feat: add HIP/ROCm support for AMD GPUs #552
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add support for building pyg-lib on ROCm (AMD GPUs) alongside existing CUDA support. This allows pyg-lib to work with PyTorch ROCm builds. CMakeLists.txt: - Add WITH_HIP option to enable HIP/ROCm builds - Detect ROCm path and find HIP, rocBLAS, hipBLAS packages - Set appropriate GPU architectures for AMD (gfx906, gfx90a, etc.) - Compile .cu files as HIP sources on ROCm - Exclude CUTLASS/cuCollections files that have HIP replacements setup.py: - Auto-detect torch.version.hip for ROCm PyTorch builds - Pass -DWITH_HIP=ON to CMake when building on ROCm New HIP implementations: - ops/hip/matmul_kernel_hip.cpp: rocBLAS-based grouped GEMM - classes/hip/hash_map_hip.cpp: Sorted array + binary search hashmap Modified CUDA kernels for HIP compatibility: - sampler/cuda/random_walk_kernel.cu: Add HIP header/API macros - ops/cuda/sampled_kernel.cu: Add HIP header/API macros - library.cpp: Use hip_runtime.h and HIP_VERSION on ROCm The simple CUDA kernels work on HIP with compatibility macros for getCurrentCUDAStream -> getCurrentHIPStream and C10_CUDA_KERNEL_LAUNCH_CHECK -> C10_HIP_KERNEL_LAUNCH_CHECK. Tested on ROCm 6.2/6.4 with PyTorch 2.6. Generated-by: Claude AI Signed-off-by: Luis Chamberlain <[email protected]>
for more information, see https://pre-commit.ci
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Add support for building pyg-lib on ROCm (AMD GPUs) alongside existing CUDA support. This allows pyg-lib to work with PyTorch ROCm builds.
CMakeLists.txt:
setup.py:
New HIP implementations:
Modified CUDA kernels for HIP compatibility:
The simple CUDA kernels work on HIP with compatibility macros for getCurrentCUDAStream -> getCurrentHIPStream and
C10_CUDA_KERNEL_LAUNCH_CHECK -> C10_HIP_KERNEL_LAUNCH_CHECK.
Tested on ROCm 6.2/6.4 with PyTorch 2.6.
Generated-by: Claude AI