Skip to content

optimize fused_seqpool_cvm ops host time 2. #56

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Feb 29, 2024
Merged

optimize fused_seqpool_cvm ops host time 2. #56

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 2 additions & 3 deletions paddle/fluid/operators/fused/fused_seqpool_cvm_op_xpu.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,7 +131,7 @@ class FusedSeqpoolCVMOpXPUKernel : public framework::OpKernel<T> {
} else {
cpu_y_addr_vec[i] = reinterpret_cast<T*>(out[i]->mutable_data<T>(place));
}
auto x_lod = ins[i]->lod()[0];
auto& x_lod = ins[i]->lod()[0];
#ifdef PADDLE_WITH_MKLML
#pragma omp parallel for
#endif
Expand Down Expand Up @@ -225,10 +225,9 @@ class FusedSeqpoolCVMGradOpXPUKernel : public framework::OpKernel<T> {
T* dx_data = dx->mutable_data<T>(place);
// T* dx_data = dx->mutable_data<T>(place);
T* dy_data = const_cast<T*>(dy->data<T>());
auto lod = dx->lod();
cpu_dx_list[k] = dx_data;
cpu_dy_list[k] = (const T*)dy_data;
auto lod_level_0 = dx->lod()[0];
auto& lod_level_0 = dx->lod()[0];
int lod_size = lod_level_0.size();
for (int i = 0; i < lod_size; i++) {
cpu_lodx[i + start_index] = lod_level_0[i];
Expand Down
5 changes: 2 additions & 3 deletions paddle/fluid/operators/fused/fused_seqpool_cvm_with_conv_op_xpu.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -103,7 +103,7 @@ class FusedSeqpoolCVMWithConvOpXPUKernel : public framework::OpKernel<T> {
} else {
cpu_y_addr_vec[i] = reinterpret_cast<T*>(out[i]->mutable_data<T>(place));
}
auto x_lod = ins[i]->lod()[0];
auto& x_lod = ins[i]->lod()[0];
#ifdef PADDLE_WITH_MKLML
#pragma omp parallel for
#endif
Expand Down Expand Up @@ -193,10 +193,9 @@ class FusedSeqpoolCVMWithConvGradOpXPUKernel : public framework::OpKernel<T> {
}
T* dx_data = dx->mutable_data<T>(place);
T* dy_data = const_cast<T*>(dy->data<T>());
auto lod = dx->lod();
cpu_dx_list[k] = dx_data;
cpu_dy_list[k] = (const T*)dy_data;
auto lod_level_0 = dx->lod()[0];
auto& lod_level_0 = dx->lod()[0];
int lod_size = lod_level_0.size();
for (int i = 0; i < lod_size; i++) {
cpu_lodx[i + start_index] = lod_level_0[i];
Expand Down
5 changes: 2 additions & 3 deletions paddle/fluid/operators/fused/fused_seqpool_cvm_with_diff_thres_op_xpu.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -120,7 +120,7 @@ class FusedSeqpoolCVMWithDiffThresOpXPUKernel : public framework::OpKernel<T> {
} else {
cpu_y_addr_vec[i] = reinterpret_cast<T*>(out[i]->mutable_data<T>(place));
}
auto x_lod = ins[i]->lod()[0];
auto& x_lod = ins[i]->lod()[0];
#ifdef PADDLE_WITH_MKLML
#pragma omp parallel for
#endif
Expand Down Expand Up @@ -213,10 +213,9 @@ class FusedSeqpoolCVMWithDiffThresGradOpXPUKernel : public framework::OpKernel<T
T* dx_data = dx->mutable_data<T>(place);
// T* dx_data = dx->mutable_data<T>(place);
T* dy_data = const_cast<T*>(dy->data<T>());
auto lod = dx->lod();
cpu_dx_list[k] = dx_data;
cpu_dy_list[k] = (const T*)dy_data;
auto lod_level_0 = dx->lod()[0];
auto& lod_level_0 = dx->lod()[0];
int lod_size = lod_level_0.size();
for (int i = 0; i < lod_size; i++) {
cpu_lodx[i + start_index] = lod_level_0[i];
Expand Down