[PATCH] D40373: [x86][icelake]GFNI

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: include/llvm/IR/IntrinsicsX86.td:1347
+  // AVX512
+  def int_x86_avx512_mask_vgf2p8affineinvqb_128 :
+         GCCBuiltin<"__builtin_ia32_vgf2p8affineinvqb_v16qi_mask">,
----------------
coby wrote:
> craig.topper wrote:
> > Can we use __builtin_ia32_selectb_512 and  __builtin_ia32_vgf2p8affineinvqb_v16qi to implement these in the clang header instead of adding a separate masked intrinsic?
> > 
> > Its different than gcc, but consistent with other intrinsics in clang/llvm.
> > 
> > You should add __builtin_ia32_vgf2p8affineinvqb_v64qi without masking so that it can be consistent.
> certainly
> so you propose introducing only non-masked variants on the llvm-side for gfni as a whole?
Yep that's what I'm proposing.


Repository:
  rL LLVM

https://reviews.llvm.org/D40373