[Mlir-commits] [mlir] [mlir] [arith] Fold trunc(extremum(ext(lhs), ext(rhs))) (PR #214658)

[Chuanqi Xu]

ChuanqiXu9 wrote:

If I read correctly, the summary of the discussion can be: the conversion the patch itself want to do is fine. But for the general case, especially for accumulating, we need to be more careful. Do I understand right?

If yes, would you like to take a look at the patch itself? Thanks.

https://github.com/llvm/llvm-project/pull/214658