[all-commits] [llvm/llvm-project] 280874: [X86] LowerVSETCC - always split 512-bit vectors b...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 2808743cbdcee806eecd7b8799cf776727dc574b
      https://github.com/llvm/llvm-project/commit/2808743cbdcee806eecd7b8799cf776727dc574b
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2022-02-15 (Tue, 15 Feb 2022)

  Changed paths:
    M llvm/lib/Target/X86/X86ISelLowering.cpp
    A llvm/test/CodeGen/X86/pr53842.ll

  Log Message:
  -----------
  [X86] LowerVSETCC - always split 512-bit vectors before lowering to PCMPEQ/GT (PR53842)

Extend the existing split where we already do this for v32i16/v64i8

We can end up trying to use PCMPEQ/GT if the result needs to be sign-extended (typically due to the DAGCombiner::foldSextSetcc fold).

Fixes #53842