[llvm] [VPlan] Implement VPWidenStoreEVLRecipe::computeCost(). (PR #109644)

[Florian Hahn via llvm-commits]

================
@@ -0,0 +1,25 @@
+; RUN: opt < %s --prefer-predicate-over-epilogue=predicate-dont-vectorize --passes=loop-vectorize -mcpu=sifive-p470 -mattr=+v,+f
+; RUN: opt < %s --prefer-predicate-over-epilogue=predicate-dont-vectorize --passes=loop-vectorize -mcpu=sifive-p470 -mattr=+v,+f -force-tail-folding-style=data-with-evl
+; Generated from issue #109468.
----------------
fhahn wrote:

Could you add an explanation to the test. IIUC the important bit is that the store doesn't need a mask with EVL?

https://github.com/llvm/llvm-project/pull/109644