[PATCH] D108539: [X86][AVX] Attempt to fold a scaled index into a gather/scatter scale immediate (PR13310)

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added a comment.

I'm going to see whether this would be better moving to matchVectorAddressRecursively now that D111595 <https://reviews.llvm.org/D111595> has landed.



================
Comment at: llvm/test/CodeGen/X86/masked_gather_scatter.ll:774
 ; KNL_64-NEXT:    kxnorw %k0, %k0, %k1
-; KNL_64-NEXT:    vgatherqps (%rax,%zmm0), %ymm1 {%k1}
+; KNL_64-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; KNL_64-NEXT:    vgatherqps (%rax,%zmm0,4), %ymm1 {%k1}
----------------
pengfei wrote:
> Why do we need an extra xor now?
D112505 - for dependency breaking I guess :) 


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D108539/new/

https://reviews.llvm.org/D108539