[llvm-dev] [FPEnv] FNEG instruction (original) (raw)

Sanjay Patel via llvm-dev llvm-dev at lists.llvm.org
Tue Oct 2 13:58:54 PDT 2018

On Tue, Oct 2, 2018 at 1:06 PM Cameron McInally <cameron.mcinally at nyu.edu> wrote:

On Mon, Oct 1, 2018 at 5:41 PM Sanjay Patel <spatel at rotateright.com> wrote:

I don't see any controversy for the preliminary requirement of removing BinaryOperator::isFNeg() and friends, so start with that? That work may reveal other potential regressions that we can patch in advance too.

This is true and I will agree to do this work... Other than that, I think there's really only a question of do we want 1 or both of fneg and fnegconstrained (and if we choose both, then I assume we'd also add fabsconstrained and copysignconstrained). but this is the real goal. Doing the BinaryOperator::isFNeg() work is in vain if we don't have at least a conditional approval of an explicit FNEG IR instruction. Would it be possible to obtain that conditional approval before work begins? That seems most prudent.

I can't speak for anyone else, but I approve for the same reasons that were mentioned early on: fneg makes the intended behavior and potential transforms easier to discern.

I don't have much knowledge of the problems/requirements in the constrained environment, so I'll leave the constrained counterpart decision to people who are working on that. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20181002/2d54ea29/attachment.html>

More information about the llvm-dev mailing list