Provide default for gpuBinaryAnnotation so that we don't need to specify it in tests.
The annotation likely only needs to be target specific if we want to lower to e.g. both CUDA and ROCDL.
Paths
| Differential D98168
[mlir] Default for gpu-binary-annotation option. ClosedPublic Authored by csigg on Mar 8 2021, 4:19 AM.
Details Summary Provide default for gpuBinaryAnnotation so that we don't need to specify it in tests. The annotation likely only needs to be target specific if we want to lower to e.g. both CUDA and ROCDL.
Diff Detail
Event TimelineHerald added subscribers: dcaballe, cota, teijeong and 18 others. · View Herald TranscriptMar 8 2021, 4:19 AM Comment Actions Move option to GpuToLLVMConversionPass (from Base) to remove the duplicate forward declaration of getDefaultGpuBinaryAnnotation(). This revision is now accepted and ready to land.Mar 9 2021, 7:45 AM Comment Actions
Could you include this in the commit summary please? csigg retitled this revision from [mlir] Add options to GPUKernelToBlobPass. to [mlir] Default for gpu-binary-annotation option..Mar 9 2021, 11:14 AM Comment Actions
I reverted part of a previous version. This change does not remove any options. csigg added a child revision: D98279: [mlir] Add base class for GpuKernelToBlobPass.Mar 9 2021, 11:47 AM This revision was landed with ongoing or failed builds.Mar 9 2021, 12:01 PM Closed by commit rG840ff84d33c7: [mlir] Default for gpu-binary-annotation option. (authored by csigg). · Explain Why This revision was automatically updated to reflect the committed changes.
Revision Contents
Diff 328969 mlir/include/mlir/Conversion/GPUCommon/GPUCommonPass.h
mlir/include/mlir/Conversion/Passes.td
mlir/lib/Conversion/GPUCommon/ConvertKernelFuncToBlob.cpp
mlir/lib/Conversion/PassDetail.h
mlir/tools/mlir-cuda-runner/mlir-cuda-runner.cpp
|
Nit: Triple /// comments please consistent with other places below.