[PATCH] D146518: [AArch64] Extend icmp bitcast to vecreduce fold to comparison with -1

[Nikita Popov via Phabricator via llvm-commits]

nikic added inline comments.


================
Comment at: llvm/lib/Target/AArch64/AArch64ISelLowering.cpp:20438
+                        LHS);
       return DAG.getSetCC(DL, VT, LHS, RHS, Cond);
     }
----------------
I found the formulation here is a bit roundabout -- is there a reason why we can't directly return the vecreduce result (together with a NOT, depending on predicate)?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D146518/new/

https://reviews.llvm.org/D146518