[PATCH] D102476: [VectorCombine] Use constant range info for index scalarization legality.

Florian Hahn via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Aug 12 03:13:58 PDT 2021


fhahn marked 2 inline comments as done.
fhahn added inline comments.


================
Comment at: llvm/test/Transforms/VectorCombine/load-insert-store.ll:195
+; CHECK-NEXT:    [[TMP0:%.*]] = getelementptr inbounds <16 x i8>, <16 x i8>* [[Q:%.*]], i32 0, i32 [[IDX_CLAMPED]]
+; CHECK-NEXT:    store i8 [[S:%.*]], i8* [[TMP0]], align 1
 ; CHECK-NEXT:    ret void
----------------
fhahn wrote:
> efriedma wrote:
> > efriedma wrote:
> > > Alive doesn't like this: https://web.ist.utl.pt/nuno.lopes/alive2/index.php?hash=50de39e1c7237e13&test=Transforms%2FVectorCombine%2Fload-insert-store.ll
> > Filed as https://bugs.llvm.org/show_bug.cgi?id=50949
> Apologies for the delay! I put up a patch limiting the transform to cases where the index is known non-poison: D107364
> 
Ok, looks like things look good now on the latest run https://web.ist.utl.pt/nuno.lopes/alive2/index.php?hash=295f040c6c7caf3e


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D102476/new/

https://reviews.llvm.org/D102476



More information about the llvm-commits mailing list