projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: a6e155f) | patch
[MLIR][GPU] Detect bounds with `arith.minsi ` in loops-to-gpu
authorChristian Sigg <csigg@google.com>
Mon, 22 Aug 2022 08:39:49 +0000 (10:39 +0200)
committerChristian Sigg <csigg@google.com>
Mon, 22 Aug 2022 09:14:04 +0000 (11:14 +0200)
commit459fd3fb342d565bbaff48673838c5ea138128f8
tree79ca73f6fe73ca883544107333369f2b24af38abtree | snapshot
parenta6e155fd9a392285ab37e12777a8c140f913bbc5commit | diff
[MLIR][GPU] Detect bounds with `arith.minsi ` in loops-to-gpu

Previously, `arith.constant`, `arith.muli` and `affine.min` were supported when deriving upper loop bounds when converting parallel loops to GPU.

Reviewed By: akuegel

Differential Revision: https://reviews.llvm.org/D132354
mlir/lib/Conversion/SCFToGPU/SCFToGPU.cpp diff | blob | history
mlir/test/Conversion/SCFToGPU/parallel_loop.mlir diff | blob | history
Domain: System / Toolchain;