[PATCH] D48174: [SLPVectorizer] Relax "alternate" opcode vectorisation to work with any SK_Select shuffle pattern

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon updated this revision to Diff 152048.
RKSimon added a comment.

Rebased - the aarch64 regression is purely due to the incorrect shuffle costs in AArch64TargetTransformInfo.cpp which https://reviews.llvm.org/D48172 is handling.

I'd like to get this patch in to simplify my next work in the SLP, instead of waiting for https://reviews.llvm.org/D48172 - what do people think?


Repository:
  rL LLVM

https://reviews.llvm.org/D48174

Files:
  lib/Transforms/Vectorize/SLPVectorizer.cpp
  test/Transforms/SLPVectorizer/AArch64/transpose.ll
  test/Transforms/SLPVectorizer/X86/addsub.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D48174.152048.patch
Type: text/x-patch
Size: 8923 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180620/bb53a814/attachment.bin>