[PATCH] D69891: [VP,Integer,#1] Vector-predicated integer intrinsics

[Andy Kaylor via Phabricator via llvm-commits]

andrew.w.kaylor added a comment.

I know it's very late for me to be bringing this up, but I have a couple of broad questions -- one minor, one less minor.

First, when we introduced the constrained FP intrinsics Chandler suggested inserting ".experimental" in the names because of the likelihood that something would need to change before we were done and the "experimental" modifier would make the auto-upgrader handling slightly less ugly. That same reasoning seems like it would apply here.

Second, I'm not 100% comfortable with how the explicit vector length argument is represented. Yes, I know, I should have brought this up about a year ago. I don't have any objections to the existence of this argument. I understand that we need some way to handle it. I'm just wondering if we can handle it in some way that makes it more like an optional argument. I'm not entirely clear where this value comes from, but it seems like whatever generated it must know that we're generating SVE code, right? My concern is that for targets that don't do SVE kinds of things we still have this entirely redundant, if not meaningless, operand to carry around. What would you think of this being provided as an operand bundle when it is needed?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D69891/new/

https://reviews.llvm.org/D69891