[all-commits] [llvm/llvm-project] b37b18: [X86] Add truncate cost model tests to min-legal-v...

[topperc via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: b37b1840eb90168952832746baba53c4f7581151
      https://github.com/llvm/llvm-project/commit/b37b1840eb90168952832746baba53c4f7581151
  Author: Craig Topper <craig.topper at intel.com>
  Date:   2020-04-13 (Mon, 13 Apr 2020)

  Changed paths:
    M llvm/test/Analysis/CostModel/X86/min-legal-vector-width.ll

  Log Message:
  -----------
  [X86] Add truncate cost model tests to min-legal-vector-width.ll for when we're avoiding 512 bit vectors.


  Commit: 071c64d68d1d0c762c6d532f5312099e596c21ae
      https://github.com/llvm/llvm-project/commit/071c64d68d1d0c762c6d532f5312099e596c21ae
  Author: Craig Topper <craig.topper at intel.com>
  Date:   2020-04-13 (Mon, 13 Apr 2020)

  Changed paths:
    M llvm/lib/Target/X86/X86TargetTransformInfo.cpp
    M llvm/test/Analysis/CostModel/X86/min-legal-vector-width.ll
    M llvm/test/Analysis/CostModel/X86/trunc.ll

  Log Message:
  -----------
  [X86] Add a more accurate truncate cost for v8i64->v8i8


  Commit: 535a566a01e5aa3808de5af7e3d7fc2888eed0da
      https://github.com/llvm/llvm-project/commit/535a566a01e5aa3808de5af7e3d7fc2888eed0da
  Author: Craig Topper <craig.topper at intel.com>
  Date:   2020-04-13 (Mon, 13 Apr 2020)

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

  Log Message:
  -----------
  [X86] Split AVX512 getCastInstrCost into tables that require useAVX512Regs() and those that just operate on 256 or smaller vectors.

Use useAVX512Regs() to skip lookups instead of using type legalization
action.


  Commit: 2f60fbce6cdac9ef72339dde32946b883b6a96a8
      https://github.com/llvm/llvm-project/commit/2f60fbce6cdac9ef72339dde32946b883b6a96a8
  Author: Craig Topper <craig.topper at intel.com>
  Date:   2020-04-13 (Mon, 13 Apr 2020)

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

  Log Message:
  -----------
  [X86] Use a more realisitic cost for truncate v16i64->v16i8 with avx512f.

Still not great and we could probably codegen this better, but
11 was clearly ridiculous.


Compare: https://github.com/llvm/llvm-project/compare/23f896a096c3...2f60fbce6cda