[all-commits] [llvm/llvm-project] c46634: [LoadStoreVectorizer] Consider if operation is fas...

[Stanislav Mekhanoshin via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: c46634554d507b6b764eeba881ed4eef3b28baac
      https://github.com/llvm/llvm-project/commit/c46634554d507b6b764eeba881ed4eef3b28baac
  Author: Stanislav Mekhanoshin <Stanislav.Mekhanoshin at amd.com>
  Date:   2022-11-28 (Mon, 28 Nov 2022)

  Changed paths:
    M llvm/lib/Transforms/Vectorize/LoadStoreVectorizer.cpp

  Log Message:
  -----------
  [LoadStoreVectorizer] Consider if operation is faster than before

Compare a relative speed of misaligned accesses before and
after vectorization, not just check the new instruction is
not going to be slower.

Since no target now returns anything but 0 or 1 for Fast
argument of the allowsMisalignedMemoryAccesses this is still NFCI.

The subsequent patch will tune actual vaues of Fast on AMDGPU.

Differential Revision: https://reviews.llvm.org/D124218