mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-07 12:21:27 +01:00
246f5c412e
Summary:
This reverts commit d73c830e23.
We have observed significant perf drop when training ResNext101 with multiple amd GPUs:
Before:
https://ci.pytorch.org/jenkins/job/caffe2-builds/job/py2-clang7-rocmdeb-ubuntu16.04-bench/1636/console
2 GPUs ResNext training got 150\~160 imgs/sec
4 GPUs ResNext training got 270\~280 imgs/sec
After:
https://ci.pytorch.org/jenkins/job/caffe2-builds/job/py2-clang7-rocmdeb-ubuntu16.04-bench/1637/console
Both 2 and 4 GPUs ResNext training drop to 110\~120 imgs/sec
Similar perf drop are seen on ResNet50 training jobs as well.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18680
Differential Revision: D14702941
Pulled By: bddppq
fbshipit-source-id: 828141805afc23f25c08d4a2eb6d4b99f817c128
52 lines
1.4 KiB
C++
52 lines
1.4 KiB
C++
#ifndef CAFFE2_OPERATORS_ENSURE_CPU_OUTPUT_OP_H_
|
|
#define CAFFE2_OPERATORS_ENSURE_CPU_OUTPUT_OP_H_
|
|
|
|
#include "caffe2/core/context.h"
|
|
#include "caffe2/core/operator.h"
|
|
#include "caffe2/utils/math.h"
|
|
|
|
namespace caffe2 {
|
|
|
|
template <class Context>
|
|
class EnsureCPUOutputOp : public Operator<Context> {
|
|
public:
|
|
USE_OPERATOR_CONTEXT_FUNCTIONS;
|
|
template <class... Args>
|
|
explicit EnsureCPUOutputOp(Args&&... args)
|
|
: Operator<Context>(std::forward<Args>(args)...) {}
|
|
|
|
bool RunOnDevice() override {
|
|
if (this->InputIsTensorType(0, CPU)) {
|
|
return CopyWithContext<CPUContext>();
|
|
} else if (this->InputIsTensorType(0, Context::GetDeviceType())) {
|
|
// CUDA Context will go this branch
|
|
return CopyWithContext<Context>();
|
|
} else {
|
|
CAFFE_THROW(
|
|
"Unexpected Input Blob: ",
|
|
OperatorBase::Inputs().at(0)->meta().name());
|
|
}
|
|
return true;
|
|
}
|
|
|
|
private:
|
|
template <class InputContext>
|
|
bool CopyWithContext() {
|
|
// Output is always on CPU
|
|
auto* output = this->template Output<Tensor>(0, CPU);
|
|
auto& input = this->template Input<Tensor>(0, InputContext::GetDeviceType());
|
|
output->ResizeLike(input);
|
|
context_.CopyItemsToCPU(
|
|
input.dtype(),
|
|
input.numel(),
|
|
input.raw_data(),
|
|
output->raw_mutable_data(input.dtype()));
|
|
context_.FinishDeviceComputation();
|
|
return true;
|
|
}
|
|
};
|
|
|
|
} // namespace caffe2
|
|
|
|
#endif // CAFFE2_OPERATORS_ENSURE_CPU_OUTPUT_OP_H_
|