Skip to content
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

Divide elementwise case from BroadcastKernel and refine transpose autotune #33051

Merged
merged 15 commits into from
Dec 14, 2022
Merged

Divide elementwise case from BroadcastKernel and refine transpose autotune #33051

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
15 commits
Select commit Hold shift + click to select a range
3f243ed
First Commit.
JamesLim-sy May 21, 2021
1595849
remove useless file
JamesLim-sy Nov 29, 2022
e20398e
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
JamesLim-sy Nov 30, 2022
32725fa
add some codes
JamesLim-sy Dec 1, 2022
ad757a9
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
JamesLim-sy Dec 4, 2022
1a9036e
add elementwise loader
JamesLim-sy Dec 5, 2022
ec53fbd
fix code styles
JamesLim-sy Dec 5, 2022
e106a63
merge with develop
JamesLim-sy Dec 5, 2022
7074f55
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
JamesLim-sy Dec 5, 2022
e94b9f2
add some changes both in elementwise and transpose
JamesLim-sy Dec 5, 2022
168842f
add init operation in broadcast kernel.
JamesLim-sy Dec 6, 2022
8cf2c83
Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into…
JamesLim-sy Dec 6, 2022
e72fea1
change codes according to pr suggestions about transpose file
JamesLim-sy Dec 6, 2022
0994db3
fix error for op-benchmark ci
JamesLim-sy Dec 7, 2022
0a70f21
fix according to ci
JamesLim-sy Dec 9, 2022
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
1 change: 0 additions & 1 deletion paddle/fluid/operators/mkldnn/transpose_mkldnn_op.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,7 +16,6 @@
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/fluid/memory/malloc.h"
#include "paddle/fluid/platform/mkldnn_reuse.h"
#include "paddle/phi/kernels/funcs/transpose_functor.h"

namespace paddle {
namespace operators {
Expand Down
1 change: 0 additions & 1 deletion paddle/fluid/operators/transpose_op.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,6 @@ limitations under the License. */
#include "paddle/fluid/platform/mkldnn_helper.h"
#endif
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/phi/kernels/funcs/transpose_functor.h"

namespace paddle {
namespace operators {
Expand Down
1 change: 0 additions & 1 deletion paddle/fluid/operators/transpose_op_mlu.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,6 @@ See the License for the specific language governing permissions and
limitations under the License. */

#include "paddle/fluid/operators/mlu/mlu_baseop.h"
#include "paddle/phi/kernels/funcs/transpose_functor.h"

namespace paddle {
namespace operators {
Expand Down
1 change: 0 additions & 1 deletion paddle/fluid/operators/unique_op.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,6 @@ limitations under the License. */
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/fluid/operators/math/concat_and_split.h"
#include "paddle/phi/kernels/funcs/math_function.h"
#include "paddle/phi/kernels/funcs/transpose_functor.h"

namespace paddle {
namespace operators {
Expand Down
2 changes: 1 addition & 1 deletion paddle/phi/kernels/autotune/auto_tune_base.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -123,7 +123,7 @@ class AutoTuneBase {
float RunAndMeasureKernel(const Context& ctx, const int idx, Args&&... args) {
// Regard 1st run as warmup, judge the compare result by the time cost
// of rest cycles.
constexpr int repeats = 4;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

6这个数字总觉得有些奇怪，是不是可以调整到10？

constexpr int repeats = 6;
phi::GpuTimer timer;
float time_cost = 0;
const auto& stream = ctx.stream();
Expand Down
Loading