[all-commits] [llvm/llvm-project] b27553: [X86] Remove LowerToHorizontalOp vectorization low...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: b2755364764f0159a0e74a7111c3fc2140ec43ee
      https://github.com/llvm/llvm-project/commit/b2755364764f0159a0e74a7111c3fc2140ec43ee
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2026-07-04 (Sat, 04 Jul 2026)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp

  Log Message:
  -----------
  [X86] Remove LowerToHorizontalOp vectorization lowering code (#207406)

Thanks to improvements to SLP, InstCombine and VectorCombine - we should
no longer see any of the scalar code hadd/sub patterns in the backend.

This allows to remove the old
build_vector(x[0]+x[1],x[2]+x[3],y[0]+y[1],y[2]+y[3]) lowering entirely,
and just rely on the combineToHorizontalAddSub folds from
add(shuffle(x,y),shuffle(x,y)) patterns.

Fixes #143000



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications