[all-commits] [llvm/llvm-project] 14d1e5: [X86] combineConcatVectorOps - fold a 512-bit spla...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 14d1e502df1613659e82290d7e31179a2ac97234
      https://github.com/llvm/llvm-project/commit/14d1e502df1613659e82290d7e31179a2ac97234
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2023-08-11 (Fri, 11 Aug 2023)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/avx512-intrinsics-upgrade.ll
    M llvm/test/CodeGen/X86/avx512dq-intrinsics-upgrade.ll
    M llvm/test/CodeGen/X86/pr36199.ll
    M llvm/test/CodeGen/X86/subvector-broadcast.ll
    M llvm/test/CodeGen/X86/vector-shuffle-512-v16.ll
    M llvm/test/CodeGen/X86/vector-shuffle-512-v8.ll

  Log Message:
  -----------
  [X86] combineConcatVectorOps - fold a 512-bit splat of a 128-bit subvector to a single X86ISD::SHUF128 node.

Replaces a pair of insert_subvectors with a single (implicitly widened) vector - also reduce uses of the src.

Hopefully this should address most of the remaining widen subvector regressions I'm seeing while trying to aggressively convert TRUNCATE to PACKSS/PACKUS.