diff options
| author | Matthias Gehre <matthias.gehre@amd.com> | 2025-01-20 08:48:28 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-01-20 08:48:28 +0100 |
| commit | 5c6db8c9e0c488585e642e0c17582ba3aaab7518 (patch) | |
| tree | e1170767e86cfd13df6fbb77d56f48118f035c2a /llvm/lib/CodeGen/CodeGen.cpp | |
| parent | 046b064df0ac9d4530e79f11077a768383b1ca16 (diff) | |
| download | llvm-5c6db8c9e0c488585e642e0c17582ba3aaab7518.zip llvm-5c6db8c9e0c488585e642e0c17582ba3aaab7518.tar.gz llvm-5c6db8c9e0c488585e642e0c17582ba3aaab7518.tar.bz2 | |
[MLIR] TosaToLinalg: Prefer to emit identity maps (#123295)
When deciding whether to emit a map like
`#map = affine_map<(d0, d1, d2, d3) -> (0, d1, d2, d3)>` or `#map =
affine_map<(d0, d1, d2, d3) -> (d0, d1, d2, d3)>` for an operand of a
`linalg.generic` when lowering element-wise TOSA ops, prefer the latter
unless broadcasting of the operand is really needed.
This helps later transformations which often require the affine map to
be a projected permuatation.
Diffstat (limited to 'llvm/lib/CodeGen/CodeGen.cpp')
0 files changed, 0 insertions, 0 deletions