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

[Florian Hahn via llvm-commits]

================
@@ -628,3 +628,118 @@ void VPlanTransforms::attachCheckBlock(VPlan &Plan, Value *Cond,
     Term->addMetadata(LLVMContext::MD_prof, BranchWeights);
   }
 }
+
+bool VPlanTransforms::handleFMaxReductionsWithoutFastMath(VPlan &Plan) {
+  VPRegionBlock *LoopRegion = Plan.getVectorLoopRegion();
+  VPReductionPHIRecipe *RedPhiR = nullptr;
+  VPRecipeWithIRFlags *MaxOp = nullptr;
+  VPWidenIntOrFpInductionRecipe *WideIV = nullptr;
+
+  // Check if there are any FCmpOGTSelect reductions using wide selects that we
+  // can fix up. To do so, we also need a  wide canonical IV to keep track of
----------------
fhahn wrote:

Updated, thanks

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