projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: c4d955d) | patch
[mlir][VectorToGPU] Fix bug generating incorrect ldmatrix ops
authorThomas Raoux <thomasraoux@google.com>
Fri, 3 Jun 2022 04:23:27 +0000 (04:23 +0000)
committerThomas Raoux <thomasraoux@google.com>
Fri, 3 Jun 2022 04:30:22 +0000 (04:30 +0000)
commit271a48e02917859cd09ee7f230adea7b6cc7a578
treef8044c920eb089a1659782881282a10be82c5c19tree | snapshot
parentc4d955dd7f68b80119529165bbd878b09271c4d3commit | diff
[mlir][VectorToGPU] Fix bug generating incorrect ldmatrix ops

ldmatrix transpose can only be used with types that are 16bits wide.

Differential Revision: https://reviews.llvm.org/D126846
mlir/lib/Conversion/VectorToGPU/VectorToGPU.cpp diff | blob | history
mlir/test/Conversion/VectorToGPU/vector-to-mma-ops-mma-sync.mlir diff | blob | history
Domain: System / Toolchain;