[PATCH] D53157: Teach the IRBuilder about constrained fadd and friends

[Ulrich Weigand via Phabricator via cfe-commits]

uweigand added a comment.

In https://reviews.llvm.org/D53157#1305233, @kpn wrote:

> In https://reviews.llvm.org/D53157#1304347, @uweigand wrote:
>
> > But given that there is still infrastructure missing in the IR optimizers, I also think that at least in the first implementation, we probably should go with the original approach and just use constrained intrinsics everywhere in the function, and possibly add some function attribute that prevent any cross-inlining of functions built with constrained intrinsics with functions built with regular floating-point operations.
>
>
> Subtle. This last sentence seems to imply that cross-inlining should be allowed when there are no regular floating point operations in the function to be inlined. This makes sense due to, for example, the common use of tiny functions just to retrieve a value. Do I interpret your statement correctly?


Sure, that would be an optimization.  Another potential optimization would be to **allow** inlining, but have the inliner automatically replace regular floating-point operations with constrained intrinsics in the target routine ...   But all of that is probably "stage2" after an initial implementation that just disables inlining.


https://reviews.llvm.org/D53157