[PATCH] D44370: [X86] Combine vXi64 multiplies to MULDQ/MULUDQ during DAG combine instead of lowering.

Tue Mar 27 17:49:59 PDT 2018

craig.topper updated this revision to Diff 140027.
craig.topper added a comment.
Herald added a subscriber: javed.absar.

Added a target independent DAG combine to create zero_extend from a build_vector of zexts from a contiguous range of an input vector. This more or less recovers the regression. I can split that change out of this patch if you want. I'll definitely split it for commit.

https://reviews.llvm.org/D44370

Files:
  lib/CodeGen/SelectionDAG/DAGCombiner.cpp
  lib/Target/X86/X86ISelLowering.cpp
  test/CodeGen/ARM/dagcombine-anyexttozeroext.ll
  test/CodeGen/X86/combine-pmuldq.ll
  test/CodeGen/X86/masked_gather_scatter.ll
  test/CodeGen/X86/mulvi32.ll
  test/CodeGen/X86/pmul.ll
  test/CodeGen/X86/xop-ifma.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D44370.140027.patch
Type: text/x-patch
Size: 17968 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20180328/97ccd124/attachment.bin>