projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 20ba079) | patch
[mlir][vector] Add fold pattern for InsertOp(Constant into Constant)
authorJakub Kuderski <kubak@google.com>
Sat, 26 Nov 2022 04:00:48 +0000 (23:00 -0500)
committerJakub Kuderski <kubak@google.com>
Sat, 26 Nov 2022 04:01:29 +0000 (23:01 -0500)
commit5d05d2966f4394e7f3fd5708ecc6f1b1c1156145
tree584c5bd96ea7c011bd2759fb06185a343e552fdetree | snapshot
parent20ba079dda7be1a72d64cebc9f55d909bf29f6c1commit | diff
[mlir][vector] Add fold pattern for InsertOp(Constant into Constant)

This pattern comes with vector size threshold to make sure we do not
introduce too many large constants.

This help clean up code created by the Wide Integer Emulation pass.

Reviewed By: dcaballe

Differential Revision: https://reviews.llvm.org/D138733
mlir/lib/Dialect/Vector/IR/VectorOps.cpp diff | blob | history
mlir/test/Dialect/Vector/canonicalize.mlir diff | blob | history
Domain: System / Toolchain;