[Mlir-commits] [mlir] [MLIR][Vector] Support vector reduction for xor/maxnumf/minnumf (PR #163310)

[Andrzej Warzyński]

https://github.com/banach-space commented:

Thanks for this contribution!

I am not familiar with the Affine part, but the tests that you added look like a pretty straightforward extension of the existing tests.

That said, for a change like this I would expect more comprehensive testing. Here are two places where new tests should be added:
* https://github.com/llvm/llvm-project/blob/b228a18e572ac7ab7a2f61b97c0a2dd3565a8201/mlir/test/Dialect/Vector/ops.mlir#L609-L629
* https://github.com/llvm/llvm-project/blob/b228a18e572ac7ab7a2f61b97c0a2dd3565a8201/mlir/test/Dialect/Vector/break-down-vector-reduction.mlir#L16-L24

There's definitely more :) (*) 

Thanks!

(*) To clarify, I am not suggesting that you add these new variants _everywhere_ `vector.reduction` is used. However, please update tests that enumerate over all the existing options.

https://github.com/llvm/llvm-project/pull/163310