[mlir] Add NVVM to CUBIN conversion to mlir-opt
authorChristian Sigg <csigg@google.com>
Thu, 11 Mar 2021 08:56:00 +0000 (09:56 +0100)
committerChristian Sigg <csigg@google.com>
Thu, 11 Mar 2021 09:07:11 +0000 (10:07 +0100)
commit2224221fb3fa9738bd84221ed048247089187fce
treedbe83433bd5d2e9a0845fd56c7e89eb39480355a
parent80d1f657a157b2e1558ecfc161bad062592ba5bf
[mlir] Add NVVM to CUBIN conversion to mlir-opt

If MLIR_CUDA_RUNNER_ENABLED, register a 'gpu-to-cubin' conversion pass to mlir-opt.

The next step is to switch CUDA integration tests from mlir-cuda-runner to mlir-opt + mlir-cpu-runner and remove mlir-cuda-runner.

Depends On D98279

Reviewed By: herhut, rriddle, mehdi_amini

Differential Revision: https://reviews.llvm.org/D98203
mlir/include/mlir/Dialect/GPU/Passes.h
mlir/include/mlir/InitAllPasses.h
mlir/lib/Conversion/GPUCommon/CMakeLists.txt
mlir/lib/Conversion/GPUCommon/ConvertKernelFuncToBlob.cpp
mlir/lib/Dialect/GPU/CMakeLists.txt
mlir/lib/Dialect/GPU/Transforms/SerializeToBlob.cpp
mlir/lib/Dialect/GPU/Transforms/SerializeToCubin.cpp [new file with mode: 0644]
mlir/test/Integration/GPU/CUDA/shuffle.mlir