projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: bfedf16) | patch
[memref][Transform][NFC] Improve the doc for masked_vectorize
authorQuentin Colombet <quentin.colombet@gmail.com>
Thu, 26 Jan 2023 14:36:44 +0000 (15:36 +0100)
committerQuentin Colombet <quentin.colombet@gmail.com>
Fri, 27 Jan 2023 10:20:05 +0000 (11:20 +0100)
commitbf5f63e59fd729efd6dd69318b365293f7ce385d
treedec61201e33c9893aece58bac000959d46fbda29tree | snapshot
parentbfedf169f4261a0fbe097f4a717d375dda2503c4commit | diff
[memref][Transform][NFC] Improve the doc for masked_vectorize

The `transform.structured.masked_vectorize` operator assumes that the
iteration space of the given linalg op is smaller than the given vector
sizes.
Explicitly states this requirement in the description of the operation.

Also fix the related comment and assert message in the vectorization code.
The wording was flipped.

NFC

Differential Revision: https://reviews.llvm.org/D142628
mlir/include/mlir/Dialect/Linalg/TransformOps/LinalgTransformOps.td diff | blob | history
mlir/lib/Dialect/Linalg/Transforms/Vectorization.cpp diff | blob | history
Domain: System / Toolchain;