[PATCH] D114779: [LV][X86] Sink `LoopVectorizationCostModel::useEmulatedMaskMemRefHack()` further into TTI, disable for X86/AVX2+

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri created this revision.
lebedev.ri added reviewers: rengolin, RKSimon, fhahn, hsaito, yamauchi, Ayal.
lebedev.ri added a project: LLVM.
Herald added subscribers: pengfei, arphaman, hiraditya.
lebedev.ri requested review of this revision.

D43208 <https://reviews.llvm.org/D43208> extracted `useEmulatedMaskMemRefHack()` from legality into cost model.
What it essentially does is prevents scalarized vectorization of masked memory operations:

  // TODO: Cost model for emulated masked load/store is completely
  // broken. This hack guides the cost model to use an artificially
  // high enough value to practically disable vectorization with such
  // operations, except where previously deployed legality hack allowed
  // using very low cost values. This is to avoid regressions coming simply
  // from moving "masked load/store" check from legality to cost model.
  // Masked Load/Gather emulation was previously never allowed.
  // Limited number of Masked Store/Scatter emulation was allowed.

While i don't really understand about what specifically `is completely broken`
was talking about, i believe that at least on X86 with AVX2-or-later,
this is no longer true. (or at least, i would like to know what is still broken).
So i would like to follow suit after D111460 <https://reviews.llvm.org/D111460>, and like wise disable that hack for AVX2+.

This adds a similarly-named TTI hook, which is default-true,
but returns false for X86-with-AVX2.

There is an unexpected regression - the `optsize.ll` and `tripcount.ll`
start being vectorized. The comments there say they should not be,
because it's unprofitable. Are there some thresholds that need to be adjusted?


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D114779

Files:
  llvm/include/llvm/Analysis/TargetTransformInfo.h
  llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
  llvm/include/llvm/CodeGen/BasicTTIImpl.h
  llvm/lib/Analysis/TargetTransformInfo.cpp
  llvm/lib/Target/X86/X86TargetTransformInfo.cpp
  llvm/lib/Target/X86/X86TargetTransformInfo.h
  llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
  llvm/test/Analysis/CostModel/X86/masked-gather-i32-with-i8-index.ll
  llvm/test/Analysis/CostModel/X86/masked-gather-i64-with-i8-index.ll
  llvm/test/Analysis/CostModel/X86/masked-interleaved-load-i16.ll
  llvm/test/Analysis/CostModel/X86/masked-interleaved-store-i16.ll
  llvm/test/Analysis/CostModel/X86/masked-load-i16.ll
  llvm/test/Analysis/CostModel/X86/masked-load-i8.ll
  llvm/test/Transforms/LoopVectorize/X86/gather_scatter.ll
  llvm/test/Transforms/LoopVectorize/X86/x86-interleaved-accesses-masked-group.ll
  llvm/test/Transforms/LoopVectorize/optsize.ll
  llvm/test/Transforms/LoopVectorize/tripcount.ll
  llvm/test/Transforms/LoopVectorize/vplan-sink-scalars-and-merge.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D114779.390609.patch
Type: text/x-patch
Size: 302420 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20211130/b4fbb0b6/attachment-0001.bin>