[PATCH] D26297: [AVX-512] Add support for lowering shuffles to VALIGND/VALIGNQ

[Craig Topper via llvm-commits]

craig.topper added inline comments.


================
Comment at: lib/Target/X86/X86ISelLowering.cpp:12215
   if (Subtarget.hasBWI())
     if (SDValue Rotate = lowerVectorShuffleAsByteRotate(
             DL, MVT::v16i32, V1, V2, Mask, Subtarget, DAG))
----------------
RKSimon wrote:
> Will this ever fire?
VALIGND/Q work on the whole vector. PALIGNR works on 128-bit lanes so they should cover different cases.


https://reviews.llvm.org/D26297