[PATCH] D43776: [SLP] Fix PR36481: vectorize reassociated instructions.

[Alexey Bataev via Phabricator via llvm-commits]

ABataev created this revision.
ABataev added reviewers: RKSimon, spatel, hfinkel, mkuper, Ayal, ashahid.

If the load/extractelement/extractvalue instructions are not originally
consecutive, the SLP vectorizer is unable to vectorize them. Patch
allows reordering of such instructions.


Repository:
  rL LLVM

https://reviews.llvm.org/D43776

Files:
  include/llvm/Analysis/LoopAccessAnalysis.h
  lib/Analysis/LoopAccessAnalysis.cpp
  lib/Transforms/Vectorize/SLPVectorizer.cpp
  test/Transforms/SLPVectorizer/X86/extract.ll
  test/Transforms/SLPVectorizer/X86/jumbled-load-multiuse.ll
  test/Transforms/SLPVectorizer/X86/jumbled-load.ll
  test/Transforms/SLPVectorizer/X86/reassociated-loads.ll
  test/Transforms/SLPVectorizer/X86/store-jumbled.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D43776.135948.patch
Type: text/x-patch
Size: 44687 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180226/5d3825a6/attachment.bin>