[all-commits] [llvm/llvm-project] 6ba0b9: [X86][SLM] Fix PBLENDVB uops and throughput

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 6ba0b9f68ac9b3a058a2e98e5c4d1e039eff662f
      https://github.com/llvm/llvm-project/commit/6ba0b9f68ac9b3a058a2e98e5c4d1e039eff662f
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-09-03 (Fri, 03 Sep 2021)

  Changed paths:
    M llvm/lib/Target/X86/X86ScheduleSLM.td
    M llvm/test/tools/llvm-mca/X86/SLM/resources-sse41.s

  Log Message:
  -----------
  [X86][SLM] Fix PBLENDVB uops and throughput

SLM PBLENDVB is just as bad as BLENDVPD/PS - so model it as such, fixing the rr vs rm uops diff as well. The Intel AoM appears to have a copy+paste typo with PBLENDW, it doesn't match Agner or InstLatX64.

Noticed while investigating some of the weird discrepancies reported by the D103695 helper script (SLM had much better vector shift throughputs than it should).