[Mlir-commits] [mlir] [mlir][AMDGPU] Add scaled wmma ops for gfx1250 (PR #169854)

[Krzysztof Drewniak]

https://github.com/krzysz00 requested changes to this pull request.

The design is too low level (exposes constants that are just types at the MLIR level and doesn't properly expose the underlying types of the "i32" scales) and also there are no tests 

https://github.com/llvm/llvm-project/pull/169854