[PATCH] D54836: [X86] Add DAG combine to combine a v8i32->v8i16 truncate with a packuswb that truncates v8i16->v8i8.

Simon Pilgrim via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Dec 3 00:00:48 PST 2018


RKSimon accepted this revision.
RKSimon added a comment.
This revision is now accepted and ready to land.

LGTM with one minor, cheers.



================
Comment at: lib/Target/X86/X86ISelLowering.cpp:35383
+  // create a larger truncate.
+  if (Subtarget.hasAVX512() && Opcode == X86ISD::PACKUS &&
+      N0.getOpcode() == ISD::TRUNCATE && N1.isUndef() && VT == MVT::v16i8 &&
----------------
craig.topper wrote:
> RKSimon wrote:
> > Can the PACKSS equivalent occur as well?
> It doesn't occur in any of our lit tests so I'm not sure.
OK - please can you add a TODO comment for now?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D54836/new/

https://reviews.llvm.org/D54836





More information about the llvm-commits mailing list