[all-commits] [llvm/llvm-project] 11bb4a: [X86] combinePredicateReduction - split vXi16 allo...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 11bb4a11116c4937b5f7e851189c593abf28e682
      https://github.com/llvm/llvm-project/commit/11bb4a11116c4937b5f7e851189c593abf28e682
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2022-01-24 (Mon, 24 Jan 2022)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    M llvm/test/CodeGen/X86/vector-reduce-and-bool.ll

  Log Message:
  -----------
  [X86] combinePredicateReduction - split vXi16 allof(cmpeq()) to vXi8 allof(cmpeq())

vXi16 patterns allof(cmp()) reduction patterns will have to be pack the comparison results to vXi8 to use PMOVMSKB.

If we're reducing cmpeq(), then we can compare the vXi8 halves directly - similar to what we already do for vXi64 -> vXi32 for cases without PCMPEQQ.