[llvm] [LV][EVL] Emit vp.merge intrinsic to enable out-loop reduction in EVL vectorization. (PR #101641)

[Florian Hahn via llvm-commits]

================
@@ -9386,10 +9397,8 @@ void LoopVectorizationPlanner::adjustRecipesForReductions(
                cast<VPInstruction>(&U)->getOpcode() ==
                    VPInstruction::ComputeReductionResult;
       });
-      if (PreferPredicatedReductionSelect ||
-          TTI.preferPredicatedReductionSelect(
-              PhiR->getRecurrenceDescriptor().getOpcode(), PhiTy,
-              TargetTransformInfo::ReductionFlags()))
+      if (CM.usePredicatedReductionSelect(
+              PhiR->getRecurrenceDescriptor().getOpcode(), PhiTy))
----------------
fhahn wrote:

At the moment, everything EVL related is applied during the transform that introduces EVL recipes; one potential issue is that we assume EVL is used here, but the transform may not apply.

Don't have any strong preferences, doing it later indeed seems to require some extra work

https://github.com/llvm/llvm-project/pull/101641