[PATCH] D69493: Add -fconvergent-functions flag

[Artem Belevich via Phabricator via cfe-commits]

tra accepted this revision.
tra added a comment.
This revision is now accepted and ready to land.

LGTM.

> The CUDA builtin library is apparently compiled in C++ mode, so the
>  assumption of convergent needs to be made in a typically non-SPMD
>  language.

I think the key here is that we sometimes may need to compile things for device **without** `-x cuda` and that changes the default assumption about convergence. This flag allows controlling the assumption explicitly without relying on input language.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D69493/new/

https://reviews.llvm.org/D69493