mirror of
https://github.com/pytorch/pytorch.git
synced 2025-10-20 21:14:14 +08:00
27cc11226d
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/68365 title. broadcast fastpath has been running fine for the enabled ops for a while now, so make it the default for these ops. Test Plan: diff is a no-op, so sandcastle Differential Revision: D32107847 fbshipit-source-id: b239b127b219985bf7df6a0eea2d879b8e9c79a4
18 lines
394 B
C++
18 lines
394 B
C++
#include "caffe2/operators/elementwise_add_op.h"
|
|
|
|
#include "caffe2/core/context_gpu.h"
|
|
|
|
namespace caffe2 {
|
|
|
|
REGISTER_CUDA_OPERATOR(
|
|
Add,
|
|
BinaryElementwiseOp<NumericTypes, CUDAContext, AddFunctor<CUDAContext>>);
|
|
REGISTER_CUDA_OPERATOR(
|
|
AddGradient,
|
|
BinaryElementwiseGradientOp<
|
|
NumericTypes,
|
|
CUDAContext,
|
|
AddFunctor<CUDAContext>>);
|
|
|
|
} // namespace caffe2
|