[PATCH] D157368: [VPlan] Consider scalar VF in masked VPReductionRecipe

[Lou via Phabricator via llvm-commits]

iamlouk marked 3 inline comments as done.
iamlouk added inline comments.


================
Comment at: llvm/test/Transforms/LoopVectorize/AArch64/scalar_interleave_masked_reduce.ll:34-35
+exit:
+  %sum.lcssa = phi double [ %sum.next, %for.body ]
+  ret double %sum.lcssa
+}
----------------
ABataev wrote:
> Also, do you need this or phi can be removed and make function just return void?
I simplified the test a tiny bit more by removing the LCSSA phi, I used a return to make the value live-out. If there is another (preferred) way, I'd be happy to change this! 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D157368/new/

https://reviews.llvm.org/D157368