[PATCH] D129537: [X86][DAGISel] Don't widen shuffle element with AVX512

[LuoYuanke via Phabricator via llvm-commits]

LuoYuanke added inline comments.


================
Comment at: llvm/test/CodeGen/X86/avx512-shuffles/shuffle-blend.ll:94
 ; CHECK-NEXT:    vpaddb %xmm1, %xmm0, %xmm2
-; CHECK-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
-; CHECK-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],xmm2[1,2,3,4,5,6,7]
+; CHECK-NEXT:    movw $3, %ax
+; CHECK-NEXT:    kmovd %eax, %k1
----------------
RKSimon wrote:
> LuoYuanke wrote:
> > For this case not sure if it is worse than left side code (previous code).
> Add a 128-bit vector limit?
I'll add 128-bit vector limit. However in this case it is 128-bit vector.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D129537/new/

https://reviews.llvm.org/D129537