[PATCH] D129890: [SDAG] narrow truncated sign_extend_inreg

[Sanjay Patel via Phabricator via llvm-commits]

spatel updated this revision to Diff 445231.
spatel retitled this revision from "[SDAG] reduce cast ops around sign_extend_inreg" to "[SDAG] narrow truncated sign_extend_inreg".
spatel edited the summary of this revision.
spatel added reviewers: foad, arsenm.
spatel added a comment.
Herald added subscribers: jsji, kosarev, kerbowa, pengfei, wdng, jvesely.

Patch updated:
On closer inspection, we can generalize the fold to only match trunc-of-sext (don't need to match the preceding extend), and that shows improvements on existing tests for x86 and (I think) AMDGPU.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D129890/new/

https://reviews.llvm.org/D129890

Files:
  llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
  llvm/test/CodeGen/AMDGPU/mul_int24.ll
  llvm/test/CodeGen/X86/pmulh.ll
  llvm/test/CodeGen/X86/sar_fold64.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D129890.445231.patch
Type: text/x-patch
Size: 4720 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220716/fbf86229/attachment.bin>