[all-commits] [llvm/llvm-project] b32915: [AArch64][LV] AArch64 does not prefer vectorized a...

Allen via All-commits all-commits at lists.llvm.org
Fri Jun 17 03:37:15 PDT 2022

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: b329156f4f1463d3937f6e9e46fb31c0fcee5532
  Author: Tiehu Zhang <zhangtiehu at huawei.com>
  Date:   2022-06-17 (Fri, 17 Jun 2022)

  Changed paths:
    M llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
    M llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
    A llvm/test/Transforms/LoopVectorize/AArch64/gather-do-not-vectorize-addressing.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/interleaved-vs-scalar.ll

  Log Message:
  [AArch64][LV] AArch64 does not prefer vectorized addressing

TTI::prefersVectorizedAddressing() try to vectorize the addresses that lead to loads.
For aarch64, only gather/scatter (supported by SVE) can deal with vectors of addresses.
This patch specializes the hook for AArch64, to return true only when we enable SVE.

Reviewed By: dmgreen

Differential Revision: https://reviews.llvm.org/D124612

More information about the All-commits mailing list