[all-commits] [llvm/llvm-project] a1c952: [X86][AVX] isHorizontalBinOp - relax no-lane-cross...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: a1c9529e602adcc25162f93fd4563fbe3d04ab62
      https://github.com/llvm/llvm-project/commit/a1c9529e602adcc25162f93fd4563fbe3d04ab62
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2020-07-29 (Wed, 29 Jul 2020)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/haddsub-4.ll

  Log Message:
  -----------
  [X86][AVX] isHorizontalBinOp - relax no-lane-crossing limit for AVX1-only targets.

Instead of never accepting v8f32/v4f64 FHADD/FHSUB if the input shuffle masks cross lanes, perform the matching and determine if the post shuffle mask simplifies to a 'whole lane shuffle' mask - in which case we are guaranteed to cheaply perform this as a VPERM2F128 shuffle.