[PATCH] D44540: [X86] Merge ADDSUB/SUBADD detection into single methods that can detect either and indicate what they found.

[Simon Pilgrim via Phabricator via llvm-commits]

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

LGTM with a couple of minors.

I haven't checked recently - is PR30633 finished now?



================
Comment at: lib/Target/X86/X86ISelLowering.cpp:7600
+  // Ensure we have found an opcode for both parities and that they are
+  // different. Don't try to fold this build_vector into an ADDSUB if the inputs
+  // are undef.
----------------
ADDSUB/SUBADD


================
Comment at: lib/Target/X86/X86ISelLowering.cpp:7695
   // 512-bit ADDSUB idiom recognition was needed only as part of FMADDSUB idiom
   // recognition.
   if (VT.is512BitVector())
----------------
What is better - splitting into 2 x 256-bit entries or BLEND(ADD.SUB) ?


https://reviews.llvm.org/D44540