review.tizen.org Git - platform/upstream/pytorch.git/commit

projects / platform / upstream / pytorch.git / commit

author	Edward Yang <ezyang@fb.com>
	Wed, 12 Dec 2018 19:19:03 +0000 (11:19 -0800)
committer	Facebook Github Bot <facebook-github-bot@users.noreply.github.com>
	Wed, 12 Dec 2018 19:24:26 +0000 (11:24 -0800)
commit	2d485ffb17ebdec54b399df591e1da031e101d46
tree	e9e9208777a42581b3e31fe5f1a22702fc9ff5f1	tree \| snapshot
parent	9943cf2378a9919c576420a743ff96306512efe1	commit \| diff

Move CUDAGuard, CUDAStream and CUDAGuardImpl to c10/cuda (#14248)

Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/14248

This diff also introduces a horrifying hack to override CUDA's DeviceGuardImpl
with a HIPGuardImplMasqueradingAsCUDA, to accommodate PyTorch's current
behavior of pretending CUDA is HIP when you build with ROCm enabled.

Reviewed By: bddppq

Differential Revision: D13145293

fbshipit-source-id: ee0e207b6fd132f0d435512957424a002d588f02

47 files changed:

aten/src/ATen/CMakeLists.txt		diff \| blob \| history
aten/src/ATen/cuda/CUDAContext.cpp		diff \| blob \| history
aten/src/ATen/cuda/CUDAContext.h		diff \| blob \| history
aten/src/ATen/cuda/CUDAEvent.h		diff \| blob \| history
aten/src/ATen/cuda/CUDAMultiStreamGuard.h		diff \| blob \| history
aten/src/ATen/hip/impl/HIPGuardImplMasqueradingAsCUDA.cpp	[new file with mode: 0644]	blob
aten/src/ATen/hip/impl/HIPGuardImplMasqueradingAsCUDA.h	[new file with mode: 0644]	blob
aten/src/ATen/miopen/Utils.h		diff \| blob \| history
aten/src/ATen/native/DispatchStub.h		diff \| blob \| history
aten/src/ATen/native/cuda/Copy.cu		diff \| blob \| history
aten/src/ATen/native/cuda/Resize.cuh		diff \| blob \| history
aten/src/ATen/test/cuda_stream_test.cpp		diff \| blob \| history
aten/src/THC/THCCachingAllocator.cpp		diff \| blob \| history
aten/src/THC/THCCachingAllocator.h		diff \| blob \| history
aten/src/THC/THCCachingHostAllocator.h		diff \| blob \| history
aten/src/THC/THCGeneral.cpp		diff \| blob \| history
aten/src/THC/THCGeneral.h.in		diff \| blob \| history
aten/src/THC/THCStream.cpp		diff \| blob \| history
c10/cuda/CMakeLists.txt		diff \| blob \| history
c10/cuda/CUDAGuard.h	[moved from aten/src/ATen/cuda/CUDAGuard.h with 96% similarity]	diff \| blob \| history
c10/cuda/CUDAStream.cpp	[moved from aten/src/ATen/cuda/CUDAStream.cpp with 98% similarity]	diff \| blob \| history
c10/cuda/CUDAStream.h	[moved from aten/src/ATen/cuda/CUDAStream.h with 97% similarity]	diff \| blob \| history
c10/cuda/impl/CUDAGuardImpl.cpp	[moved from aten/src/ATen/cuda/detail/CUDAGuardImpl.cpp with 58% similarity]	diff \| blob \| history
c10/cuda/impl/CUDAGuardImpl.h	[moved from aten/src/ATen/cuda/detail/CUDAGuardImpl.h with 73% similarity]	diff \| blob \| history
c10/impl/InlineDeviceGuard.h		diff \| blob \| history
docs/cpp/source/Doxyfile		diff \| blob \| history
tools/amd_build/pyHIPIFY/cuda_to_hip_mappings.py		diff \| blob \| history
tools/amd_build/pyHIPIFY/hipify_python.py		diff \| blob \| history
tools/cwrap/plugins/AutoGPU.py		diff \| blob \| history
tools/cwrap/plugins/NNExtension.py		diff \| blob \| history
torch/csrc/autograd/engine.cpp		diff \| blob \| history
torch/csrc/autograd/profiler.cpp		diff \| blob \| history
torch/csrc/cuda/Stream.cpp		diff \| blob \| history
torch/csrc/cuda/comm.cpp		diff \| blob \| history
torch/csrc/cuda/nccl.cpp		diff \| blob \| history
torch/csrc/cuda/python_nccl.cpp		diff \| blob \| history
torch/csrc/distributed/c10d/ddp.cpp		diff \| blob \| history
torch/csrc/generic/StorageSharing.cpp		diff \| blob \| history
torch/csrc/jit/fuser/cuda/fused_kernel.cpp		diff \| blob \| history
torch/csrc/utils.h		diff \| blob \| history
torch/lib/THD/base/data_channels/DataChannelNccl.cpp		diff \| blob \| history
torch/lib/c10d/ProcessGroupGloo.cpp		diff \| blob \| history
torch/lib/c10d/ProcessGroupGloo.hpp		diff \| blob \| history
torch/lib/c10d/ProcessGroupNCCL.cpp		diff \| blob \| history
torch/lib/c10d/test/CUDATest.hpp		diff \| blob \| history
torch/lib/c10d/test/ProcessGroupGlooAsyncTest.cpp		diff \| blob \| history
torch/lib/c10d/test/ProcessGroupNCCLTest.cpp		diff \| blob \| history

Domain: Machine Learning / ML Framework;

RSS Atom