[all-commits] [llvm/llvm-project] bad4de: Don't disable loop unroll for vectorized loops on ...

[alex-t via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: bad4de1ae7fa6dffec2f038118d6acedb9fce38e
      https://github.com/llvm/llvm-project/commit/bad4de1ae7fa6dffec2f038118d6acedb9fce38e
  Author: Alexander Timofeev <alexander.timofeev at amd.com>
  Date:   2023-05-25 (Thu, 25 May 2023)

  Changed paths:
    M llvm/include/llvm/Analysis/TargetTransformInfo.h
    M llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
    M llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
    A llvm/test/CodeGen/AMDGPU/vectorize-unroll-metadata.ll

  Log Message:
  -----------
  Don't disable loop unroll for vectorized loops on AMDGPU target

We've got a performance regression after the https://reviews.llvm.org/D115261.
Despite the loop being vectorized unroll is still required.

Reviewed By: rampitec

Differential Revision: https://reviews.llvm.org/D149281