[llvm-branch-commits] [llvm] release/21.x: [LV] Vectorize maxnum/minnum w/o fast-math flags. (#148239) (PR #149736)

[via llvm-branch-commits]

ayalz wrote:

> @ayalz What do you think about merging this PR to the release branch?

Sure, fine w/ me, looks profitable and stable enough.

Nit: name could be shortened from `‏handleMaxMinNumReductionsWithoutFastMath()` to `‏handleMaxMinNumReductions()`

https://github.com/llvm/llvm-project/pull/149736