file stringlengths 6 242 | name stringlengths 1 11 | kernel stringclasses 4
values | input stringlengths 531 14.6k | output stringlengths 176 17.2k |
|---|---|---|---|---|
Convolution_Broadcast_Add_4531 | 21 | null | .visible .entry _Z7Group16PfS_S_S_(
.param .u64 _Z7Group16PfS_S_S__param_0,
.param .u64 _Z7Group16PfS_S_S__param_1,
.param .u64 _Z7Group16PfS_S_S__param_2,
.param .u64 _Z7Group16PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<3>;
.reg .f32 %f<126>;
.reg .b32 %r<135>;
.reg .b64 %rd<17>;
// demoted var... | __global__ void __launch_bounds__(128) Group16(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[8];
__shared__ float input1_shared[256];
__shared__ float input2_shared[4096];
mediate1_local[0] = 0.000000e+00f;
mediate1_loca... |
Convolution_Broadcast_Add_4531 | 11 | null | .visible .entry _Z7Group16PfS_S_S_(
.param .u64 _Z7Group16PfS_S_S__param_0,
.param .u64 _Z7Group16PfS_S_S__param_1,
.param .u64 _Z7Group16PfS_S_S__param_2,
.param .u64 _Z7Group16PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<4>;
.reg .f32 %f<150>;
.reg .b32 %r<203>;
.reg .b64 %rd<17>;
// demoted var... | __global__ void __launch_bounds__(128) Group16(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[16];
__shared__ float input1_shared[8192];
__shared__ float input2_shared[256];
mediate1_local[0] = 0.000000e+00f;
mediate1_loc... |
Convolution_Broadcast_Add_4531 | 2 | null | .visible .entry _Z7Group16PfS_S_S_(
.param .u64 _Z7Group16PfS_S_S__param_0,
.param .u64 _Z7Group16PfS_S_S__param_1,
.param .u64 _Z7Group16PfS_S_S__param_2,
.param .u64 _Z7Group16PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<4>;
.reg .f32 %f<242>;
.reg .b32 %r<158>;
.reg .b64 %rd<19>;
// demoted var... | __global__ void __launch_bounds__(128) Group16(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[32];
__shared__ float input1_shared[4096];
__shared__ float input2_shared[1024];
mediate1_local[0] = 0.000000e+00f;
mediate1_lo... |
Convolution_Broadcast_Add_4531 | 3 | null | .visible .entry _Z7Group16PfS_S_S_(
.param .u64 _Z7Group16PfS_S_S__param_0,
.param .u64 _Z7Group16PfS_S_S__param_1,
.param .u64 _Z7Group16PfS_S_S__param_2,
.param .u64 _Z7Group16PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<4>;
.reg .f32 %f<150>;
.reg .b32 %r<140>;
.reg .b64 %rd<17>;
// demoted var... | __global__ void __launch_bounds__(128) Group16(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[16];
__shared__ float input1_shared[4096];
__shared__ float input2_shared[512];
mediate1_local[0] = 0.000000e+00f;
mediate1_loc... |
Reshape_Add_4296 | 53 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 64, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<11>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS_... | __global__ void __launch_bounds__(64) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = (int)blockIdx.x * 64 + (int)threadIdx.x;
int index1 = ((int)blockIdx.x & 7) * 32768 + (int)threadIdx.x * 512 + ((int)blockIdx.x >> 3);
output0[index0] = input1... |
Reshape_Add_4296 | 63 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<385>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = (int)blockIdx.x * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)threadIdx.x & 7);
int index1 = ((int)threadIdx.x & 7) * 512 + (int)blockIdx.x * 32 + ((int)threa... |
Reshape_Add_4296 | 21 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<97>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)threadIdx.x >> 3) * 512 + (int)blockIdx.x * 8 + ((int)threadIdx.x & 7);
int index1 = (int)blockIdx.x * 4096 + ((int)threadIdx.x & 7) * 512 + ((int)threadI... |
Reshape_Add_4296 | 20 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<97>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 5) * 131072 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 31) * 16 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 31) * 81... |
Reshape_Add_4296 | 48 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<11>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = (int)blockIdx.x * 128 + (int)threadIdx.x;
int index1 = ((int)blockIdx.x & 3) * 65536 + (int)threadIdx.x * 512 + ((int)blockIdx.x >> 2);
output0[index0] = inpu... |
Reshape_Add_4296 | 16 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<49>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 2) * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 3) * 65536... |
Reshape_Add_4296 | 82 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 16, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS_... | __global__ void __launch_bounds__(16) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 9) * 8192 + (int)threadIdx.x * 512 + ((int)blockIdx.x & 511);
int index1 = ((int)blockIdx.x & 511) * 512 + ((int)blockIdx.x >> 9) * 16 + (int... |
Reshape_Add_4296 | 73 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<7>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 9) * 131072 + (int)threadIdx.x * 512 + ((int)blockIdx.x & 511);
int index1 = ((int)blockIdx.x & 511) * 512 + ((int)blockIdx.x >> 9) * 256 + ... |
Reshape_Add_4296 | 55 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 64, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS_... | __global__ void __launch_bounds__(64) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 8) * 16384 + ((int)threadIdx.x >> 1) * 512 + ((int)blockIdx.x & 255) * 2 + ((int)threadIdx.x & 1);
int index1 = ((int)blockIdx.x & 255) * 102... |
Reshape_Add_4296 | 38 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<193>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = (int)blockIdx.x * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)threadIdx.x & 7);
int index1 = ((int)threadIdx.x & 7) * 512 + (int)blockIdx.x * 16 + ((int)thread... |
Reshape_Add_4296 | 62 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<385>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 2) * 65536 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 3) * 6553... |
Reshape_Add_4296 | 42 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<193>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 4) * 131072 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 15) * 32 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 15) * 16... |
Reshape_Add_4296 | 17 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<49>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 1) * 4096 + ((int)threadIdx.x >> 4) * 512 + ((int)blockIdx.x & 1) * 256 + ((int)threadIdx.x & 15);
int index1 = ((int)blockIdx.x & 1) * 1310... |
Reshape_Add_4296 | 57 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<13>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 8) * 131072 + ((int)threadIdx.x >> 1) * 512 + ((int)blockIdx.x & 255) * 2 + ((int)threadIdx.x & 1);
int index1 = ((int)blockIdx.x & 255) * 1... |
Reshape_Add_4296 | 31 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<49>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = (int)blockIdx.x * 2048 + ((int)threadIdx.x >> 5) * 512 + ((int)threadIdx.x & 31);
int index1 = ((int)threadIdx.x & 31) * 512 + (int)blockIdx.x * 4 + ((int)threa... |
Reshape_Add_4296 | 22 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<97>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 3) * 32768 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 7) * 64 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 7) * 32768... |
Reshape_Add_4296 | 41 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<193>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 3) * 65536 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 7) * 64 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 7) * 32768... |
Reshape_Add_4296 | 11 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<49>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 3) * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 7) * 64 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 7) * 32768... |
Reshape_Add_4296 | 68 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 32, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS_... | __global__ void __launch_bounds__(32) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 8) * 8192 + ((int)threadIdx.x >> 1) * 512 + ((int)blockIdx.x & 255) * 2 + ((int)threadIdx.x & 1);
int index1 = ((int)blockIdx.x & 255) * 1024... |
Reshape_Add_4296 | 18 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<97>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 2) * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 3) * 6553... |
Reshape_Add_4296 | 60 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<385>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 1) * 32768 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 1) * 256 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 1) * 1310... |
Reshape_Add_4296 | 13 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<25>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 3) * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 7) * 64 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 7) * 32768 ... |
Reshape_Add_4296 | 1 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<13>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 5) * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 31) * 16 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 31) * 819... |
Reshape_Add_4296 | 64 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<385>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)threadIdx.x >> 3) * 512 + (int)blockIdx.x * 32 + ((int)threadIdx.x & 7);
int index1 = (int)blockIdx.x * 16384 + ((int)threadIdx.x & 7) * 512 + ((int)threa... |
Reshape_Add_4296 | 12 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<49>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 6) * 131072 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 63) * 8 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 63) * 409... |
Reshape_Add_4296 | 67 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 32, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS_... | __global__ void __launch_bounds__(32) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 5) * 1024 + ((int)threadIdx.x >> 4) * 512 + ((int)blockIdx.x & 31) * 16 + ((int)threadIdx.x & 15);
int index1 = ((int)blockIdx.x & 31) * 8192... |
Reshape_Add_4296 | 26 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 5) * 4096 + ((int)threadIdx.x >> 4) * 512 + ((int)blockIdx.x & 31) * 16 + ((int)threadIdx.x & 15);
int index1 = ((int)blockIdx.x & 31) * 819... |
Reshape_Add_4296 | 14 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<49>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 4) * 32768 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 15) * 32 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 15) * 163... |
Reshape_Add_4296 | 40 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<193>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 2) * 32768 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 3) * 6553... |
Reshape_Add_4296 | 44 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<193>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)threadIdx.x >> 3) * 512 + (int)blockIdx.x * 16 + ((int)threadIdx.x & 7);
int index1 = (int)blockIdx.x * 8192 + ((int)threadIdx.x & 7) * 512 + ((int)thread... |
Reshape_Add_4296 | 84 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 8, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS_S... | __global__ void __launch_bounds__(8) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 9) * 4096 + (int)threadIdx.x * 512 + ((int)blockIdx.x & 511);
int index1 = ((int)blockIdx.x & 511) * 512 + ((int)blockIdx.x >> 9) * 8 + (int)t... |
Reshape_Add_4296 | 70 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 32, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS_... | __global__ void __launch_bounds__(32) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 9) * 16384 + (int)threadIdx.x * 512 + ((int)blockIdx.x & 511);
int index1 = ((int)blockIdx.x & 511) * 512 + ((int)blockIdx.x >> 9) * 32 + (in... |
Reshape_Add_4296 | 19 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<97>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 4) * 65536 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 15) * 32 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 15) * 163... |
Reshape_Add_4296 | 24 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<97>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 1) * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 1) * 256 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 1) * 13107... |
Reshape_Add_4296 | 30 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 3) * 1024 + ((int)threadIdx.x >> 6) * 512 + ((int)blockIdx.x & 7) * 64 + ((int)threadIdx.x & 63);
int index1 = ((int)blockIdx.x & 7) * 32768... |
Reshape_Add_4296 | 0 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<13>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 4) * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 15) * 32 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 15) * 1638... |
Reshape_Add_4296 | 61 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<385>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 3) * 131072 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 7) * 64 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 7) * 3276... |
Reshape_Add_4296 | 87 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 8, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<11>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS_S... | __global__ void __launch_bounds__(8) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = (int)blockIdx.x * 8 + (int)threadIdx.x;
int index1 = ((int)blockIdx.x & 63) * 4096 + (int)threadIdx.x * 512 + ((int)blockIdx.x >> 6);
output0[index0] = input1[i... |
Reshape_Add_4296 | 6 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<25>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 2) * 4096 + ((int)threadIdx.x >> 4) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 15);
int index1 = ((int)blockIdx.x & 3) * 6553... |
Reshape_Add_4296 | 36 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<4>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 7) * 16384 + ((int)threadIdx.x >> 2) * 512 + ((int)blockIdx.x & 127) * 4 + ((int)threadIdx.x & 3);
int index1 = ((int)blockIdx.x & 127) * 20... |
Reshape_Add_4296 | 25 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<13>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 2) * 2048 + ((int)threadIdx.x >> 5) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 31);
int index1 = ((int)blockIdx.x & 3) * 6553... |
Reshape_Add_4296 | 37 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<193>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296P... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 1) * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 1) * 256 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 1) * 1310... |
Reshape_Add_4296 | 3 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<7>;
.reg .b32 %r<20>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296PfS... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 6) * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 63) * 8 + ((int)threadIdx.x & 7);
int index1 = ((int)blockIdx.x & 63) * 4096... |
Reshape_Add_4296 | 35 | null | .visible .entry _Z16Reshape_Add_4296PfS_S_(
.param .u64 _Z16Reshape_Add_4296PfS_S__param_0,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_1,
.param .u64 _Z16Reshape_Add_4296PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<25>;
.reg .b32 %r<13>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z16Reshape_Add_4296Pf... | __global__ void __launch_bounds__(128) Reshape_Add_4296(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = (int)blockIdx.x * 1024 + ((int)threadIdx.x >> 6) * 512 + ((int)threadIdx.x & 63);
int index1 = ((int)threadIdx.x & 63) * 512 + (int)blockIdx.x * 2 + ((int)threa... |
Sum_Broadcast_Divide_4492-Reshape_Reshape_Broadcast_Subtract_Multiply_4383 | 2 | null | .visible .entry _Z6Group5PfS_S_S_S_(
.param .u64 _Z6Group5PfS_S_S_S__param_0,
.param .u64 _Z6Group5PfS_S_S_S__param_1,
.param .u64 _Z6Group5PfS_S_S_S__param_2,
.param .u64 _Z6Group5PfS_S_S_S__param_3,
.param .u64 _Z6Group5PfS_S_S_S__param_4
)
.maxntid 128, 1, 1
{
.reg .pred %p<10>;
.reg .f32 %f<96>;
.reg .b32... | __device__ void Group5_0_Sum_Broadcast_Divide_4492(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
int __flatten_tid = threadIdx.x;
const dim3 threadIdx(__flatten_tid % 32, __flatten_tid / 32, 0);
float normal_reduce_temp0[1];
float* input1_shared = (float*)(... |
Sum_Broadcast_Divide_4492-Reshape_Reshape_Broadcast_Subtract_Multiply_4383 | 0 | null | .visible .entry _Z6Group5PfS_S_S_S_(
.param .u64 _Z6Group5PfS_S_S_S__param_0,
.param .u64 _Z6Group5PfS_S_S_S__param_1,
.param .u64 _Z6Group5PfS_S_S_S__param_2,
.param .u64 _Z6Group5PfS_S_S_S__param_3,
.param .u64 _Z6Group5PfS_S_S_S__param_4
)
.maxntid 128, 1, 1
{
.reg .pred %p<4>;
.reg .f32 %f<46>;
.reg .b32 ... | __device__ void Group5_0_Sum_Broadcast_Divide_4492(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float normal_reduce_temp0[1];
float* input1_shared = (float*)(shared+0);
__shared__ float red_buf0[128];
normal_reduce_temp0[0] = 0.000000e+00f;
*(float4*)(in... |
Sum_Broadcast_Divide_4492-Reshape_Reshape_Broadcast_Subtract_Multiply_4383 | 1 | null | .visible .entry _Z6Group5PfS_S_S_S_(
.param .u64 _Z6Group5PfS_S_S_S__param_0,
.param .u64 _Z6Group5PfS_S_S_S__param_1,
.param .u64 _Z6Group5PfS_S_S_S__param_2,
.param .u64 _Z6Group5PfS_S_S_S__param_3,
.param .u64 _Z6Group5PfS_S_S_S__param_4
)
.maxntid 128, 1, 1
{
.reg .pred %p<6>;
.reg .f32 %f<64>;
.reg .b32 ... | __device__ void Group5_0_Sum_Broadcast_Divide_4492(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
int __flatten_tid = threadIdx.x;
const dim3 threadIdx(__flatten_tid % 64, __flatten_tid / 64, 0);
float normal_reduce_temp0[1];
float* input1_shared = (float*)(... |
DotSplitK_3974-Broadcast_Add_Broadcast_Multiply_Sum_4209 | 15 | null | .visible .entry _Z7Group22PfS_S_S_S_(
.param .u64 _Z7Group22PfS_S_S_S__param_0,
.param .u64 _Z7Group22PfS_S_S_S__param_1,
.param .u64 _Z7Group22PfS_S_S_S__param_2,
.param .u64 _Z7Group22PfS_S_S_S__param_3,
.param .u64 _Z7Group22PfS_S_S_S__param_4
)
.maxntid 32, 1, 1
{
.reg .pred %p<11>;
.reg .b16 %rs<4>;
.reg... | __device__ void Group22_0_DotSplitK_3974(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float output0_local[1];
float* input0_shared = (float*)(shared+0);
float* input1_shared = (float*)(shared+8192);
output0_local[0] = 0.000000e+00f;
for (int ax0_ax1_fuse... |
DotSplitK_3974-Broadcast_Add_Broadcast_Multiply_Sum_4209 | 3 | null | .visible .entry _Z7Group22PfS_S_S_S_(
.param .u64 _Z7Group22PfS_S_S_S__param_0,
.param .u64 _Z7Group22PfS_S_S_S__param_1,
.param .u64 _Z7Group22PfS_S_S_S__param_2,
.param .u64 _Z7Group22PfS_S_S_S__param_3,
.param .u64 _Z7Group22PfS_S_S_S__param_4
)
.maxntid 128, 1, 1
{
.reg .pred %p<8>;
.reg .f32 %f<60>;
.reg... | __device__ void Group22_0_DotSplitK_3974(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float output0_local[2];
float* input0_shared = (float*)(shared+0);
float* input1_shared = (float*)(shared+16384);
output0_local[0] = 0.000000e+00f;
output0_local[1] = 0... |
DotSplitK_3974-Broadcast_Add_Broadcast_Multiply_Sum_4209 | 16 | null | .visible .entry _Z7Group22PfS_S_S_S_(
.param .u64 _Z7Group22PfS_S_S_S__param_0,
.param .u64 _Z7Group22PfS_S_S_S__param_1,
.param .u64 _Z7Group22PfS_S_S_S__param_2,
.param .u64 _Z7Group22PfS_S_S_S__param_3,
.param .u64 _Z7Group22PfS_S_S_S__param_4
)
.maxntid 32, 1, 1
{
.reg .pred %p<10>;
.reg .f32 %f<81>;
.reg... | __device__ void Group22_0_DotSplitK_3974(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float output0_local[1];
float* input0_shared = (float*)(shared+0);
float* input1_shared = (float*)(shared+16384);
output0_local[0] = 0.000000e+00f;
for (int ax0_ax1_fus... |
Not_564 | 0 | null | .visible .entry _Z7Not_564PaS_(
.param .u64 _Z7Not_564PaS__param_0,
.param .u64 _Z7Not_564PaS__param_1
)
.maxntid 128, 1, 1
{
.reg .pred %p<3>;
.reg .b16 %rs<7>;
.reg .b32 %r<5>;
.reg .b64 %rd<8>;
ld.param.u64 %rd1, [_Z7Not_564PaS__param_0];
ld.param.u64 %rd2, [_Z7Not_564PaS__param_1];
cvta.to.global.u... | __global__ void __launch_bounds__(128) Not_564(int8_t* __restrict__ input0, int8_t* __restrict__ output0) {
int index0 = (int)blockIdx.x * 256 + (int)threadIdx.x;
output0[index0] = signed char((int)input0[index0] == 0);
output0[index0 + 128] = signed char((int)input0[index0 + 128] == 0);
}
|
Reshape_Multiply_4315 | 10 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<33>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = (int)blockIdx.x * 2048 + ((int)threadIdx.x >> 5) * 512 + ((int)threadIdx.x & 31);
output_proxy[index0] = input0[index0];
output_proxy[index0... |
Reshape_Multiply_4315 | 40 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<129>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = (int)blockIdx.x * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)threadIdx.x & 7);
output_proxy[index0] = input0[index0];
output_proxy[index0 ... |
Reshape_Multiply_4315 | 6 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<17>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 2) * 4096 + ((int)threadIdx.x >> 4) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 15);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 23 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<65>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 5) * 131072 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 31) * 16 + ((int)threadIdx.x & 7);
output_proxy[index0] =... |
Reshape_Multiply_4315 | 32 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<65>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 3) * 32768 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 7) * 64 + ((int)threadIdx.x & 7);
output_proxy[index0] = i... |
Reshape_Multiply_4315 | 29 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<33>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 3) * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 7) * 64 + ((int)threadIdx.x & 7);
output_proxy[index0] = i... |
Reshape_Multiply_4315 | 1 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<9>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 2) * 2048 + ((int)threadIdx.x >> 5) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 31);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 31 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<65>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 4) * 65536 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 15) * 32 + ((int)threadIdx.x & 7);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 54 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<33>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)threadIdx.x >> 2) * 512 + (int)blockIdx.x * 4 + ((int)threadIdx.x & 3);
output_proxy[index0] = input0[index0];
output_proxy[index0 + 1... |
Reshape_Multiply_4315 | 33 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<65>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)threadIdx.x >> 3) * 512 + (int)blockIdx.x * 8 + ((int)threadIdx.x & 7);
output_proxy[index0] = input0[index0];
output_proxy[index0 + 8... |
Reshape_Multiply_4315 | 44 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<129>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 4) * 131072 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 15) * 32 + ((int)threadIdx.x & 7);
output_proxy[index0] =... |
Reshape_Multiply_4315 | 5 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<17>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 1) * 2048 + ((int)threadIdx.x >> 5) * 512 + ((int)blockIdx.x & 1) * 256 + ((int)threadIdx.x & 31);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 74 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 32, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z21R... | __global__ void __launch_bounds__(32) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 9) * 16384 + (int)threadIdx.x * 512 + ((int)blockIdx.x & 511);
output_proxy[index0] = input0[index0];
output_proxy_1[inde... |
Reshape_Multiply_4315 | 13 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<9>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 4) * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 15) * 32 + ((int)threadIdx.x & 7);
output_proxy[index0] = i... |
Reshape_Multiply_4315 | 12 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<9>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 5) * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 31) * 16 + ((int)threadIdx.x & 7);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 24 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<33>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 2) * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 7);
output_proxy[index0] = i... |
Reshape_Multiply_4315 | 93 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 1, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<2>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z21Re... | __global__ void __launch_bounds__(1) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = (int)blockIdx.x;
output_proxy[index0] = input0[index0];
output_proxy_1[index0] = input0[index0] * input0[index0];
}
|
Reshape_Multiply_4315 | 47 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 64, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z21... | __global__ void __launch_bounds__(64) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 7) * 8192 + ((int)threadIdx.x >> 2) * 512 + ((int)blockIdx.x & 127) * 4 + ((int)threadIdx.x & 3);
output_proxy[index0] = in... |
Reshape_Multiply_4315 | 35 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 6) * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 63) * 8 + ((int)threadIdx.x & 7);
output_proxy[index0] = in... |
Reshape_Multiply_4315 | 87 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 8, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z21Re... | __global__ void __launch_bounds__(8) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 9) * 4096 + (int)threadIdx.x * 512 + ((int)blockIdx.x & 511);
output_proxy[index0] = input0[index0];
output_proxy_1[index0... |
Reshape_Multiply_4315 | 82 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 16, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z21R... | __global__ void __launch_bounds__(16) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 9) * 8192 + (int)threadIdx.x * 512 + ((int)blockIdx.x & 511);
output_proxy[index0] = input0[index0];
output_proxy_1[index... |
Reshape_Multiply_4315 | 26 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<33>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 5) * 65536 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 31) * 16 + ((int)threadIdx.x & 7);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 42 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<129>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 2) * 32768 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 7);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 7 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<9>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 3) * 4096 + ((int)threadIdx.x >> 4) * 512 + ((int)blockIdx.x & 7) * 64 + ((int)threadIdx.x & 15);
output_proxy[index0] = i... |
Reshape_Multiply_4315 | 8 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<5>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 2) * 1024 + ((int)threadIdx.x >> 6) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 63);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 30 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<65>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 1) * 8192 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 1) * 256 + ((int)threadIdx.x & 7);
output_proxy[index0] = i... |
Reshape_Multiply_4315 | 41 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<129>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 1) * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 1) * 256 + ((int)threadIdx.x & 7);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 27 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<33>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 6) * 131072 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 63) * 8 + ((int)threadIdx.x & 7);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 51 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<9>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 7) * 65536 + ((int)threadIdx.x >> 2) * 512 + ((int)blockIdx.x & 127) * 4 + ((int)threadIdx.x & 3);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 25 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<33>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 4) * 32768 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 15) * 32 + ((int)threadIdx.x & 7);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 50 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 64, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z21... | __global__ void __launch_bounds__(64) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 4) * 1024 + ((int)threadIdx.x >> 5) * 512 + ((int)blockIdx.x & 15) * 32 + ((int)threadIdx.x & 31);
output_proxy[index0] = i... |
Reshape_Multiply_4315 | 36 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<5>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z21... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = (int)blockIdx.x * 128 + (int)threadIdx.x;
output_proxy[index0] = input0[index0];
output_proxy_1[index0] = input0[index0] * input0[index0];
}... |
Reshape_Multiply_4315 | 22 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<65>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 2) * 16384 + ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x & 3) * 128 + ((int)threadIdx.x & 7);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 15 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<33>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 1) * 4096 + ((int)threadIdx.x >> 4) * 512 + ((int)blockIdx.x & 1) * 256 + ((int)threadIdx.x & 15);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 69 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<17>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)threadIdx.x >> 1) * 512 + (int)blockIdx.x * 2 + ((int)threadIdx.x & 1);
output_proxy[index0] = input0[index0];
output_proxy[index0 + 3... |
Reshape_Multiply_4315 | 39 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<129>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)threadIdx.x >> 3) * 512 + (int)blockIdx.x * 16 + ((int)threadIdx.x & 7);
output_proxy[index0] = input0[index0];
output_proxy[index0 + ... |
Reshape_Multiply_4315 | 38 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 5) * 4096 + ((int)threadIdx.x >> 4) * 512 + ((int)blockIdx.x & 31) * 16 + ((int)threadIdx.x & 15);
output_proxy[index0] = ... |
Reshape_Multiply_4315 | 68 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 32, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z21... | __global__ void __launch_bounds__(32) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 8) * 8192 + ((int)threadIdx.x >> 1) * 512 + ((int)blockIdx.x & 255) * 2 + ((int)threadIdx.x & 1);
output_proxy[index0] = in... |
Reshape_Multiply_4315 | 60 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 64, 1, 1
{
.reg .f32 %f<3>;
.reg .b32 %r<12>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z21... | __global__ void __launch_bounds__(64) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = ((int)blockIdx.x >> 8) * 16384 + ((int)threadIdx.x >> 1) * 512 + ((int)blockIdx.x & 255) * 2 + ((int)threadIdx.x & 1);
output_proxy[index0] = i... |
Reshape_Multiply_4315 | 0 | null | .visible .entry _Z21Reshape_Multiply_4315PfS_S_(
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_0,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_1,
.param .u64 _Z21Reshape_Multiply_4315PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<17>;
.reg .b32 %r<8>;
.reg .b64 %rd<11>;
ld.param.u64 %rd1, [_Z2... | __global__ void __launch_bounds__(128) Reshape_Multiply_4315(float* __restrict__ input0, float* __restrict__ output_proxy, float* __restrict__ output_proxy_1) {
int index0 = (int)blockIdx.x * 1024 + ((int)threadIdx.x >> 6) * 512 + ((int)threadIdx.x & 63);
output_proxy[index0] = input0[index0];
output_proxy[index0... |
Reshape_Reshape_Broadcast_Reshape_Broadcast_Select_Broadcast_Multiply_4502 | 5 | null | .visible .entry _Z7Group10PfS_PaS_S_(
.param .u64 _Z7Group10PfS_PaS_S__param_0,
.param .u64 _Z7Group10PfS_PaS_S__param_1,
.param .u64 _Z7Group10PfS_PaS_S__param_2,
.param .u64 _Z7Group10PfS_PaS_S__param_3,
.param .u64 _Z7Group10PfS_PaS_S__param_4
)
.maxntid 128, 1, 1
{
.reg .pred %p<4>;
.reg .b16 %rs<4>;
.reg... | __global__ void __launch_bounds__(128) Group10(float* __restrict__ input0, float* __restrict__ input1, int8_t* __restrict__ input2, float* __restrict__ input3, float* __restrict__ output0) {
__shared__ float input1_shared[8];
__shared__ float input0_shared[8];
if ((int)threadIdx.x < 1) {
int index0 = (int)thr... |
Reshape_Reshape_Broadcast_Reshape_Broadcast_Select_Broadcast_Multiply_4502 | 39 | null | .visible .entry _Z7Group10PfS_PaS_S_(
.param .u64 _Z7Group10PfS_PaS_S__param_0,
.param .u64 _Z7Group10PfS_PaS_S__param_1,
.param .u64 _Z7Group10PfS_PaS_S__param_2,
.param .u64 _Z7Group10PfS_PaS_S__param_3,
.param .u64 _Z7Group10PfS_PaS_S__param_4
)
.maxntid 1, 1, 1
{
.reg .pred %p<2>;
.reg .b16 %rs<3>;
.reg .... | __global__ void __launch_bounds__(1) Group10(float* __restrict__ input0, float* __restrict__ input1, int8_t* __restrict__ input2, float* __restrict__ input3, float* __restrict__ output0) {
int index0 = (int)blockIdx.x;
output0[index0] = (((int)input2[index0 & 511]) == 0 ? input3[index0] : input1[0]) * input0[0];
}
... |
Reshape_Reshape_Broadcast_Reshape_Broadcast_Select_Broadcast_Multiply_4502 | 2 | null | .visible .entry _Z7Group10PfS_PaS_S_(
.param .u64 _Z7Group10PfS_PaS_S__param_0,
.param .u64 _Z7Group10PfS_PaS_S__param_1,
.param .u64 _Z7Group10PfS_PaS_S__param_2,
.param .u64 _Z7Group10PfS_PaS_S__param_3,
.param .u64 _Z7Group10PfS_PaS_S__param_4
)
.maxntid 32, 1, 1
{
.reg .pred %p<3>;
.reg .b16 %rs<3>;
.reg ... | __global__ void __launch_bounds__(32) Group10(float* __restrict__ input0, float* __restrict__ input1, int8_t* __restrict__ input2, float* __restrict__ input3, float* __restrict__ output0) {
__shared__ float input1_shared[8];
__shared__ float input0_shared[8];
if ((int)threadIdx.x < 1) {
int index0 = (int)thre... |
Reshape_Reshape_Broadcast_Reshape_Broadcast_Select_Broadcast_Multiply_4502 | 29 | null | .visible .entry _Z7Group10PfS_PaS_S_(
.param .u64 _Z7Group10PfS_PaS_S__param_0,
.param .u64 _Z7Group10PfS_PaS_S__param_1,
.param .u64 _Z7Group10PfS_PaS_S__param_2,
.param .u64 _Z7Group10PfS_PaS_S__param_3,
.param .u64 _Z7Group10PfS_PaS_S__param_4
)
.maxntid 128, 1, 1
{
.reg .pred %p<34>;
.reg .b16 %rs<37>;
.r... | __global__ void __launch_bounds__(128) Group10(float* __restrict__ input0, float* __restrict__ input1, int8_t* __restrict__ input2, float* __restrict__ input3, float* __restrict__ output0) {
__shared__ float input1_shared[8];
__shared__ float input0_shared[8];
if ((int)threadIdx.x < 1) {
int index0 = (int)thr... |
Reshape_Reshape_Broadcast_Reshape_Broadcast_Select_Broadcast_Multiply_4502 | 0 | null | .visible .entry _Z7Group10PfS_PaS_S_(
.param .u64 _Z7Group10PfS_PaS_S__param_0,
.param .u64 _Z7Group10PfS_PaS_S__param_1,
.param .u64 _Z7Group10PfS_PaS_S__param_2,
.param .u64 _Z7Group10PfS_PaS_S__param_3,
.param .u64 _Z7Group10PfS_PaS_S__param_4
)
.maxntid 128, 1, 1
{
.reg .pred %p<10>;
.reg .b16 %rs<10>;
.r... | __global__ void __launch_bounds__(128) Group10(float* __restrict__ input0, float* __restrict__ input1, int8_t* __restrict__ input2, float* __restrict__ input3, float* __restrict__ output0) {
__shared__ float input1_shared[8];
__shared__ float input0_shared[8];
if ((int)threadIdx.x < 1) {
int index0 = (int)thr... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.