Change ConvPoolOp<Context>::SetOutputSize to ConvPoolOp<Context>::GetOutputSize ...
authorJerry Zhang <jerryzh@fb.com>
Fri, 8 Mar 2019 02:31:33 +0000 (18:31 -0800)
committerFacebook Github Bot <facebook-github-bot@users.noreply.github.com>
Fri, 8 Mar 2019 02:38:53 +0000 (18:38 -0800)
commitac87488bd3dad62fb4d9d556a8e1ccc277f71aeb
tree724d737f2bab3ad7abf7a15a519626c21b3ae882
parentcc7aec12fd76678d5e702cbf49b5339b7125486f
Change ConvPoolOp<Context>::SetOutputSize to ConvPoolOp<Context>::GetOutputSize (#17764)

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17764

Original commit changeset: f1923fdca4a1

reverted int8 ops fixes the original runtime regression.
We'll ignore the memory regression since it is flaky, see D14228484

Reviewed By: dzhulgakov

Differential Revision: D13885233

fbshipit-source-id: ccbe4b94acb44b7b4cb3ae4d73e3f6091e1e1195
16 files changed:
caffe2/cuda_rtc/pool_op_rtc_gpu.cc
caffe2/mobile/contrib/ios/mpscnn/mpscnn.mm
caffe2/operators/conv_op_cudnn.cc
caffe2/operators/conv_pool_op_base.h
caffe2/operators/depthwise_3x3_conv_op_cudnn.cu
caffe2/operators/hip/conv_op_miopen.hip
caffe2/operators/hip/pool_op_miopen.hip
caffe2/operators/max_pool_with_index.cu
caffe2/operators/pad_op_gpu.cu
caffe2/operators/pool_op_cudnn.cc
caffe2/quantization/server/conv_dnnlowp_acc16_op.cc
caffe2/quantization/server/conv_dnnlowp_op.cc
caffe2/quantization/server/conv_pool_dnnlowp_op_base.h
caffe2/quantization/server/dnnlowp_op.h
caffe2/quantization/server/pool_dnnlowp_op.cc
caffe2/share/contrib/depthwise/depthwise3x3_conv_op.cc