[all-commits] [llvm/llvm-project] d181fd: [CostModel][X86] Drop some hard coded fp<->int sca...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: d181fd918d18cbd99768f025e14a69d35d275f14
      https://github.com/llvm/llvm-project/commit/d181fd918d18cbd99768f025e14a69d35d275f14
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-07-02 (Fri, 02 Jul 2021)

  Changed paths:
    M llvm/lib/Target/X86/X86TargetTransformInfo.cpp
    M llvm/test/Analysis/CostModel/X86/sitofp.ll

  Log Message:
  -----------
  [CostModel][X86] Drop some hard coded fp<->int scalarization costs

Scalarization costs handling is a lot better now, and the hard coded costs were higher than the worse case numbers from the script in D103695


  Commit: d867634fbdb1f1f521981c244c410a9b839c8325
      https://github.com/llvm/llvm-project/commit/d867634fbdb1f1f521981c244c410a9b839c8325
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-07-02 (Fri, 02 Jul 2021)

  Changed paths:
    M llvm/lib/Target/X86/X86TargetTransformInfo.cpp

  Log Message:
  -----------
  [CostModel][X86] Update comment describing source of costs - we now use llvm-mca more than IACA


  Commit: e5fdff1cf822e8687ddb5568d14b8a609a5831d0
      https://github.com/llvm/llvm-project/commit/e5fdff1cf822e8687ddb5568d14b8a609a5831d0
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2021-07-02 (Fri, 02 Jul 2021)

  Changed paths:
    M llvm/lib/Target/X86/X86ScheduleSLM.td

  Log Message:
  -----------
  [X86][SLM] Keep similar scheduler costs types together. NFCI.

The SLM model is inconsistent about where it kept its 'unsupported' schedule classes - better to keep them close to similar classes.

I'm not sure why some ymm classes are defined and others are unsupported though (but I haven't altered them) - the only SLM-like CPU supporting any ymm is KNL and that currently uses the HSW model.


Compare: https://github.com/llvm/llvm-project/compare/5df556ac8bb8...e5fdff1cf822