[all-commits] [llvm/llvm-project] 94620e: [CostModel][X86] Add CostKinds handling for vector...

[Simon Pilgrim via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 94620e4fc340dc392e863af1c86441a20511ebd7
      https://github.com/llvm/llvm-project/commit/94620e4fc340dc392e863af1c86441a20511ebd7
  Author: Simon Pilgrim <llvm-dev at redking.me.uk>
  Date:   2022-09-15 (Thu, 15 Sep 2022)

  Changed paths:
    M llvm/lib/Target/X86/X86TargetTransformInfo.cpp
    M llvm/test/Analysis/CostModel/X86/div-codesize.ll
    M llvm/test/Analysis/CostModel/X86/div-latency.ll
    M llvm/test/Analysis/CostModel/X86/div-sizelatency.ll
    M llvm/test/Analysis/CostModel/X86/div.ll
    M llvm/test/Analysis/CostModel/X86/fshl.ll
    M llvm/test/Analysis/CostModel/X86/fshr.ll
    M llvm/test/Analysis/CostModel/X86/intrinsic-cost-kinds.ll
    M llvm/test/Analysis/CostModel/X86/mul.ll
    M llvm/test/Analysis/CostModel/X86/rem-codesize.ll
    M llvm/test/Analysis/CostModel/X86/rem-latency.ll
    M llvm/test/Analysis/CostModel/X86/rem-sizelatency.ll
    M llvm/test/Analysis/CostModel/X86/rem.ll
    M llvm/test/Analysis/CostModel/X86/vshift-ashr-codesize.ll
    M llvm/test/Analysis/CostModel/X86/vshift-ashr-cost-inseltpoison.ll
    M llvm/test/Analysis/CostModel/X86/vshift-ashr-cost.ll
    M llvm/test/Analysis/CostModel/X86/vshift-ashr-latency.ll
    M llvm/test/Analysis/CostModel/X86/vshift-ashr-sizelatency.ll
    M llvm/test/Analysis/CostModel/X86/vshift-lshr-codesize.ll
    M llvm/test/Analysis/CostModel/X86/vshift-lshr-cost-inseltpoison.ll
    M llvm/test/Analysis/CostModel/X86/vshift-lshr-cost.ll
    M llvm/test/Analysis/CostModel/X86/vshift-lshr-latency.ll
    M llvm/test/Analysis/CostModel/X86/vshift-lshr-sizelatency.ll
    M llvm/test/Analysis/CostModel/X86/vshift-shl-codesize.ll
    M llvm/test/Analysis/CostModel/X86/vshift-shl-cost-inseltpoison.ll
    M llvm/test/Analysis/CostModel/X86/vshift-shl-cost.ll
    M llvm/test/Analysis/CostModel/X86/vshift-shl-latency.ll
    M llvm/test/Analysis/CostModel/X86/vshift-shl-sizelatency.ll
    M llvm/test/Transforms/SLPVectorizer/X86/alternate-int-inseltpoison.ll
    M llvm/test/Transforms/SLPVectorizer/X86/alternate-int.ll

  Log Message:
  -----------
  [CostModel][X86] Add CostKinds handling for vector shift by generic/non-uniform shift amounts

These are the worst case generic vector shift costs, where nothing is known about the shift amounts - in particular this should stop us using the default sizelatency cost of 1 for so many pre-AVX2 vector shifts that can often actually expand during lowering to +20 uops, just for 128-bit vectors, resulting in some horrible inline/unroll decisions.

This was achieved with an updated version of the 'cost-tables vs llvm-mca' script D103695 (I'll update the patch soon for reference)