[PATCH] D115329: [LoopVectorize] Pass a vector type to isLegalMaskedGather/Scatter

[Sander de Smalen via Phabricator via llvm-commits]

sdesmalen added a comment.

Hi @RosieSumpter, just wanted to let you know I'm happy with the patch. This looks like a good improvement to me. It conceptually makes more sense and it also allows AArch64TTI to better distinguish between Neon/SVE when using fixed-width vectors.
I want to accept the patch, but I'll defer this to @lebedev.ri because there's still an open question about the X86 cost-model.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D115329/new/

https://reviews.llvm.org/D115329