AlexMaclean wrote: @Artem-B / @jhuber6 You're both correct. This change won't work as written due to the PTX and SM version restrictions clang and NVPTX enforce on the builtin and intrinsic. It still seems like a bit of a hack to use inline assembly to circumvent these checks. Do either of you have ideas for how this could be made to work? maybe with `__has_builtin` instead of `__nvvm_reflect`?
https://github.com/llvm/llvm-project/pull/198815 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
