diff options
author | Artemiy Bulavin <artemiyb@graphcore.ai> | 2025-03-12 13:16:49 -0400 |
---|---|---|
committer | GitHub <noreply@github.com> | 2025-03-12 17:16:49 +0000 |
commit | fc127ff53d0c816e9e9a64ef55868479e0b84ebd (patch) | |
tree | ad872da677795ab18b8013e0f3c09024b02f47b2 /lldb/packages/Python/lldbsuite/test/gdbclientutils.py | |
parent | 1a626e63b5a009075eea87c01f0661144b1ec010 (diff) | |
download | llvm-fc127ff53d0c816e9e9a64ef55868479e0b84ebd.zip llvm-fc127ff53d0c816e9e9a64ef55868479e0b84ebd.tar.gz llvm-fc127ff53d0c816e9e9a64ef55868479e0b84ebd.tar.bz2 |
[mlir] Extract RHS rows once when lowering vector.contract to dot (#130130)
The `vector.contract` op on two matrices A and B will be lowered to
individual dot products of each row and column of A and B respectively.
The existing lowering will extract each column of B for each row of A,
which leads to multiple values in the IR representing the same columns
of B.
This PR makes changes to the `ContractOpToDotLowering` to make sure that
the columns of B are only ever extracted once, so then the SSA values
representing the extracted columns are then re-used in the IR for later
dot products.
I have updated the existing vector-contract-to-dot-transforms test.
Diffstat (limited to 'lldb/packages/Python/lldbsuite/test/gdbclientutils.py')
0 files changed, 0 insertions, 0 deletions