projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 23dcdbf) | patch
[mlir][NVGPU][transform] Add `create_async_groups` transform op
authorMatthias Springer <me@m-sp.org>
Tue, 18 Jul 2023 12:13:31 +0000 (14:13 +0200)
committerMatthias Springer <me@m-sp.org>
Tue, 18 Jul 2023 12:36:41 +0000 (14:36 +0200)
commitdb393288ff9b07d2c13b2be6a61f91ba180cb601
treec9a248a91691bcd4cbfc0bacbfd8143ddd18d525tree | snapshot
parent23dcdbfba790eb24141d0d2523aab34a2a4d1ef5commit | diff
[mlir][NVGPU][transform] Add `create_async_groups` transform op

This transform looks for suitable vector transfers from global memory to shared memory and converts them to async device copies.

Differential Revision: https://reviews.llvm.org/D155569
mlir/include/mlir/Dialect/NVGPU/TransformOps/NVGPUTransformOps.td diff | blob | history
mlir/include/mlir/Dialect/NVGPU/Transforms/Transforms.h diff | blob | history
mlir/include/mlir/Dialect/NVGPU/Transforms/Utils.h diff | blob | history
mlir/include/mlir/Dialect/Vector/IR/VectorOps.td diff | blob | history
mlir/lib/Dialect/NVGPU/TransformOps/CMakeLists.txt diff | blob | history
mlir/lib/Dialect/NVGPU/TransformOps/NVGPUTransformOps.cpp diff | blob | history
mlir/lib/Dialect/NVGPU/Transforms/CMakeLists.txt diff | blob | history
mlir/lib/Dialect/NVGPU/Transforms/CreateAsyncGroups.cpp [new file with mode: 0644] blob
mlir/lib/Dialect/NVGPU/Transforms/Utils.cpp diff | blob | history
mlir/test/Dialect/NVGPU/transform-create-async-groups.mlir [new file with mode: 0644] blob
utils/bazel/llvm-project-overlay/mlir/BUILD.bazel diff | blob | history
Domain: System / Toolchain;