[all-commits] [llvm/llvm-project] e6eb84: [LoopVectorize] Use DataLayout::getIndexType inste...

[sdesmalen-arm via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: e6eb84a191ca2a1afd5789c5bb398da68bb6065e
      https://github.com/llvm/llvm-project/commit/e6eb84a191ca2a1afd5789c5bb398da68bb6065e
  Author: Sander de Smalen <sander.desmalen at arm.com>
  Date:   2023-02-07 (Tue, 07 Feb 2023)

  Changed paths:
    M llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-epilog-vect-inloop-reductions.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-epilog-vect-reductions.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-epilog-vect-strict-reductions.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-epilog-vect.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-fneg.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-gather-scatter.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-live-out-pointer-induction.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-runtime-check-size-based-threshold.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-tail-folding-unroll.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-vector-reverse.ll
    M llvm/test/Transforms/LoopVectorize/AArch64/sve-widen-phi.ll
    M llvm/test/Transforms/LoopVectorize/RISCV/divrem.ll
    M llvm/test/Transforms/LoopVectorize/RISCV/inloop-reduction.ll
    M llvm/test/Transforms/LoopVectorize/RISCV/mask-index-type.ll
    M llvm/test/Transforms/LoopVectorize/RISCV/scalable-basics.ll
    M llvm/test/Transforms/LoopVectorize/scalable-inductions.ll
    M llvm/test/Transforms/LoopVectorize/scalable-loop-unpredicated-body-scalar-tail.ll

  Log Message:
  -----------
  [LoopVectorize] Use DataLayout::getIndexType instead of i32 for non-constant GEP indices.

This is specifically relevant for loops that vectorize using a scalable VF,
where the code results in:

  %vscale = call i32 llvm.vscale.i32()
  %vf.part1 = mul i32 %vscale, 4
  %gep = getelementptr  ..., i32 %vf.part1

Which InstCombine then changes into:

  %vscale = call i32 llvm.vscale.i32()
  %vf.part1 = mul i32 %vscale, 4
  %vf.part1.zext = sext i32 %vf.part1 to i64
  %gep = getelementptr  ..., i32 %vf.part1.zext

D143016 tried to remove these extends, but that only works when
the call to llvm.vscale.i32() has a single use. After doing any
kind of CSE on these calls the combine no longer kicks in.

It seems more sensible to ask DataLayout what type to use, rather
than relying on InstCombine to insert the extend and hoping it can
fold it away.

I've only changed this for indices that are not constant, because
I vaguely remember there was a reason for sticking with i32. It
would also mean patching up loads more tests.

Reviewed By: paulwalker-arm

Differential Revision: https://reviews.llvm.org/D143267