[all-commits] [llvm/llvm-project] a1c892: [X86][SSE] lowerVECTOR_SHUFFLE - canonicalize with...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: a1c892b439899a78e39f6ea0aaf8871ca385f1cd
      https://github.com/llvm/llvm-project/commit/a1c892b439899a78e39f6ea0aaf8871ca385f1cd
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-08-22 (Sun, 22 Aug 2021)

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

  Log Message:
  -----------
  [X86][SSE] lowerVECTOR_SHUFFLE - canonicalize with horizontal ops.

Before lowering shuffles, see if we can merge horizontal ops or canonicalize the shuffle mask to point to the same LHS/RHS of the HOps when an HOp's args are repeated.


  Commit: 7b7ac4b16a2f2bab1f961e3b362c6a231fa923d2
      https://github.com/llvm/llvm-project/commit/7b7ac4b16a2f2bab1f961e3b362c6a231fa923d2
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-08-22 (Sun, 22 Aug 2021)

  Changed paths:
    M llvm/test/CodeGen/X86/avx.ll
    M llvm/test/CodeGen/X86/sse41.ll

  Log Message:
  -----------
  [X86] Expose memory codegen in element insert load tests to improve accuracy of checks

Also replace X32 with X86 check prefixes for i686 tests (we tend to try to use X32 for gnux32 targets)


Compare: https://github.com/llvm/llvm-project/compare/dcf659e8219b...7b7ac4b16a2f