commit | 70dd3b8ff958c4cc268d0c324a56cfabe26fcdcf | [log] [tgz] |
---|---|---|
author | Cullen Rhodes <cullen.rhodes@arm.com> | Tue Apr 23 07:21:06 2024 +0100 |
committer | Copybara-Service <copybara-worker@google.com> | Mon Apr 22 23:27:48 2024 -0700 |
tree | d006cc2ed1d15d08463321b1447f167670426477 | |
parent | 761715c4268441545f29b157aa8565284a6c4e1e [diff] |
[mlir][linalg] Add patterns to convert matmul to transposed variants (#89075) This adds patterns to convert from the Linalg matmul and batch_matmul ops to the transposed variants. By default the LHS matrix is transposed. Our work enabling a lowering path from linalg.matmul to ArmSME has revealed the current lowering results in non-contiguous memory accesses for the A matrix and very poor performance. These patterns provide a simple option to fix this. GitOrigin-RevId: 79225349748bb556fd027cc0bfeb73b1e9a632f4
See https://mlir.llvm.org/ for more information.