[PATCH] D106859: [Sink] allow sinking convergent operations across uniform branches

[Sameer Sahasrabuddhe via Phabricator via llvm-commits]

sameerds updated this revision to Diff 364354.
sameerds added a comment.

Replaced the full-fledged divergence analysis with a trivial
DivergenceInfo object instead. The change also allows short-circuiting
the divergence analysis in lit tests, as demonstrated by changes to
LoopUnswitch.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D106859/new/

https://reviews.llvm.org/D106859

Files:
  llvm/include/llvm/Analysis/DivergenceAnalysis.h
  llvm/include/llvm/Analysis/LegacyDivergenceAnalysis.h
  llvm/lib/Analysis/DivergenceAnalysis.cpp
  llvm/lib/Analysis/LegacyDivergenceAnalysis.cpp
  llvm/lib/Transforms/Scalar/LoopUnswitch.cpp
  llvm/lib/Transforms/Scalar/SimpleLoopUnswitch.cpp
  llvm/lib/Transforms/Scalar/Sink.cpp
  llvm/test/Analysis/DivergenceAnalysis/AMDGPU/irreducible.ll
  llvm/test/Transforms/LoopUnswitch/AMDGPU/divergent-unswitch.ll
  llvm/test/Transforms/LoopUnswitch/divergent.ll
  llvm/test/Transforms/Sink/convergent.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D106859.364354.patch
Type: text/x-patch
Size: 27910 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210805/3c55cae1/attachment.bin>