file stringlengths 6 242 | name stringlengths 1 11 | kernel stringclasses 4
values | input stringlengths 531 14.6k | output stringlengths 176 17.2k |
|---|---|---|---|---|
merged_1 | 326 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 16, 1, 1
{
.reg .f32 %f<6>;
.reg .b32 %r<16>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z17SoftmaxBasic_391... | __global__ void __launch_bounds__(16) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)threadIdx.x >> 1) * 262144 + ((int)blockIdx.x >> 9) * 1024 + ((int)threadIdx.x & 1) * 512 + ((int)blockIdx.x & 511);
int index1 = ((int)threadIdx.x >> 1) ... |
merged_1 | 287 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 64, 1, 1
{
.reg .f32 %f<6>;
.reg .b32 %r<10>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z17SoftmaxBasic_391... | __global__ void __launch_bounds__(64) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 9) * 32768 + (int)threadIdx.x * 512 + ((int)blockIdx.x & 511);
int index1 = ((int)blockIdx.x >> 9) * 64 + (int)threadIdx.x;
output0[index0... |
merged_1 | 288 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 64, 1, 1
{
.reg .f32 %f<6>;
.reg .b32 %r<22>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z17SoftmaxBasic_391... | __global__ void __launch_bounds__(64) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 13) * 524288 + ((int)threadIdx.x >> 5) * 262144 + (((int)blockIdx.x & 8191) >> 9) * 16384 + ((int)threadIdx.x & 31) * 512 + ((int)blockIdx.x &... |
merged_1 | 343 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 4, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<7>;
.reg .b32 %r<26>;
.reg .b64 %rd<12>;
// demoted variable
.sh... | __global__ void __launch_bounds__(4) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 17) * 1024 + (int)threadIdx.x * 512 + (... |
merged_1 | 145 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<516>;
.reg .b32 %r<27>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[32];
if ((int)threadIdx.x < 32) {
int index0 = (int)threadIdx.x;
int index1 = ((int)threadIdx.x >> 2) * 512 + (int)blockIdx.x * 4 + (... |
merged_1 | 128 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<138>;
.reg .b32 %r<34>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[128];
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 7) * 1024 + ((int)threadIdx.x >> 6) * 512 + (((int)blockIdx.x & 127) ... |
merged_1 | 2 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<19>;
.reg .b32 %r<21>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 4) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 2) * 4 + (int)threadIdx.x;
input1... |
merged_1 | 106 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<132>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[32];
if ((int)threadIdx.x < 32) {
int index0 = (int)threadIdx.x;
int index1 = ((int)threadIdx.x >> 2) * 512 + ((int)blockIdx.x >> 2) ... |
merged_1 | 25 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<22>;
.reg .b32 %r<21>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[64];
if ((int)threadIdx.x < 64) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 6) * 64 + (int)threadIdx.x;
inp... |
merged_1 | 63 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<74>;
.reg .b32 %r<30>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b8... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[128];
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 9) * 2048 + ((int)threadIdx.x >> 5) * 512 + (((int)blockIdx.x & 511) ... |
merged_1 | 17 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<35>;
.reg .b32 %r<35>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 8) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 9) * 1024 + ((int)threadIdx.x >> 2) *... |
merged_1 | 161 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<329>;
.reg .b32 %r<31>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[1024];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 8; ++ax0_ax1_fused_outer_outer_outer) {
int index0... |
merged_1 | 27 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<19>;
.reg .b32 %r<25>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 2) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 10) * 1024 + (int)threadIdx.x * 512 +... |
merged_1 | 138 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<531>;
.reg .b32 %r<30>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[256];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 2; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 38 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<74>;
.reg .b32 %r<19>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b8... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[128];
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 5) * 128 + (int)threadIdx.x;
input1_shared[index0] = input1[index1]... |
merged_1 | 109 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<165>;
.reg .b32 %r<19>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[512];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 4; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 13 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<35>;
.reg .b32 %r<22>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 4) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 1) * 4 + (int)threadIdx.x;
input1... |
merged_1 | 291 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<6>;
.reg .b32 %r<22>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z17SoftmaxBasic_39... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 12) * 524288 + ((int)threadIdx.x >> 6) * 262144 + (((int)blockIdx.x & 4095) >> 9) * 32768 + ((int)threadIdx.x & 63) * 512 + ((int)blockIdx.x ... |
merged_1 | 304 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<21>;
.reg .b32 %r<23>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z17SoftmaxBasic_3... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 11) * 1048576 + ((int)threadIdx.x >> 5) * 262144 + (((int)blockIdx.x & 2047) >> 9) * 65536 + ((int)threadIdx.x & 31) * 512 + ((int)blockIdx.x... |
merged_1 | 108 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<293>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[512];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 4; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 39 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<35>;
.reg .b32 %r<34>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[16];
if ((int)threadIdx.x < 16) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 9) * 1024 + ((int)threadIdx.x >> 3)... |
merged_1 | 113 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<260>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[32];
if ((int)threadIdx.x < 32) {
int index0 = (int)threadIdx.x;
int index1 = ((int)threadIdx.x >> 2) * 512 + ((int)blockIdx.x >> 1) ... |
merged_1 | 159 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<549>;
.reg .b32 %r<19>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[512];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 4; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 220 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<43>;
.reg .b32 %r<34>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b8... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[256];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 2; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 194 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<22>;
.reg .b32 %r<35>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b8... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[128];
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 10) * 1024 + ((int)threadIdx.x >> 6) * 512 + (((int)blockIdx.x & 1023... |
merged_1 | 83 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<22>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[64];
if ((int)threadIdx.x < 64) {
int index0 = (int)threadIdx.x;
int index1 = ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x >> 6) ... |
merged_1 | 143 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<165>;
.reg .b32 %r<30>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[512];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 4; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 142 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<138>;
.reg .b32 %r<29>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[128];
int index0 = (int)threadIdx.x;
int index1 = ((int)threadIdx.x >> 4) * 512 + ((int)blockIdx.x >> 4) * 16 + ((int)threadIdx.x & 15);
... |
merged_1 | 207 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<43>;
.reg .b32 %r<19>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b8... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[256];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 2; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 311 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 32, 1, 1
{
.reg .f32 %f<6>;
.reg .b32 %r<16>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z17SoftmaxBasic_391... | __global__ void __launch_bounds__(32) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)threadIdx.x >> 2) * 262144 + ((int)blockIdx.x >> 9) * 2048 + ((int)threadIdx.x & 3) * 512 + ((int)blockIdx.x & 511);
int index1 = ((int)threadIdx.x >> 2) ... |
merged_1 | 22 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<67>;
.reg .b32 %r<24>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 4) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 9) * 2048 + (int)threadIdx.x * 512 + ... |
merged_1 | 149 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<585>;
.reg .b32 %r<31>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[1024];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 8; ++ax0_ax1_fused_outer_outer_outer) {
int index0... |
merged_1 | 235 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 64, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<7>;
.reg .b32 %r<35>;
.reg .b64 %rd<12>;
// demoted variable
.s... | __global__ void __launch_bounds__(64) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 4) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 13) * 1024 + ((int)threadIdx.x >> 1) *... |
merged_1 | 134 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<262>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[64];
if ((int)threadIdx.x < 64) {
int index0 = (int)threadIdx.x;
int index1 = ((int)threadIdx.x >> 3) * 512 + ((int)blockIdx.x >> 2) ... |
merged_1 | 43 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<42>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b8... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[128];
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 10) * 2048 + ((int)threadIdx.x >> 5) * 512 + (((int)blockIdx.x & 1023... |
merged_1 | 229 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .b16 %rs<3>;
.reg .f32 %f<73>;
.reg .b32 %r<24>;
.reg .b64 %rd<18>;
/... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[4096];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 32; ++ax0_ax1_fused_outer_outer_outer) {
int index... |
merged_1 | 187 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .b16 %rs<3>;
.reg .f32 %f<57>;
.reg .b32 %r<30>;
.reg .b64 %rd<18>;
/... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[2048];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 16; ++ax0_ax1_fused_outer_outer_outer) {
int index... |
merged_1 | 7 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<19>;
.reg .b32 %r<35>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 8) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 11) * 2048 + ((int)threadIdx.x >> 1) ... |
merged_1 | 60 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<67>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 4) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 8) * 1024 + ((int)threadIdx.x >> 1) *... |
merged_1 | 336 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 8, 1, 1
{
.reg .f32 %f<6>;
.reg .b32 %r<22>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z17SoftmaxBasic_3919... | __global__ void __launch_bounds__(8) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)blockIdx.x >> 16) * 524288 + ((int)threadIdx.x >> 2) * 262144 + (((int)blockIdx.x & 65535) >> 9) * 2048 + ((int)threadIdx.x & 3) * 512 + ((int)blockIdx.x & 5... |
merged_1 | 80 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<132>;
.reg .b32 %r<22>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[32];
if ((int)threadIdx.x < 32) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 2) * 32 + (int)threadIdx.x;
inp... |
merged_1 | 37 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<38>;
.reg .b32 %r<37>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[64];
if ((int)threadIdx.x < 64) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 10) * 2048 + ((int)threadIdx.x >> 4... |
merged_1 | 122 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<266>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[128];
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 7) * 2048 + ((int)threadIdx.x >> 5) * 512 + (((int)blockIdx.x & 127) ... |
merged_1 | 96 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<131>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 8) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 7) * 1024 + ((int)threadIdx.x >> 2) *... |
merged_1 | 24 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<70>;
.reg .b32 %r<35>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[64];
if ((int)threadIdx.x < 64) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 9) * 2048 + ((int)threadIdx.x >> 4)... |
merged_1 | 79 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<147>;
.reg .b32 %r<34>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[256];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 2; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 165 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<165>;
.reg .b32 %r<31>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[512];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 4; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 36 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<68>;
.reg .b32 %r<36>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[32];
if ((int)threadIdx.x < 32) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 8) * 1024 + ((int)threadIdx.x >> 4)... |
merged_1 | 167 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<329>;
.reg .b32 %r<26>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[1024];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 8; ++ax0_ax1_fused_outer_outer_outer) {
int index0... |
merged_1 | 297 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<11>;
.reg .b32 %r<16>;
.reg .b64 %rd<12>;
ld.param.u64 %rd1, [_Z17SoftmaxBasic_3... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
int index0 = ((int)threadIdx.x >> 4) * 262144 + ((int)blockIdx.x >> 9) * 16384 + ((int)threadIdx.x & 15) * 512 + ((int)blockIdx.x & 511);
int index1 = ((int)threadIdx.x >> ... |
merged_1 | 163 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<7>;
.reg .b32 %r<18>;
.reg .b64 %rd<12>;
// demoted variable
.... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 8) {
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 512 + ((int)blockIdx.x >> 5);
inpu... |
merged_1 | 95 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<131>;
.reg .b32 %r<17>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 8) {
int index0 = (int)threadIdx.x;
int index1 = (int)blockIdx.x * 8 + (int)threadIdx.x;
input1_shared... |
merged_1 | 127 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<293>;
.reg .b32 %r<29>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[512];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 4; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 114 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<275>;
.reg .b32 %r<29>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[256];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 2; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 120 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<531>;
.reg .b32 %r<27>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[256];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 2; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 35 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<68>;
.reg .b32 %r<22>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[32];
if ((int)threadIdx.x < 32) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 3) * 32 + (int)threadIdx.x;
inp... |
merged_1 | 184 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<57>;
.reg .b32 %r<31>;
.reg .b64 %rd<19>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[2048];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 16; ++ax0_ax1_fused_outer_outer_outer) {
int index... |
merged_1 | 4 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<36>;
.reg .b32 %r<22>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[32];
if ((int)threadIdx.x < 32) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 4) * 32 + (int)threadIdx.x;
inp... |
merged_1 | 151 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<260>;
.reg .b32 %r<37>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[32];
if ((int)threadIdx.x < 32) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 7) * 2048 + ((int)threadIdx.x >> 3)... |
merged_1 | 162 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<585>;
.reg .b32 %r<26>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[1024];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 8; ++ax0_ax1_fused_outer_outer_outer) {
int index0... |
merged_1 | 331 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 8, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<7>;
.reg .b32 %r<26>;
.reg .b64 %rd<12>;
// demoted variable
.sh... | __global__ void __launch_bounds__(8) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 4) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 17) * 2048 + (int)threadIdx.x * 512 + (... |
merged_1 | 9 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<19>;
.reg .b32 %r<11>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[8];
if ((int)threadIdx.x < 1) {
int index0 = (int)threadIdx.x;
int index1 = (int)threadIdx.x * 512 + (int)blockIdx.x;
input1_shar... |
merged_1 | 147 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<549>;
.reg .b32 %r<32>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[512];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 4; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 92 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<275>;
.reg .b32 %r<34>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[256];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 2; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 141 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<262>;
.reg .b32 %r<22>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[64];
if ((int)threadIdx.x < 64) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 2) * 64 + (int)threadIdx.x;
inp... |
merged_1 | 88 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<147>;
.reg .b32 %r<19>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[256];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 2; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 21 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<70>;
.reg .b32 %r<38>;
.reg .b64 %rd<12>;
// demoted variable
... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[64];
if ((int)threadIdx.x < 64) {
int index0 = (int)threadIdx.x;
int index1 = ((int)blockIdx.x >> 8) * 1024 + ((int)threadIdx.x >> 5)... |
merged_1 | 124 | null | .visible .entry _Z17SoftmaxBasic_3919PfS_S_(
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_0,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_1,
.param .u64 _Z17SoftmaxBasic_3919PfS_S__param_2
)
.maxntid 128, 1, 1
{
.reg .f32 %f<275>;
.reg .b32 %r<19>;
.reg .b64 %rd<12>;
// demoted variable
.shared .align 4 .b... | __global__ void __launch_bounds__(128) SoftmaxBasic_3919(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0) {
__shared__ float input1_shared[256];
for (int ax0_ax1_fused_outer_outer_outer = 0; ax0_ax1_fused_outer_outer_outer < 2; ++ax0_ax1_fused_outer_outer_outer) {
int index0 ... |
merged_1 | 37 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<3>;
.reg .f32 %f<291>;
.reg .b32 %r<114>;
.reg .b64 %rd<20>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[32];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 27 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<195>;
.reg .b32 %r<161>;
.reg .b64 %rd<16>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[16];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 22 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<195>;
.reg .b32 %r<161>;
.reg .b64 %rd<16>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[16];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 42 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<3>;
.reg .f32 %f<291>;
.reg .b32 %r<102>;
.reg .b64 %rd<20>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[32];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 21 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .f32 %f<195>;
.reg .b32 %r<154>;
.reg .b64 %rd<16>;
// demoted variable
.shared .alig... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[16];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 31 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<297>;
.reg .b32 %r<90>;
.reg .b64 %rd<20>;
// demoted vari... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[32];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[4];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[8] = 0.0000... |
merged_1 | 26 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<195>;
.reg .b32 %r<162>;
.reg .b64 %rd<16>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[16];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 14 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .f32 %f<99>;
.reg .b32 %r<90>;
.reg .b64 %rd<16>;
// demoted variable
.shared .align ... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[8];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.00000... |
merged_1 | 83 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 2, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<19>;
.reg .b32 %r<15>;
.reg .b64 %rd<16>;
// demoted variabl... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[1];
float* input1_shared = (float*)(shared+0);
float* input0_shared = (float*)(shared+32);
float input0_shared_local[1];
mediate1_local[0] ... |
merged_1 | 59 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<27>;
.reg .b32 %r<41>;
.reg .b64 %rd<16>;
// demoted varia... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[2];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.00000... |
merged_1 | 6 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .f32 %f<53>;
.reg .b32 %r<59>;
.reg .b64 %rd<16>;
// demoted variable
.shared .align ... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[4];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[2];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[2] = 0.00000... |
merged_1 | 52 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<99>;
.reg .b32 %r<89>;
.reg .b64 %rd<16>;
// demoted varia... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[8];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.00000... |
merged_1 | 69 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 32, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<19>;
.reg .b32 %r<20>;
.reg .b64 %rd<16>;
// demoted variab... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[1];
float* input1_shared = (float*)(shared+0);
float* input0_shared = (float*)(shared+512);
float input0_shared_local[1];
mediate1_local[0]... |
merged_1 | 40 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<195>;
.reg .b32 %r<162>;
.reg .b64 %rd<16>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[16];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 19 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<99>;
.reg .b32 %r<95>;
.reg .b64 %rd<16>;
// demoted varia... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[8];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.00000... |
merged_1 | 17 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .f32 %f<201>;
.reg .b32 %r<144>;
.reg .b64 %rd<16>;
// demoted variable
.shared .alig... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[16];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[4];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[4] = 0.0000... |
merged_1 | 41 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<3>;
.reg .f32 %f<291>;
.reg .b32 %r<114>;
.reg .b64 %rd<20>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[32];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 16 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<51>;
.reg .b32 %r<63>;
.reg .b64 %rd<16>;
// demoted varia... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[4];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.00000... |
merged_1 | 54 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<195>;
.reg .b32 %r<155>;
.reg .b64 %rd<16>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[16];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 35 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 64, 1, 1
{
.reg .f32 %f<14>;
.reg .b32 %r<25>;
.reg .b64 %rd<16>;
// demoted variable
.shared .align 1... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[1];
float* input1_shared = (float*)(shared+0);
mediate1_local[0] = 0.000000e+00f;
*(float4*)(input1_shared + (int)threadIdx.x * 4) = *(float4... |
merged_1 | 32 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<291>;
.reg .b32 %r<103>;
.reg .b64 %rd<20>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[32];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 71 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 16, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<19>;
.reg .b32 %r<20>;
.reg .b64 %rd<16>;
// demoted variab... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[1];
float* input1_shared = (float*)(shared+0);
float* input0_shared = (float*)(shared+256);
float input0_shared_local[1];
mediate1_local[0]... |
merged_1 | 34 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<3>;
.reg .f32 %f<291>;
.reg .b32 %r<114>;
.reg .b64 %rd<20>;
// demoted var... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[32];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[1];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[1] = 0.0000... |
merged_1 | 15 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .f32 %f<101>;
.reg .b32 %r<92>;
.reg .b64 %rd<16>;
// demoted variable
.shared .align... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[8];
float* input1_shared = (float*)(shared+0);
float input0_shared_local[2];
mediate1_local[0] = 0.000000e+00f;
mediate1_local[4] = 0.00000... |
merged_1 | 72 | null | .visible .entry _Z7Group14PfS_S_S_(
.param .u64 _Z7Group14PfS_S_S__param_0,
.param .u64 _Z7Group14PfS_S_S__param_1,
.param .u64 _Z7Group14PfS_S_S__param_2,
.param .u64 _Z7Group14PfS_S_S__param_3
)
.maxntid 128, 1, 1
{
.reg .pred %p<2>;
.reg .f32 %f<35>;
.reg .b32 %r<21>;
.reg .b64 %rd<16>;
// demoted varia... | __device__ void Group14_0_Broadcast_Add_Sum_4516(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ output0, char* shared) {
float mediate1_local[2];
float* input1_shared = (float*)(shared+0);
float* input0_shared = (float*)(shared+4096);
float input0_shared_local[1];
mediate1_local[0... |
merged_1 | 5 | null | .visible .entry _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S_(
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_0,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_1,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_... | __global__ void __launch_bounds__(128) DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[4];
__shared__ float input1_shared[632];
__shared__ float input2_shared[128];
m... |
merged_1 | 0 | null | .visible .entry _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S_(
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_0,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_1,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_... | __global__ void __launch_bounds__(128) DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[4];
__shared__ float input1_shared[512];
__shared__ float input2_shared[32];
__... |
merged_1 | 8 | null | .visible .entry _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S_(
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_0,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_1,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_... | __global__ void __launch_bounds__(128) DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[16];
__shared__ float input1_shared[2048];
__shared__ float input2_shared[128];
... |
merged_1 | 65 | null | .visible .entry _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S_(
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_0,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_1,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_... | __global__ void __launch_bounds__(124) DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float normal_reduce_temp0[1];
__shared__ float input1_shared[64];
__shared__ float input2_shared[64];
... |
merged_1 | 4 | null | .visible .entry _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S_(
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_0,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_1,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_... | __global__ void __launch_bounds__(128) DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[4];
__shared__ float input1_shared[576];
__shared__ float input2_shared[64];
me... |
merged_1 | 1 | null | .visible .entry _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S_(
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_0,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_1,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_... | __global__ void __launch_bounds__(128) DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[2];
__shared__ float input1_shared[288];
__shared__ float input2_shared[32];
__... |
merged_1 | 51 | null | .visible .entry _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S_(
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_0,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469PfS_S_S__param_1,
.param .u64 _Z57DepthwiseConv1dNative_Broadcast_... | __global__ void __launch_bounds__(32) DepthwiseConv1dNative_Broadcast_Add_Sigmoid_Multiply_4469(float* __restrict__ input0, float* __restrict__ input1, float* __restrict__ input2, float* __restrict__ output0) {
float mediate1_local[1];
__shared__ float input1_shared[992];
__shared__ float input2_shared[992];
me... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.