[Mlir-commits] [mlir] [mlir][amdgpu] Add scaled_ext_packed{8, 16} operations (PR #159830)

[Jakub Kuderski]

================

----------------
kuhar wrote:

nit/optional: Why functional type though? We can't have multiple return values, so I think the parens don't help much... This is fine, but I'd slightly prefer `a, b -> c` like a bunch of existing code does, e.g.: https://mlir.llvm.org/docs/Dialects/SPIR-V/#spirvgroupnonuniformfmax-spirvgroupnonuniformfmaxop

https://github.com/llvm/llvm-project/pull/159830