[Mlir-commits] [mlir] [MLIR][Arith] expand-ops: Support mini/maxi (PR #90575)

[Matthias Gehre]

mgehre-amd wrote:

> The implementation looks good, but can you explain why we need to expand them in the first place? I'm not familiar with the LLVM lowering; it seems like LLVM has matching intrinsics: [llvm.org/docs/LangRef.html#llvm-smin-intrinsic](https://llvm.org/docs/LangRef.html#llvm-smin-intrinsic)

Sure: We don't target LLVM and our target dialect has no representation for them.

https://github.com/llvm/llvm-project/pull/90575