[PATCH] D113371: [X86] combinePMULH - recognise 'cheap' trunctions via PACKS/PACKUS as well as SEXT/ZEXT (WIP)

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon updated this revision to Diff 386210.
RKSimon retitled this revision from "[X86] combinePMULH - recognise 'cheap' trunctions via PACKS/PACKUS as well as SEXT/ZEXT" to "[X86] combinePMULH - recognise 'cheap' trunctions via PACKS/PACKUS as well as SEXT/ZEXT (WIP)".
RKSimon added a comment.
Herald added a subscriber: ecnelises.

Use ComputeMinSignedBits/countMaxActiveBits

Add SimplifyVBinOp to DAGCombiner::visitMULHS/U to fold MULH(shuffle(x,u,m),shuffle(y,u,m)) -> shuffle(MULH(x,y),u,m) - annoyingly its tricky to expose this with existing MULH generation :(

There's still a few AVX2 regressions to address.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D113371/new/

https://reviews.llvm.org/D113371

Files:
  llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
  llvm/lib/Target/X86/X86ISelLowering.cpp
  llvm/test/CodeGen/X86/pmulh.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D113371.386210.patch
Type: text/x-patch
Size: 21812 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20211110/cf420520/attachment.bin>