[llvm] [LV] Vectorize FMax w/o fast-math flags. (PR #146711)

[Florian Hahn via llvm-commits]

================
@@ -47,6 +47,8 @@ enum class RecurKind {
   FMul,     ///< Product of floats.
   FMin,     ///< FP min implemented in terms of select(cmp()).
   FMax,     ///< FP max implemented in terms of select(cmp()).
+  FMaxNoFMFs, ///< FP max implemented in terms of select(cmp()), but without
+              ///any fast-math flags. Users need to handle NaNs and signed zeros when generating code.
----------------
fhahn wrote:

Yeah, I originally named if FMaxNoFMFs in line with existing `FMax`, but the current version doesn't match all pattersn `FMax` currently does. Renamed as suggested, we can always rename once/if we reach parity

https://github.com/llvm/llvm-project/pull/146711