder_shared[index63];
int index64 = (int)threadIdx.x * 64 + 62;
T_dense[3] = T_dense[3] + placeholder_d_shared[14] * placeholder_shared[index64];
int index65 = (int)threadIdx.x * 64 + 63;
T_dense[3] = T_dense[3] + placeholder_d_shared[15] * placeholder_shared[index65];
}
for (int ax1_inner = 0; ax1_inner < 4; ++ax1_inner) {
int index66 = (int)threadIdx.x * 4 + ax1_inner;
int index67 = ax1_inner;
T_relu[index66] = max(T_dense[index67] + placeholder2[index66], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpfl_mw0l0 | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<3>;
.reg .f32 %f<73>;
.reg .b32 %r<18>;
.reg .b64 %rd<24>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[64];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[8192];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd11, [default_function_kernel0_param_1];
ld.param.u64 %rd12, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd13, %rd11;
cvta.to.global.u64 %rd1, %rd12;
mov.u32 %r7, %tid.x;
shl.b32 %r8, %r7, 2;
mov.u32 %r9, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r1, %r9, %r8;
shr.s32 %r10, %r7, 4;
and.b32 %r11, %r7, 15;
mov.u32 %r12, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r2, %r12, %r8;
shl.b32 %r13, %r7, 6;
add.s32 %r3, %r12, %r13;
cvt.s64.s32 %rd2, %r7;
cvta.to.global.u64 %rd14, %rd10;
mul.wide.s32 %rd15, %r7, 4;
add.s64 %rd23, %rd14, %rd15;
mul.lo.s32 %r14, %r10, 784;
or.b32 %r15, %r14, %r11;
mul.wide.s32 %rd16, %r15, 4;
add.s64 %rd17, %rd13, %rd16;
add.s64 %rd22, %rd17, 200704;
mov.f32 %f72, 0f00000000;
mov.u32 %r17, 0;
$L__BB0_1:
cvt.u32.u64 %r16, %rd2;
setp.gt.s32 %p1, %r16, 15;
bar.sync 0;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f4, [%rd23];
st.shared.f32 [%r1], %f4;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(5+i*1), [%rd22+(-200704+i*25088)];
st.shared.f32 [%r2+(0+i*512)], %f(5+i*1);
bar.sync 0;
for.size.3 i in range(0, 2, 1):
ld.shared.f32 %f(21+i*3), [%r3+(0+i*4)];
ld.shared.f32 %f(22+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(0+i*4)];
fma.rn.f32 %f(23+i*3), %f(22+i*3), %f(21+i*3), %f(72+i*-49);
for.size.3 i in range(0, 13, 1):
ld.shared.f32 %f(27+i*3), [%r3+(8+i*4)];
ld.shared.f32 %f(28+i*3), [_ZZ24default_function_kernel0E20placeholder_d_shared+(8+i*4)];
fma.rn.f32 %f(29+i*3), %f(28+i*3), %f(27+i*3), %f(26+i*3);
ld.shared.f32 %f66, [%r3+60];
ld.shared.f32 %f67, [_ZZ24default_function_kernel0E20placeholder_d_shared+60];
fma.rn.f32 %f72, %f67, %f66, %f65;
for.size.1 i in range(0, 2, 1):
add.s64 %rd(23+i*-1), %rd(23+i*-1), 64;
add.s32 %r17, %r17, 1;
setp.ne.s32 %p2, %r17, 49;
@%p2 bra $L__BB0_1;
cvta.to.global.u64 %rd18, %rd9;
shl.b64 %rd19, %rd2, 2;
add.s64 %rd20, %rd18, %rd19;
ld.global.nc.f32 %f68, [%rd20];
add.f32 %f69, %f72, %f68;
mov.f32 %f70, 0f00000000;
max.f32 %f71, %f69, %f70;
add.s64 %rd21, %rd1, %rd19;
st.global.f32 [%rd21], %f71;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[1];
__shared__ float placeholder_d_shared[16];
__shared__ float placeholder_shared[2048];
T_dense[0] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 49; ++k_outer_outer) {
__syncthreads();
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 16 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
}
int index0 = (int)threadIdx.x;
int index2 = ((int)threadIdx.x >> 4) * 784 + k_outer_outer * 16 + ((int)threadIdx.x & 15);
placeholder_shared[index0] = placeholder1[index2];
placeholder_shared[index0 + 128] = placeholder1[index2 + 6272];
placeholder_shared[index0 + 256] = placeholder1[index2 + 12544];
placeholder_shared[index0 + 384] = placeholder1[index2 + 18816];
placeholder_shared[index0 + 512] = placeholder1[index2 + 25088];
placeholder_shared[index0 + 640] = placeholder1[index2 + 31360];
placeholder_shared[index0 + 768] = placeholder1[index2 + 37632];
placeholder_shared[index0 + 896] = placeholder1[index2 + 43904];
placeholder_shared[index0 + 1024] = placeholder1[index2 + 50176];
placeholder_shared[index0 + 1152] = placeholder1[index2 + 56448];
placeholder_shared[index0 + 1280] = placeholder1[index2 + 62720];
placeholder_shared[index0 + 1408] = placeholder1[index2 + 68992];
placeholder_shared[index0 + 1536] = placeholder1[index2 + 75264];
placeholder_shared[index0 + 1664] = placeholder1[index2 + 81536];
placeholder_shared[index0 + 1792] = placeholder1[index2 + 87808];
placeholder_shared[index0 + 1920] = placeholder1[index2 + 94080];
__syncthreads();
T_dense[0] = T_dense[0] + placeholder_d_shared[0] * placeholder_shared[index0 * 16];
int index3 = (int)threadIdx.x * 16 + 1;
T_dense[0] = T_dense[0] + placeholder_d_shared[1] * placeholder_shared[index3];
int index4 = (int)threadIdx.x * 16 + 2;
T_dense[0] = T_dense[0] + placeholder_d_shared[2] * placeholder_shared[index4];
int index5 = (int)threadIdx.x * 16 + 3;
T_dense[0] = T_dense[0] + placeholder_d_shared[3] * placeholder_shared[index5];
int index6 = (int)threadIdx.x * 16 + 4;
T_dense[0] = T_dense[0] + placeholder_d_shared[4] * placeholder_shared[index6];
int index7 = (int)threadIdx.x * 16 + 5;
T_dense[0] = T_dense[0] + placeholder_d_shared[5] * placeholder_shared[index7];
int index8 = (int)threadIdx.x * 16 + 6;
T_dense[0] = T_dense[0] + placeholder_d_shared[6] * placeholder_shared[index8];
int index9 = (int)threadIdx.x * 16 + 7;
T_dense[0] = T_dense[0] + placeholder_d_shared[7] * placeholder_shared[index9];
int index10 = (int)threadIdx.x * 16 + 8;
T_dense[0] = T_dense[0] + placeholder_d_shared[8] * placeholder_shared[index10];
int index11 = (int)threadIdx.x * 16 + 9;
T_dense[0] = T_dense[0] + placeholder_d_shared[9] * placeholder_shared[index11];
int index12 = (int)threadIdx.x * 16 + 10;
T_dense[0] = T_dense[0] + placeholder_d_shared[10] * placeholder_shared[index12];
int index13 = (int)threadIdx.x * 16 + 11;
T_dense[0] = T_dense[0] + placeholder_d_shared[11] * placeholder_shared[index13];
int index14 = (int)threadIdx.x * 16 + 12;
T_dense[0] = T_dense[0] + placeholder_d_shared[12] * placeholder_shared[index14];
int index15 = (int)threadIdx.x * 16 + 13;
T_dense[0] = T_dense[0] + placeholder_d_shared[13] * placeholder_shared[index15];
int index16 = (int)threadIdx.x * 16 + 14;
T_dense[0] = T_dense[0] + placeholder_d_shared[14] * placeholder_shared[index16];
int index17 = (int)threadIdx.x * 16 + 15;
T_dense[0] = T_dense[0] + placeholder_d_shared[15] * placeholder_shared[index17];
}
int index0 = (int)threadIdx.x;
T_relu[index0] = max(T_dense[0] + placeholder2[index0], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpzd__rqis | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<316>;
.reg .b32 %r<126>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[448];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[57344];
ld.param.u64 %rd5, [default_function_kernel0_param_0];
ld.param.u64 %rd6, [default_function_kernel0_param_1];
ld.param.u64 %rd7, [default_function_kernel0_param_2];
ld.param.u64 %rd8, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd5;
cvta.to.global.u64 %rd2, %rd7;
cvta.to.global.u64 %rd3, %rd6;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r1, %tid.x;
mul.lo.s32 %r2, %r1, 112;
shl.b32 %r20, %r1, 2;
mov.u32 %r21, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r3, %r21, %r20;
add.s32 %r4, %r1, 56;
mov.u32 %r22, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r22, %r20;
mov.u32 %r120, 0;
mov.f32 %f268, 0f00000000;
for.size.1 i in range(0, 15, 1):
mov.f32 %f(269+i*1), %f268;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r7, %r120, 112;
add.s32 %r24, %r7, %r1;
mul.wide.s32 %rd9, %r24, 4;
add.s64 %rd10, %rd1, %rd9;
for.size.2 i in range(0, 4, 1):
ld.global.nc.f32 %f(97+i*1), [%rd10+(0+i*32)];
st.shared.f32 [%r3+(0+i*32)], %f(97+i*1);
ld.global.nc.f32 %f101, [%rd10+128];
mov.u32 %r123, 128;
for.size.2 i in range(0, 9, 1):
st.shared.f32 [%r3+(128+i*32)], %f(101+i*1);
ld.global.nc.f32 %f(102+i*1), [%rd10+(160+i*32)];
st.shared.f32 [%r3+416], %f110;
mov.u32 %r121, %r1;
mov.u32 %r122, %r4;
$L__BB0_2:
mul.hi.s32 %r25, %r121, -1840700269;
add.s32 %r26, %r25, %r121;
shr.u32 %r27, %r26, 31;
shr.s32 %r28, %r26, 6;
add.s32 %r29, %r28, %r27;
mul.lo.s32 %r30, %r29, 112;
sub.s32 %r31, %r121, %r30;
add.s32 %r32, %r31, %r7;
mad.lo.s32 %r33, %r29, 784, %r32;
mul.wide.s32 %rd11, %r33, 4;
add.s64 %rd12, %rd3, %rd11;
ld.global.nc.f32 %f111, [%rd12];
add.s32 %r34, %r5, %r123;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r34+(-128+i*32)], %f(111+i*1);
add.s32 %r(35+i*10), %r121, (8+i*8);
mul.hi.s32 %r(36+i*10), %r(35+i*10), -1840700269;
add.s32 %r(37+i*10), %r(36+i*10), %r(35+i*10);
shr.u32 %r(38+i*10), %r(37+i*10), 31;
shr.s32 %r(39+i*10), %r(37+i*10), 6;
add.s32 %r(40+i*10), %r(39+i*10), %r(38+i*10);
mul.lo.s32 %r(41+i*10), %r(40+i*10), 112;
sub.s32 %r(42+i*10), %r(35+i*10), %r(41+i*10);
add.s32 %r(43+i*10), %r(42+i*10), %r7;
mad.lo.s32 %r(44+i*10), %r(40+i*10), 784, %r(43+i*10);
mul.wide.s32 %rd(13+i*2), %r(44+i*10), 4;
add.s64 %rd(14+i*2), %rd3, %rd(13+i*2);
ld.global.nc.f32 %f(112+i*1), [%rd(14+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r34+(0+i*32)], %f(115+i*1);
add.s32 %r(75+i*10), %r122, (-16+i*8);
mul.hi.s32 %r(76+i*10), %r(75+i*10), -1840700269;
add.s32 %r(77+i*10), %r(76+i*10), %r(75+i*10);
shr.u32 %r(78+i*10), %r(77+i*10), 31;
shr.s32 %r(79+i*10), %r(77+i*10), 6;
add.s32 %r(80+i*10), %r(79+i*10), %r(78+i*10);
mul.lo.s32 %r(81+i*10), %r(80+i*10), 112;
sub.s32 %r(82+i*10), %r(75+i*10), %r(81+i*10);
add.s32 %r(83+i*10), %r(82+i*10), %r7;
mad.lo.s32 %r(84+i*10), %r(80+i*10), 784, %r(83+i*10);
mul.wide.s32 %rd(21+i*2), %r(84+i*10), 4;
add.s64 %rd(22+i*2), %rd3, %rd(21+i*2);
ld.global.nc.f32 %f(116+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r34+64], %f117;
mul.hi.s32 %r95, %r122, -1840700269;
add.s32 %r96, %r95, %r122;
shr.u32 %r97, %r96, 31;
shr.s32 %r98, %r96, 6;
add.s32 %r99, %r98, %r97;
mul.lo.s32 %r100, %r99, 112;
sub.s32 %r101, %r122, %r100;
add.s32 %r102, %r101, %r7;
mad.lo.s32 %r103, %r99, 784, %r102;
mul.wide.s32 %rd25, %r103, 4;
add.s64 %rd26, %rd3, %rd25;
ld.global.nc.f32 %f118, [%rd26];
st.shared.f32 [%r34+96], %f118;
add.s32 %r123, %r123, 256;
setp.ne.s32 %p1, %r123, 57472;
for.size.1 i in range(0, 2, 1):
add.s32 %r(12+i*109), %r122, (64+i*-56);
mov.u32 %r122, %r12;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r124, 0;
$L__BB0_4:
add.s32 %r105, %r2, %r124;
for.size.3 i in range(0, 2, 1):
shl.b32 %r(106+i*3), %r(105+i*19), 2;
add.s32 %r(108+i*3), %r(22+i*-1), %r(106+i*3);
ld.shared.f32 %f(119+i*1), [%r(108+i*3)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(121+i*2), %f120, %f(119+i*3), %f(283+i*-1);
ld.shared.f32 %f(122+i*2), [%r108+(3584+i*3584)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(125+i*2), %f120, %f(124+i*2), %f(281+i*-1);
ld.shared.f32 %f(126+i*2), [%r108+(10752+i*3584)];
fma.rn.f32 %f151, %f120, %f150, %f268;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(152+i*1), [%r(108+i*3)+4];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(283+i*-1), %f153, %f(152+i*2), %f(121+i*2);
ld.shared.f32 %f(154+i*1), [%r108+(3588+i*3584)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(281+i*-1), %f153, %f(155+i*1), %f(125+i*2);
ld.shared.f32 %f(156+i*1), [%r108+(10756+i*3584)];
fma.rn.f32 %f268, %f153, %f168, %f151;
add.s32 %r124, %r124, 2;
setp.ne.s32 %p2, %r124, 56;
@%p2 bra $L__BB0_4;
mov.u32 %r125, 0;
$L__BB0_6:
shl.b32 %r113, %r125, 2;
for.size.1 i in range(0, 2, 1):
add.s32 %r(115+i*1), %r(21+i*104), %r(113+i*-111);
shl.b32 %r117, %r116, 2;
add.s32 %r119, %r22, %r117;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(169+i*1), [%r(119+i*-4)+224];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(171+i*2), %f170, %f(169+i*3), %f(283+i*-1);
ld.shared.f32 %f(172+i*2), [%r119+(3808+i*3584)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(175+i*2), %f170, %f(174+i*2), %f(281+i*-1);
ld.shared.f32 %f(176+i*2), [%r119+(10976+i*3584)];
fma.rn.f32 %f201, %f170, %f200, %f268;
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(202+i*1), [%r(119+i*-4)+228];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(283+i*-1), %f203, %f(202+i*2), %f(171+i*2);
ld.shared.f32 %f(204+i*1), [%r119+(3812+i*3584)];
for.size.2 i in range(0, 13, 1):
fma.rn.f32 %f(281+i*-1), %f203, %f(205+i*1), %f(175+i*2);
ld.shared.f32 %f(206+i*1), [%r119+(10980+i*3584)];
fma.rn.f32 %f268, %f203, %f218, %f201;
add.s32 %r125, %r125, 2;
setp.ne.s32 %p3, %r125, 56;
@%p3 bra $L__BB0_6;
add.s32 %r120, %r120, 1;
setp.lt.u32 %p4, %r120, 7;
@%p4 bra $L__BB0_1;
mul.wide.s32 %rd27, %r1, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f219, [%rd28];
add.f32 %f220, %f283, %f219;
mov.f32 %f221, 0f00000000;
max.f32 %f222, %f220, %f221;
add.s64 %rd29, %rd2, %rd27;
for.size.4 i in range(0, 15, 1):
st.global.f32 [%rd29+(0+i*32)], %f(222+i*3);
ld.global.nc.f32 %f(223+i*3), [%rd28+(32+i*32)];
add.f32 %f(224+i*3), %f(282+i*-1), %f(223+i*3);
max.f32 %f(225+i*3), %f(224+i*3), %f221;
st.global.f32 [%rd29+480], %f267;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[16];
__shared__ float placeholder_d_shared[112];
__shared__ float placeholder_shared[14336];
for (int vthread_s = 0; vthread_s < 16; ++vthread_s) {
int index0 = vthread_s;
T_dense[index0] = 0.000000e+00f;
}
for (int k_outer_outer = 0; k_outer_outer < 7; ++k_outer_outer) {
__syncthreads();
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 14; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
int index2 = k_outer_outer * 112 + ax0_ax1_fused_outer_outer * 8 + (int)threadIdx.x;
placeholder_d_shared[index1] = placeholder[index2];
}
for (int ax0_ax1_fused_outer_outer1 = 0; ax0_ax1_fused_outer_outer1 < 1792; ++ax0_ax1_fused_outer_outer1) {
int index3 = ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x;
int index4 = (ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x) / 112 * 784 + k_outer_outer * 112 + (ax0_ax1_fused_outer_outer1 * 8 + (int)threadIdx.x) % 112;
placeholder_shared[index3] = placeholder1[index4];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 56; ++k_inner) {
for (int vthread_s1 = 0; vthread_s1 < 16; ++vthread_s1) {
int index5 = vthread_s1;
int index6 = k_outer_inner * 56 + k_inner;
int index7 = vthread_s1 * 896 + (int)threadIdx.x * 112 + k_outer_inner * 56 + k_inner;
T_dense[index5] = T_dense[index5] + placeholder_d_shared[index6] * placeholder_shared[index7];
}
}
}
}
for (int vthread_s2 = 0; vthread_s2 < 16; ++vthread_s2) {
int index8 = vthread_s2 * 8 + (int)threadIdx.x;
int index9 = vthread_s2;
T_relu[index8] = max(T_dense[index9] + placeholder2[index8], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpejcyy8yo | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<110>;
.reg .b32 %r<52>;
.reg .b64 %rd<16>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[200704];
ld.param.u64 %rd2, [default_function_kernel0_param_0];
ld.param.u64 %rd3, [default_function_kernel0_param_1];
ld.param.u64 %rd4, [default_function_kernel0_param_2];
ld.param.u64 %rd5, [default_function_kernel0_param_3];
mov.u32 %r1, %tid.x;
setp.gt.s32 %p1, %r1, 13;
@%p1 bra $L__BB0_2;
mul.lo.s32 %r16, %r1, 56;
cvta.to.global.u64 %rd6, %rd2;
mul.wide.s32 %rd7, %r16, 4;
add.s64 %rd8, %rd6, %rd7;
ld.global.nc.f32 %f9, [%rd8];
shl.b32 %r17, %r16, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r19, %r18, %r17;
for.size.2 i in range(0, 55, 1):
st.shared.f32 [%r19+(0+i*4)], %f(9+i*1);
ld.global.nc.f32 %f(10+i*1), [%rd8+(4+i*4)];
st.shared.f32 [%r19+220], %f64;
$L__BB0_2:
mov.u32 %r21, %ctaid.x;
mad.lo.s32 %r2, %r21, 50176, %r1;
cvta.to.global.u64 %rd1, %rd3;
mov.u32 %r48, 0;
mov.u32 %r26, _ZZ24default_function_kernel0E18placeholder_shared;
$L__BB0_3:
shl.b32 %r22, %r48, 5;
add.s32 %r23, %r2, %r22;
mul.wide.s32 %rd9, %r23, 4;
add.s64 %rd10, %rd1, %rd9;
ld.global.nc.f32 %f65, [%rd10];
add.s32 %r24, %r22, %r1;
shl.b32 %r25, %r24, 2;
add.s32 %r27, %r26, %r25;
for.size.2 i in range(0, 15, 1):
st.shared.f32 [%r27+(0+i*128)], %f(65+i*1);
ld.global.nc.f32 %f(66+i*1), [%rd10+(128+i*128)];
st.shared.f32 [%r27+1920], %f80;
add.s32 %r48, %r48, 16;
setp.ne.s32 %p2, %r48, 1568;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r5, %r1, 6272, %r26;
mov.u32 %r49, 0;
mov.f32 %f108, 0f00000000;
mov.f32 %f109, %f108;
$L__BB0_5:
mul.lo.s32 %r31, %r49, 196;
add.s32 %r7, %r5, %r31;
mov.u32 %r32, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r33, %r32, %r31;
add.s32 %r50, %r33, 8;
mov.u32 %r51, 3136;
bra.uni $L__BB0_6;
$L__BB0_7:
for.size.2 i in range(0, 2, 1):
add.s32 %r(39+i*1), %r(34+i*16), (-3136+i*3128);
ld.shared.f32 %f(86+i*1), [%r(39+i*1)+4];
fma.rn.f32 %f88, %f87, %f86, %f109;
add.s32 %r41, %r34, -3136;
ld.shared.f32 %f89, [%r41+3140];
fma.rn.f32 %f90, %f87, %f89, %f108;
for.size.2 i in range(0, 2, 1):
add.s32 %r(42+i*1), %r(34+i*16), (-3136+i*3128);
ld.shared.f32 %f(91+i*1), [%r(42+i*1)+8];
fma.rn.f32 %f93, %f92, %f91, %f88;
add.s32 %r44, %r34, -3136;
ld.shared.f32 %f94, [%r44+3144];
fma.rn.f32 %f95, %f92, %f94, %f90;
for.size.2 i in range(0, 2, 1):
add.s32 %r(45+i*1), %r(34+i*16), (-3136+i*3128);
ld.shared.f32 %f(96+i*1), [%r(45+i*1)+12];
fma.rn.f32 %f109, %f97, %f96, %f93;
add.s32 %r47, %r34, -3136;
ld.shared.f32 %f98, [%r47+3148];
fma.rn.f32 %f108, %f97, %f98, %f95;
for.size.1 i in range(0, 2, 1):
add.s32 %r(51+i*-1), %r(51+i*-1), 16;
$L__BB0_6:
add.s32 %r34, %r7, %r51;
ld.shared.f32 %f83, [%r34+-3136];
ld.shared.f32 %f84, [%r50+-8];
fma.rn.f32 %f109, %f84, %f83, %f109;
ld.shared.f32 %f85, [%r34];
fma.rn.f32 %f108, %f84, %f85, %f108;
setp.eq.s32 %p3, %r51, 3328;
@%p3 bra $L__BB0_8;
bra.uni $L__BB0_7;
$L__BB0_8:
add.s32 %r49, %r49, 1;
setp.lt.u32 %p4, %r49, 16;
@%p4 bra $L__BB0_5;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(36+i*1), %r(21+i*-20), (6+i*-5);
add.s32 %r38, %r36, %r37;
cvta.to.global.u64 %rd11, %rd5;
mul.wide.s32 %rd12, %r38, 4;
add.s64 %rd13, %rd11, %rd12;
ld.global.nc.f32 %f99, [%rd13];
add.f32 %f100, %f109, %f99;
mov.f32 %f101, 0f00000000;
max.f32 %f102, %f100, %f101;
cvta.to.global.u64 %rd14, %rd4;
add.s64 %rd15, %rd14, %rd12;
st.global.f32 [%rd15], %f102;
ld.global.nc.f32 %f103, [%rd13+4];
add.f32 %f104, %f108, %f103;
max.f32 %f105, %f104, %f101;
st.global.f32 [%rd15+4], %f105;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[2];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[50176];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0] = placeholder[index0];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 1] = placeholder[index0 + 1];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 2] = placeholder[index0 + 2];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 3] = placeholder[index0 + 3];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 4] = placeholder[index0 + 4];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 5] = placeholder[index0 + 5];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 6] = placeholder[index0 + 6];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 7] = placeholder[index0 + 7];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 8] = placeholder[index0 + 8];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 9] = placeholder[index0 + 9];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 10] = placeholder[index0 + 10];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 11] = placeholder[index0 + 11];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 12] = placeholder[index0 + 12];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 13] = placeholder[index0 + 13];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 14] = placeholder[index0 + 14];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 15] = placeholder[index0 + 15];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 16] = placeholder[index0 + 16];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 17] = placeholder[index0 + 17];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 18] = placeholder[index0 + 18];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 19] = placeholder[index0 + 19];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 20] = placeholder[index0 + 20];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 21] = placeholder[index0 + 21];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 22] = placeholder[index0 + 22];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 23] = placeholder[index0 + 23];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 24] = placeholder[index0 + 24];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 25] = placeholder[index0 + 25];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 26] = placeholder[index0 + 26];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 27] = placeholder[index0 + 27];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 28] = placeholder[index0 + 28];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 29] = placeholder[index0 + 29];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 30] = placeholder[index0 + 30];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 31] = placeholder[index0 + 31];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 32] = placeholder[index0 + 32];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 33] = placeholder[index0 + 33];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 34] = placeholder[index0 + 34];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 35] = placeholder[index0 + 35];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 36] = placeholder[index0 + 36];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 37] = placeholder[index0 + 37];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 38] = placeholder[index0 + 38];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 39] = placeholder[index0 + 39];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 40] = placeholder[index0 + 40];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 41] = placeholder[index0 + 41];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 42] = placeholder[index0 + 42];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 43] = placeholder[index0 + 43];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 44] = placeholder[index0 + 44];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 45] = placeholder[index0 + 45];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 46] = placeholder[index0 + 46];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 47] = placeholder[index0 + 47];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 48] = placeholder[index0 + 48];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 49] = placeholder[index0 + 49];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 50] = placeholder[index0 + 50];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 51] = placeholder[index0 + 51];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 52] = placeholder[index0 + 52];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 53] = placeholder[index0 + 53];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 54] = placeholder[index0 + 54];
}
if ((int)threadIdx.x < 14) {
int index0 = (int)threadIdx.x * 56;
placeholder_d_shared[index0 + 55] = placeholder[index0 + 55];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 1568; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
int index2 = (int)blockIdx.x * 50176 + ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index2];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 16; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 49; ++k_inner) {
int index3 = k_outer_inner * 49 + k_inner;
int index4 = (int)threadIdx.x * 1568 + k_outer_inner * 49 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index3] * placeholder_shared[index4];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index3] * placeholder_shared[index4 + 784]);
}
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index5 = (int)blockIdx.x * 64 + (int)threadIdx.x * 2 + ax1_inner;
int index6 = ax1_inner;
T_relu[index5] = max(T_dense[index6] + placeholder2[index5], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmplwjh6otx | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<4>;
.reg .f32 %f<142>;
.reg .b32 %r<30>;
.reg .b64 %rd<20>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[3136];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[401408];
ld.param.u64 %rd10, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
mov.u32 %r1, %tid.x;
mul.lo.s32 %r13, %r1, 14;
cvta.to.global.u64 %rd11, %rd10;
mul.wide.s32 %rd12, %r13, 4;
add.s64 %rd1, %rd11, %rd12;
ld.global.nc.f32 %f9, [%rd1];
shl.b32 %r14, %r13, 2;
mov.u32 %r15, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r15, %r14;
for.size.2 i in range(0, 13, 1):
st.shared.f32 [%r2+(0+i*4)], %f(9+i*1);
ld.global.nc.f32 %f(10+i*1), [%rd1+(4+i*4)];
st.shared.f32 [%r2+52], %f22;
setp.gt.s32 %p1, %r1, 23;
@%p1 bra $L__BB0_2;
for.size.2 i in range(0, 14, 1):
ld.global.nc.f32 %f(23+i*1), [%rd1+(1792+i*4)];
st.shared.f32 [%r2+(1792+i*4)], %f(23+i*1);
$L__BB0_2:
shl.b32 %r17, %r1, 2;
mov.u32 %r18, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r19, %r18, %r17;
add.s32 %r26, %r19, 1024;
cvta.to.global.u64 %rd13, %rd7;
mul.wide.s32 %rd14, %r1, 4;
add.s64 %rd15, %rd13, %rd14;
add.s64 %rd19, %rd15, 1024;
cvta.to.global.u64 %rd3, %rd9;
cvta.to.global.u64 %rd4, %rd8;
mov.u32 %r27, 0;
$L__BB0_3:
for.size.2 i in range(0, 16, 1):
ld.global.nc.f32 %f(37+i*1), [%rd19+(-1024+i*128)];
st.shared.f32 [%r26+(-1024+i*128)], %f(37+i*1);
add.s32 %r26, %r26, 2048;
add.s64 %rd19, %rd19, 2048;
add.s32 %r27, %r27, 16;
setp.ne.s32 %p2, %r27, 3136;
@%p2 bra $L__BB0_3;
bar.sync 0;
mad.lo.s32 %r22, %r1, 6272, %r18;
add.s32 %r28, %r22, 203868;
mov.f32 %f138, 0f00000000;
mov.u32 %r29, 200704;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(139+i*1), %f138;
$L__BB0_5:
add.s32 %r24, %r15, %r29;
for.size.9 i in range(0, 2, 1):
ld.shared.f32 %f(57+i*9), [%r28+(-203868+i*4)];
ld.shared.f32 %f(58+i*9), [%r24+(-200704+i*4)];
fma.rn.f32 %f(59+i*9), %f(58+i*9), %f(57+i*9), %f(141+i*-82);
ld.shared.f32 %f(60+i*9), [%r28+(-3164+i*4)];
fma.rn.f32 %f(61+i*9), %f(58+i*9), %f(60+i*9), %f(140+i*-79);
ld.shared.f32 %f(62+i*9), [%r28+(-200732+i*4)];
fma.rn.f32 %f(63+i*9), %f(58+i*9), %f(62+i*9), %f(139+i*-76);
ld.shared.f32 %f(64+i*9), [%r28+(-28+i*4)];
fma.rn.f32 %f(65+i*9), %f(58+i*9), %f(64+i*9), %f(138+i*-73);
for.size.9 i in range(0, 5, 1):
ld.shared.f32 %f(75+i*9), [%r28+(-203860+i*4)];
ld.shared.f32 %f(76+i*9), [%r24+(-200696+i*4)];
fma.rn.f32 %f(77+i*9), %f(76+i*9), %f(75+i*9), %f(68+i*9);
ld.shared.f32 %f(78+i*9), [%r28+(-3156+i*4)];
fma.rn.f32 %f(79+i*9), %f(76+i*9), %f(78+i*9), %f(70+i*9);
ld.shared.f32 %f(80+i*9), [%r28+(-200724+i*4)];
fma.rn.f32 %f(81+i*9), %f(76+i*9), %f(80+i*9), %f(72+i*9);
ld.shared.f32 %f(82+i*9), [%r28+(-20+i*4)];
fma.rn.f32 %f(83+i*9), %f(76+i*9), %f(82+i*9), %f(74+i*9);
ld.shared.f32 %f120, [%r28+-203840];
ld.shared.f32 %f121, [%r24+-200676];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(141+i*-1), %f121, %f(120+i*2), %f(113+i*2);
ld.shared.f32 %f(122+i*1), [%r28+(-3136+i*-197568)];
fma.rn.f32 %f139, %f121, %f123, %f117;
ld.shared.f32 %f124, [%r28];
fma.rn.f32 %f138, %f121, %f124, %f119;
for.size.1 i in range(0, 2, 1):
add.s32 %r(28+i*1), %r(28+i*1), 32;
setp.ne.s32 %p3, %r29, 203840;
@%p3 bra $L__BB0_5;
shl.b32 %r25, %r1, 1;
mul.wide.s32 %rd16, %r25, 4;
add.s64 %rd17, %rd3, %rd16;
ld.global.nc.f32 %f125, [%rd17];
add.f32 %f126, %f141, %f125;
mov.f32 %f127, 0f00000000;
max.f32 %f128, %f126, %f127;
add.s64 %rd18, %rd4, %rd16;
for.size.4 i in range(0, 2, 1):
st.global.f32 [%rd18+(0+i*256)], %f(128+i*3);
ld.global.nc.f32 %f(129+i*3), [%rd17+(256+i*-252)];
add.f32 %f(130+i*3), %f(140+i*-1), %f(129+i*3);
max.f32 %f(131+i*3), %f(130+i*3), %f127;
st.global.f32 [%rd18+4], %f134;
ld.global.nc.f32 %f135, [%rd17+260];
add.f32 %f136, %f138, %f135;
max.f32 %f137, %f136, %f127;
st.global.f32 [%rd18+260], %f137;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[784];
__shared__ float placeholder_shared[100352];
T_dense[0] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
int index0 = (int)threadIdx.x * 14;
placeholder_d_shared[index0] = placeholder[index0];
placeholder_d_shared[index0 + 1] = placeholder[index0 + 1];
placeholder_d_shared[index0 + 2] = placeholder[index0 + 2];
placeholder_d_shared[index0 + 3] = placeholder[index0 + 3];
placeholder_d_shared[index0 + 4] = placeholder[index0 + 4];
placeholder_d_shared[index0 + 5] = placeholder[index0 + 5];
placeholder_d_shared[index0 + 6] = placeholder[index0 + 6];
placeholder_d_shared[index0 + 7] = placeholder[index0 + 7];
placeholder_d_shared[index0 + 8] = placeholder[index0 + 8];
placeholder_d_shared[index0 + 9] = placeholder[index0 + 9];
placeholder_d_shared[index0 + 10] = placeholder[index0 + 10];
placeholder_d_shared[index0 + 11] = placeholder[index0 + 11];
placeholder_d_shared[index0 + 12] = placeholder[index0 + 12];
placeholder_d_shared[index0 + 13] = placeholder[index0 + 13];
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 448] = placeholder[index0 + 448];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 449] = placeholder[index0 + 449];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 450] = placeholder[index0 + 450];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 451] = placeholder[index0 + 451];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 452] = placeholder[index0 + 452];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 453] = placeholder[index0 + 453];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 454] = placeholder[index0 + 454];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 455] = placeholder[index0 + 455];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 456] = placeholder[index0 + 456];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 457] = placeholder[index0 + 457];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 458] = placeholder[index0 + 458];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 459] = placeholder[index0 + 459];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 460] = placeholder[index0 + 460];
}
if ((int)threadIdx.x < 24) {
placeholder_d_shared[index0 + 461] = placeholder[index0 + 461];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 3136; ++ax0_ax1_fused_outer_outer) {
int index1 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
placeholder_shared[index1] = placeholder1[index1];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 784; ++k_outer_inner) {
int index2 = k_outer_inner;
int index3 = (int)threadIdx.x * 1568 + k_outer_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 50176]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 784]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 50960]);
}
for (int ax1_inner = 0; ax1_inner < 2; ++ax1_inner) {
int index4 = (int)threadIdx.x * 2 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[index5 + 2] + placeholder2[index4 + 64], 0.000000e+00f);
}
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmpoelv2dad | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<6>;
.reg .f32 %f<126>;
.reg .b32 %r<122>;
.reg .b64 %rd<30>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[448];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[57344];
ld.param.u64 %rd6, [default_function_kernel0_param_0];
ld.param.u64 %rd7, [default_function_kernel0_param_1];
ld.param.u64 %rd8, [default_function_kernel0_param_2];
ld.param.u64 %rd9, [default_function_kernel0_param_3];
cvta.to.global.u64 %rd1, %rd7;
cvta.to.global.u64 %rd2, %rd6;
cvta.to.global.u64 %rd3, %rd8;
cvta.to.global.u64 %rd4, %rd9;
mov.u32 %r1, %tid.x;
shl.b32 %r22, %r1, 2;
mov.u32 %r23, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r2, %r23, %r22;
mul.lo.s32 %r3, %r1, 112;
add.s32 %r4, %r1, 224;
mov.u32 %r24, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r5, %r24, %r22;
mov.f32 %f122, 0f00000000;
mov.u32 %r116, 0;
for.size.1 i in range(0, 3, 1):
mov.f32 %f(123+i*1), %f122;
$L__BB0_1:
bar.sync 0;
mul.lo.s32 %r7, %r116, 112;
add.s32 %r25, %r7, %r1;
mul.wide.s32 %rd10, %r25, 4;
add.s64 %rd5, %rd2, %rd10;
for.size.2 i in range(0, 3, 1):
ld.global.nc.f32 %f(21+i*1), [%rd5+(0+i*128)];
st.shared.f32 [%r2+(0+i*128)], %f(21+i*1);
setp.gt.s32 %p1, %r1, 15;
@%p1 bra $L__BB0_3;
ld.global.nc.f32 %f24, [%rd5+384];
st.shared.f32 [%r2+384], %f24;
$L__BB0_3:
mov.u32 %r119, 512;
mov.u32 %r117, %r1;
mov.u32 %r118, %r4;
$L__BB0_4:
mul.hi.s32 %r27, %r117, -1840700269;
add.s32 %r28, %r27, %r117;
shr.u32 %r29, %r28, 31;
shr.s32 %r30, %r28, 6;
add.s32 %r31, %r30, %r29;
mul.lo.s32 %r32, %r31, 112;
sub.s32 %r33, %r117, %r32;
add.s32 %r34, %r33, %r7;
mad.lo.s32 %r35, %r31, 784, %r34;
mul.wide.s32 %rd11, %r35, 4;
add.s64 %rd12, %rd1, %rd11;
ld.global.nc.f32 %f25, [%rd12];
add.s32 %r36, %r5, %r119;
for.size.14 i in range(0, 4, 1):
st.shared.f32 [%r36+(-512+i*128)], %f(25+i*1);
add.s32 %r(37+i*10), %r117, (32+i*32);
mul.hi.s32 %r(38+i*10), %r(37+i*10), -1840700269;
add.s32 %r(39+i*10), %r(38+i*10), %r(37+i*10);
shr.u32 %r(40+i*10), %r(39+i*10), 31;
shr.s32 %r(41+i*10), %r(39+i*10), 6;
add.s32 %r(42+i*10), %r(41+i*10), %r(40+i*10);
mul.lo.s32 %r(43+i*10), %r(42+i*10), 112;
sub.s32 %r(44+i*10), %r(37+i*10), %r(43+i*10);
add.s32 %r(45+i*10), %r(44+i*10), %r7;
mad.lo.s32 %r(46+i*10), %r(42+i*10), 784, %r(45+i*10);
mul.wide.s32 %rd(13+i*2), %r(46+i*10), 4;
add.s64 %rd(14+i*2), %rd1, %rd(13+i*2);
ld.global.nc.f32 %f(26+i*1), [%rd(14+i*2)+0];
for.size.14 i in range(0, 2, 1):
st.shared.f32 [%r36+(0+i*128)], %f(29+i*1);
add.s32 %r(77+i*10), %r118, (-64+i*32);
mul.hi.s32 %r(78+i*10), %r(77+i*10), -1840700269;
add.s32 %r(79+i*10), %r(78+i*10), %r(77+i*10);
shr.u32 %r(80+i*10), %r(79+i*10), 31;
shr.s32 %r(81+i*10), %r(79+i*10), 6;
add.s32 %r(82+i*10), %r(81+i*10), %r(80+i*10);
mul.lo.s32 %r(83+i*10), %r(82+i*10), 112;
sub.s32 %r(84+i*10), %r(77+i*10), %r(83+i*10);
add.s32 %r(85+i*10), %r(84+i*10), %r7;
mad.lo.s32 %r(86+i*10), %r(82+i*10), 784, %r(85+i*10);
mul.wide.s32 %rd(21+i*2), %r(86+i*10), 4;
add.s64 %rd(22+i*2), %rd1, %rd(21+i*2);
ld.global.nc.f32 %f(30+i*1), [%rd(22+i*2)+0];
st.shared.f32 [%r36+256], %f31;
mul.hi.s32 %r97, %r118, -1840700269;
add.s32 %r98, %r97, %r118;
shr.u32 %r99, %r98, 31;
shr.s32 %r100, %r98, 6;
add.s32 %r101, %r100, %r99;
mul.lo.s32 %r102, %r101, 112;
sub.s32 %r103, %r118, %r102;
add.s32 %r104, %r103, %r7;
mad.lo.s32 %r105, %r101, 784, %r104;
mul.wide.s32 %rd25, %r105, 4;
add.s64 %rd26, %rd1, %rd25;
ld.global.nc.f32 %f32, [%rd26];
st.shared.f32 [%r36+384], %f32;
for.size.1 i in range(0, 2, 1):
add.s32 %r(118+i*-1), %r(118+i*-1), 256;
add.s32 %r119, %r119, 1024;
setp.ne.s32 %p2, %r119, 57856;
@%p2 bra $L__BB0_4;
bar.sync 0;
mov.u32 %r106, 0;
mov.u32 %r120, %r106;
$L__BB0_6:
mul.lo.s32 %r15, %r120, 56;
add.s32 %r16, %r15, %r3;
mov.u32 %r121, %r106;
$L__BB0_7:
for.size.3 i in range(0, 2, 1):
add.s32 %r(108+i*4), %r(121+i*-105), %r(15+i*106);
shl.b32 %r(109+i*4), %r(108+i*4), 2;
add.s32 %r(111+i*4), %r(23+i*1), %r(109+i*4);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(33+i*1), [%r(115+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(35+i*2), %f34, %f(33+i*3), %f(122+i*1);
ld.shared.f32 %f(36+i*2), [%r115+(14336+i*14336)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(39+i*2), %f34, %f(38+i*2), %f(124+i*1);
ld.shared.f32 %f(40+i*2), [%r115+(43008+i*-43004)];
for.size.9 i in range(0, 6, 1):
ld.shared.f32 %f(43+i*9), [%r111+(4+i*4)];
fma.rn.f32 %f(44+i*9), %f(43+i*9), %f(42+i*9), %f(35+i*9);
ld.shared.f32 %f(45+i*9), [%r115+(14340+i*4)];
fma.rn.f32 %f(46+i*9), %f(43+i*9), %f(45+i*9), %f(37+i*9);
ld.shared.f32 %f(47+i*9), [%r115+(28676+i*4)];
fma.rn.f32 %f(48+i*9), %f(43+i*9), %f(47+i*9), %f(39+i*9);
ld.shared.f32 %f(49+i*9), [%r115+(43012+i*4)];
fma.rn.f32 %f(50+i*9), %f(43+i*9), %f(49+i*9), %f(41+i*9);
ld.shared.f32 %f(51+i*9), [%r115+(8+i*4)];
ld.shared.f32 %f97, [%r111+28];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(122+i*1), %f97, %f(96+i*2), %f(89+i*2);
ld.shared.f32 %f(98+i*1), [%r115+(14364+i*14336)];
fma.rn.f32 %f124, %f97, %f99, %f93;
ld.shared.f32 %f100, [%r115+43036];
fma.rn.f32 %f125, %f97, %f100, %f95;
add.s32 %r121, %r121, 8;
setp.ne.s32 %p3, %r121, 56;
@%p3 bra $L__BB0_7;
add.s32 %r120, %r120, 1;
setp.lt.u32 %p4, %r120, 2;
@%p4 bra $L__BB0_6;
add.s32 %r116, %r116, 1;
setp.lt.u32 %p5, %r116, 7;
@%p5 bra $L__BB0_1;
mul.wide.s32 %rd27, %r1, 4;
add.s64 %rd28, %rd4, %rd27;
ld.global.nc.f32 %f101, [%rd28];
add.f32 %f102, %f122, %f101;
mov.f32 %f103, 0f00000000;
max.f32 %f104, %f102, %f103;
add.s64 %rd29, %rd3, %rd27;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd29+(0+i*128)], %f(104+i*3);
ld.global.nc.f32 %f(105+i*3), [%rd28+(128+i*128)];
add.f32 %f(106+i*3), %f(123+i*1), %f(105+i*3);
max.f32 %f(107+i*3), %f(106+i*3), %f103;
st.global.f32 [%rd29+384], %f113;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[4];
__shared__ float placeholder_d_shared[112];
__shared__ float placeholder_shared[14336];
T_dense[0] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 7; ++k_outer_outer) {
__syncthreads();
int index0 = (int)threadIdx.x;
int index1 = k_outer_outer * 112 + (int)threadIdx.x;
placeholder_d_shared[index0] = placeholder[index1];
placeholder_d_shared[index0 + 32] = placeholder[index1 + 32];
placeholder_d_shared[index0 + 64] = placeholder[index1 + 64];
if ((int)threadIdx.x < 16) {
placeholder_d_shared[index0 + 96] = placeholder[index1 + 96];
}
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 448; ++ax0_ax1_fused_outer_outer) {
int index2 = ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x;
int index3 = (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) / 112 * 784 + k_outer_outer * 112 + (ax0_ax1_fused_outer_outer * 32 + (int)threadIdx.x) % 112;
placeholder_shared[index2] = placeholder1[index3];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 56; ++k_inner) {
int index4 = k_outer_inner * 56 + k_inner;
int index5 = (int)threadIdx.x * 112 + k_outer_inner * 56 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index4] * placeholder_shared[index5];
T_dense[1] = T_dense[1] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 3584]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 7168]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index4] * placeholder_shared[index5 + 10752]);
}
}
}
int index0 = (int)threadIdx.x;
T_relu[index0] = max(T_dense[0] + placeholder2[index0], 0.000000e+00f);
T_relu[index0 + 32] = max(T_dense[1] + placeholder2[index0 + 32], 0.000000e+00f);
T_relu[index0 + 64] = max(T_dense[2] + placeholder2[index0 + 64], 0.000000e+00f);
T_relu[index0 + 96] = max(T_dense[3] + placeholder2[index0 + 96], 0.000000e+00f);
} |
[407f97a0b39db63cc31f3e635caadfd0,1,784,128,784,128,1,128] | tmp5_hly50l | default_function_kernel0 | .visible .entry default_function_kernel0(
.param .u64 default_function_kernel0_param_0,
.param .u64 default_function_kernel0_param_1,
.param .u64 default_function_kernel0_param_2,
.param .u64 default_function_kernel0_param_3
)
{
.reg .pred %p<5>;
.reg .f32 %f<779>;
.reg .b32 %r<95>;
.reg .b64 %rd<32>;
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E20placeholder_d_shared[32];
// demoted variable
.shared .align 4 .b8 _ZZ24default_function_kernel0E18placeholder_shared[4096];
ld.param.u64 %rd1, [default_function_kernel0_param_0];
ld.param.u64 %rd2, [default_function_kernel0_param_1];
mov.u32 %r12, %tid.x;
mov.u32 %r91, 0;
mov.f32 %f715, 0f00000000;
cvta.to.global.u64 %rd5, %rd1;
cvta.to.global.u64 %rd8, %rd2;
for.size.1 i in range(0, 63, 1):
mov.f32 %f(716+i*1), %f715;
$L__BB0_1:
mov.u32 %r88, %tid.x;
mov.u32 %r92, 0;
for.size.1 i in range(0, 2, 1):
shl.b32 %r(86+i*-1), %r88, (3+i*-2);
bar.sync 0;
shl.b32 %r16, %r91, 3;
add.s32 %r17, %r16, %r85;
mul.wide.s32 %rd6, %r17, 4;
add.s64 %rd7, %rd5, %rd6;
ld.global.nc.v2.u32 {%r18, %r19}, [%rd7];
mov.u32 %r23, _ZZ24default_function_kernel0E20placeholder_d_shared;
add.s32 %r24, %r23, %r86;
st.shared.v2.u32 [%r24], {%r18, %r19};
ld.global.nc.v2.u32 {%r25, %r26}, [%rd7+16];
st.shared.v2.u32 [%r24+16], {%r25, %r26};
$L__BB0_2:
for.size.1 i in range(0, 2, 1):
shl.b32 %r(81+i*-51), %r(91+i*1), (3+i*-2);
add.s32 %r31, %r30, %r12;
shr.s32 %r32, %r31, 3;
and.b32 %r33, %r12, 7;
bfi.b32 %r35, %r91, %r33, 3, 29;
mad.lo.s32 %r36, %r32, 784, %r35;
mul.wide.s32 %rd9, %r36, 4;
add.s64 %rd10, %rd8, %rd9;
ld.global.nc.f32 %f321, [%rd10];
shl.b32 %r37, %r31, 2;
mov.u32 %r38, _ZZ24default_function_kernel0E18placeholder_shared;
add.s32 %r39, %r38, %r37;
st.shared.f32 [%r39], %f321;
add.s32 %r40, %r31, 2;
shr.s32 %r41, %r40, 3;
add.s32 %r42, %r12, 2;
and.b32 %r43, %r42, 7;
bfi.b32 %r44, %r91, %r43, 3, 29;
mad.lo.s32 %r45, %r41, 784, %r44;
mul.wide.s32 %rd11, %r45, 4;
add.s64 %rd12, %rd8, %rd11;
ld.global.nc.f32 %f322, [%rd12];
st.shared.f32 [%r39+8], %f322;
add.s32 %r46, %r31, 4;
shr.s32 %r47, %r46, 3;
xor.b32 %r48, %r33, 4;
or.b32 %r49, %r48, %r81;
mad.lo.s32 %r50, %r47, 784, %r49;
mul.wide.s32 %rd13, %r50, 4;
add.s64 %rd14, %rd8, %rd13;
ld.global.nc.f32 %f323, [%rd14];
st.shared.f32 [%r39+16], %f323;
add.s32 %r51, %r31, 6;
shr.s32 %r52, %r51, 3;
add.s32 %r53, %r12, 6;
and.b32 %r54, %r53, 7;
bfi.b32 %r55, %r91, %r54, 3, 29;
for.size.7 i in range(0, 2, 1):
mad.lo.s32 %r(56+i*3), %r(52+i*6), 784, %r(55+i*-20);
mul.wide.s32 %rd(15+i*2), %r(56+i*3), 4;
add.s64 %rd(16+i*2), %rd8, %rd(15+i*2);
ld.global.nc.f32 %f(324+i*1), [%rd(16+i*2)+0];
st.shared.f32 [%r39+(24+i*8)], %f(324+i*1);
add.s32 %r(57+i*3), %r31, (8+i*2);
shr.s32 %r(58+i*3), %r(57+i*3), 3;
for.size.7 i in range(0, 2, 1):
mad.lo.s32 %r(62+i*3), %r(61+i*3), 784, %r(44+i*5);
mul.wide.s32 %rd(19+i*2), %r(62+i*3), 4;
add.s64 %rd(20+i*2), %rd8, %rd(19+i*2);
ld.global.nc.f32 %f(326+i*1), [%rd(20+i*2)+0];
st.shared.f32 [%r39+(40+i*8)], %f(326+i*1);
add.s32 %r(63+i*3), %r31, (12+i*2);
shr.s32 %r(64+i*3), %r(63+i*3), 3;
mad.lo.s32 %r68, %r67, 784, %r55;
mul.wide.s32 %rd23, %r68, 4;
add.s64 %rd24, %rd8, %rd23;
ld.global.nc.f32 %f328, [%rd24];
st.shared.f32 [%r39+56], %f328;
add.s32 %r92, %r92, 8;
setp.ne.s32 %p1, %r92, 512;
@%p1 bra $L__BB0_2;
bar.sync 0;
mov.u32 %r93, 0;
$L__BB0_4:
mov.u32 %r94, 0;
$L__BB0_5:
shl.b32 %r89, %r93, 2;
mov.u32 %r84, _ZZ24default_function_kernel0E18placeholder_shared;
shl.b32 %r83, %r12, 7;
mov.u32 %r82, _ZZ24default_function_kernel0E20placeholder_d_shared;
for.size.3 i in range(0, 2, 1):
add.s32 %r(71+i*4), %r(94+i*-23), %r(89+i*-6);
shl.b32 %r(72+i*4), %r(71+i*4), 2;
add.s32 %r(74+i*4), %r(82+i*2), %r(72+i*4);
for.size.1 i in range(0, 2, 1):
ld.shared.f32 %f(329+i*1), [%r(78+i*-4)+0];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(778+i*-1), %f330, %f(329+i*2), %f(778+i*-1);
ld.shared.f32 %f(331+i*1), [%r78+(1024+i*1024)];
for.size.2 i in range(0, 2, 1):
fma.rn.f32 %f(776+i*-1), %f330, %f(332+i*1), %f(776+i*-1);
ld.shared.f32 %f(333+i*1), [%r78+(3072+i*-3040)];
for.size.2 i in range(0, 3, 1):
fma.rn.f32 %f(774+i*-1), %f330, %f(334+i*1), %f(774+i*-1);
ld.shared.f32 %f(335+i*1), [%r78+(1056+i*1024)];
for.size.3 i in range(0, 14, 1):
for.size.2 j in range(0, 4, 1):
fma.rn.f32 %f(771+i*-4+j*-1), %f330, %f(337+i*4+j*1), %f(771+i*-4+j*-1);
ld.shared.f32 %f(338+i*4+j*1), [%r78+(64+i*32+j*1024)];
fma.rn.f32 %f715, %f330, %f393, %f715;
add.s32 %r94, %r94, 1;
setp.ne.s32 %p2, %r94, 4;
@%p2 bra $L__BB0_5;
add.s32 %r93, %r93, 1;
setp.lt.u32 %p3, %r93, 2;
@%p3 bra $L__BB0_4;
add.s32 %r91, %r91, 1;
setp.lt.u32 %p4, %r91, 98;
@%p4 bra $L__BB0_1;
ld.param.u64 %rd31, [default_function_kernel0_param_2];
ld.param.u64 %rd30, [default_function_kernel0_param_3];
shl.b32 %r80, %r12, 4;
cvta.to.global.u64 %rd25, %rd30;
mul.wide.s32 %rd26, %r80, 4;
add.s64 %rd27, %rd25, %rd26;
ld.global.nc.f32 %f394, [%rd27];
add.f32 %f395, %f778, %f394;
mov.f32 %f396, 0f00000000;
max.f32 %f397, %f395, %f396;
cvta.to.global.u64 %rd28, %rd31;
add.s64 %rd29, %rd28, %rd26;
for.size.4 i in range(0, 3, 1):
st.global.f32 [%rd29+(0+i*128)], %f(397+i*3);
ld.global.nc.f32 %f(398+i*3), [%rd27+(128+i*128)];
add.f32 %f(399+i*3), %f(777+i*-1), %f(398+i*3);
max.f32 %f(400+i*3), %f(399+i*3), %f396;
for.size.10 i in range(0, 15, 1):
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(384+i*4+j*-380)], %f(406+i*12+j*3);
ld.global.nc.f32 %f(407+i*12+j*3), [%rd27+(4+i*4+j*128)];
add.f32 %f(408+i*12+j*3), %f(774+i*-4+j*-1), %f(407+i*12+j*3);
max.f32 %f(409+i*12+j*3), %f(408+i*12+j*3), %f396;
for.size.4 j in range(0, 2, 1):
st.global.f32 [%rd29+(132+i*4+j*128)], %f(412+i*12+j*3);
ld.global.nc.f32 %f(413+i*12+j*3), [%rd27+(260+i*4+j*128)];
add.f32 %f(414+i*12+j*3), %f(772+i*-4+j*-1), %f(413+i*12+j*3);
max.f32 %f(415+i*12+j*3), %f(414+i*12+j*3), %f396;
st.global.f32 [%rd29+444], %f586;
ret;
}
| extern "C" __global__ void default_function_kernel0(float* __restrict__ placeholder, float* __restrict__ placeholder1, float* __restrict__ T_relu, float* __restrict__ placeholder2) {
float T_dense[64];
__shared__ float placeholder_d_shared[8];
__shared__ float placeholder_shared[1024];
T_dense[0] = 0.000000e+00f;
T_dense[16] = 0.000000e+00f;
T_dense[32] = 0.000000e+00f;
T_dense[48] = 0.000000e+00f;
T_dense[1] = 0.000000e+00f;
T_dense[17] = 0.000000e+00f;
T_dense[33] = 0.000000e+00f;
T_dense[49] = 0.000000e+00f;
T_dense[2] = 0.000000e+00f;
T_dense[18] = 0.000000e+00f;
T_dense[34] = 0.000000e+00f;
T_dense[50] = 0.000000e+00f;
T_dense[3] = 0.000000e+00f;
T_dense[19] = 0.000000e+00f;
T_dense[35] = 0.000000e+00f;
T_dense[51] = 0.000000e+00f;
T_dense[4] = 0.000000e+00f;
T_dense[20] = 0.000000e+00f;
T_dense[36] = 0.000000e+00f;
T_dense[52] = 0.000000e+00f;
T_dense[5] = 0.000000e+00f;
T_dense[21] = 0.000000e+00f;
T_dense[37] = 0.000000e+00f;
T_dense[53] = 0.000000e+00f;
T_dense[6] = 0.000000e+00f;
T_dense[22] = 0.000000e+00f;
T_dense[38] = 0.000000e+00f;
T_dense[54] = 0.000000e+00f;
T_dense[7] = 0.000000e+00f;
T_dense[23] = 0.000000e+00f;
T_dense[39] = 0.000000e+00f;
T_dense[55] = 0.000000e+00f;
T_dense[8] = 0.000000e+00f;
T_dense[24] = 0.000000e+00f;
T_dense[40] = 0.000000e+00f;
T_dense[56] = 0.000000e+00f;
T_dense[9] = 0.000000e+00f;
T_dense[25] = 0.000000e+00f;
T_dense[41] = 0.000000e+00f;
T_dense[57] = 0.000000e+00f;
T_dense[10] = 0.000000e+00f;
T_dense[26] = 0.000000e+00f;
T_dense[42] = 0.000000e+00f;
T_dense[58] = 0.000000e+00f;
T_dense[11] = 0.000000e+00f;
T_dense[27] = 0.000000e+00f;
T_dense[43] = 0.000000e+00f;
T_dense[59] = 0.000000e+00f;
T_dense[12] = 0.000000e+00f;
T_dense[28] = 0.000000e+00f;
T_dense[44] = 0.000000e+00f;
T_dense[60] = 0.000000e+00f;
T_dense[13] = 0.000000e+00f;
T_dense[29] = 0.000000e+00f;
T_dense[45] = 0.000000e+00f;
T_dense[61] = 0.000000e+00f;
T_dense[14] = 0.000000e+00f;
T_dense[30] = 0.000000e+00f;
T_dense[46] = 0.000000e+00f;
T_dense[62] = 0.000000e+00f;
T_dense[15] = 0.000000e+00f;
T_dense[31] = 0.000000e+00f;
T_dense[47] = 0.000000e+00f;
T_dense[63] = 0.000000e+00f;
for (int k_outer_outer = 0; k_outer_outer < 98; ++k_outer_outer) {
__syncthreads();
((float2*)(placeholder_d_shared + (int)threadIdx.x * 2))[0] = ((float2*)(placeholder + k_outer_outer * 8 + (int)threadIdx.x * 2))[0];
((float2*)(placeholder_d_shared + (int)threadIdx.x * 2 + 4))[0] = ((float2*)(placeholder + k_outer_outer * 8 + (int)threadIdx.x * 2 + 4))[0];
for (int ax0_ax1_fused_outer_outer = 0; ax0_ax1_fused_outer_outer < 512; ++ax0_ax1_fused_outer_outer) {
int index0 = ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x;
int index1 = (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x >> 3) * 784 + k_outer_outer * 8 + (ax0_ax1_fused_outer_outer * 2 + (int)threadIdx.x & 7);
placeholder_shared[index0] = placeholder1[index1];
}
__syncthreads();
for (int k_outer_inner = 0; k_outer_inner < 2; ++k_outer_inner) {
for (int k_inner = 0; k_inner < 4; ++k_inner) {
int index2 = k_outer_inner * 4 + k_inner;
int index3 = (int)threadIdx.x * 128 + k_outer_inner * 4 + k_inner;
T_dense[0] = T_dense[0] + placeholder_d_shared[index2] * placeholder_shared[index3];
T_dense[16] = T_dense[16] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 256]);
T_dense[32] = T_dense[32] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 512]);
T_dense[48] = T_dense[48] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 768]);
T_dense[1] = T_dense[1] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 8]);
T_dense[17] = T_dense[17] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 264]);
T_dense[33] = T_dense[33] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 520]);
T_dense[49] = T_dense[49] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 776]);
T_dense[2] = T_dense[2] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 16]);
T_dense[18] = T_dense[18] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 272]);
T_dense[34] = T_dense[34] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 528]);
T_dense[50] = T_dense[50] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 784]);
T_dense[3] = T_dense[3] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 24]);
T_dense[19] = T_dense[19] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 280]);
T_dense[35] = T_dense[35] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 536]);
T_dense[51] = T_dense[51] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 792]);
T_dense[4] = T_dense[4] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 32]);
T_dense[20] = T_dense[20] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 288]);
T_dense[36] = T_dense[36] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 544]);
T_dense[52] = T_dense[52] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 800]);
T_dense[5] = T_dense[5] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 40]);
T_dense[21] = T_dense[21] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 296]);
T_dense[37] = T_dense[37] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 552]);
T_dense[53] = T_dense[53] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 808]);
T_dense[6] = T_dense[6] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 48]);
T_dense[22] = T_dense[22] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 304]);
T_dense[38] = T_dense[38] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 560]);
T_dense[54] = T_dense[54] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 816]);
T_dense[7] = T_dense[7] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 56]);
T_dense[23] = T_dense[23] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 312]);
T_dense[39] = T_dense[39] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 568]);
T_dense[55] = T_dense[55] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 824]);
T_dense[8] = T_dense[8] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 64]);
T_dense[24] = T_dense[24] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 320]);
T_dense[40] = T_dense[40] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 576]);
T_dense[56] = T_dense[56] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 832]);
T_dense[9] = T_dense[9] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 72]);
T_dense[25] = T_dense[25] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 328]);
T_dense[41] = T_dense[41] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 584]);
T_dense[57] = T_dense[57] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 840]);
T_dense[10] = T_dense[10] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 80]);
T_dense[26] = T_dense[26] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 336]);
T_dense[42] = T_dense[42] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 592]);
T_dense[58] = T_dense[58] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 848]);
T_dense[11] = T_dense[11] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 88]);
T_dense[27] = T_dense[27] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 344]);
T_dense[43] = T_dense[43] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 600]);
T_dense[59] = T_dense[59] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 856]);
T_dense[12] = T_dense[12] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 96]);
T_dense[28] = T_dense[28] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 352]);
T_dense[44] = T_dense[44] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 608]);
T_dense[60] = T_dense[60] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 864]);
T_dense[13] = T_dense[13] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 104]);
T_dense[29] = T_dense[29] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 360]);
T_dense[45] = T_dense[45] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 616]);
T_dense[61] = T_dense[61] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 872]);
T_dense[14] = T_dense[14] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 112]);
T_dense[30] = T_dense[30] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 368]);
T_dense[46] = T_dense[46] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 624]);
T_dense[62] = T_dense[62] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 880]);
T_dense[15] = T_dense[15] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 120]);
T_dense[31] = T_dense[31] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 376]);
T_dense[47] = T_dense[47] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 632]);
T_dense[63] = T_dense[63] + (placeholder_d_shared[index2] * placeholder_shared[index3 + 888]);
}
}
}
for (int ax1_inner = 0; ax1_inner < 16; ++ax1_inner) {
int index4 = (int)threadIdx.x * 16 + ax1_inner;
int index5 = ax1_inner;
T_relu[index4] = max(T_dense[index5] + placeholder2[index4], 0.000000e+00f);
T_relu[index4 + 32] = max(T_dense[index5 + 16] + placeholder2[index4 + 32], 0.000000e+00f);
T_relu[index4 + 64] = max(T_dense[index5 + 32] + placeholder2[index4 + 64], 0.000000e+00f);
T_relu[index4 + 96] = max(T_dense[index5 + 48] + placeholder2[index4 + 96], 0.000000e+00f);
}
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.