[PATCH] D144413: [InstCombine] Extend SVEVectorFuseMulAddSub to support newly added "undef" intrinsics.

[Matt Devereau via Phabricator via llvm-commits]

MattDevereau added inline comments.


================
Comment at: llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp:1622-1625
+  case Intrinsic::aarch64_sve_fadd_u:
+    return instCombineSVEVectorFuseMulAddSub<Intrinsic::aarch64_sve_fmul_u,
+                                             Intrinsic::aarch64_sve_fmla_u>(
+        IC, II, true);
----------------
Is this case separate from `instCombineSVEVectorAdd` because the "undef" variant can't combine to fmla_u unless both the fmul and fadd are of the _u variants? Or because this case can't benefit from combines in `instCombineSVEVectorBinOp`?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D144413/new/

https://reviews.llvm.org/D144413