[PATCH] D71432: [AArch64][SVE] Proposal to use op+select to match scalable predicated operations

[Eli Friedman via Phabricator via llvm-commits]

efriedma added a comment.

> My current understanding (and I could be wrong) is that the native predication intrinsics and the constrained intrinsics will be merge into one set

My understanding is that we're going to get masked constrained intrinsics, yes.  Not sure if that means the other versions are going away completely; we still probably want some way to represent vector operations on targets without predication, at least in SelectionDAG.

In the meantime, we have to generate all the exceptions implied by a strict vector operation.  We can't change the exceptions based on the uses of the instruction.  Consider the following:

  `
  for(...)
    double c = a[i]/b[i];
    if (b[i] != 0)
      a[i] = c;
  `


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D71432/new/

https://reviews.llvm.org/D71432