jhuber6 wrote: > Hope the bot will be happy as well
I need to fix one thing unfortunately, there's this `--nvptx-lower-global-ctor-dtor` option that we pass for these 'direct' GPU compilations on NVPTX. That's defined in the NVPTX backend so if the user didn't build with it, then it is an error. It's an opt-in thing since CUDA can't handle it and they wanted it to remain an error. I'm thinking the easiest solution is to just define it somewhere in LLVM so it's always there. Don't know where a good spot would be though. https://github.com/llvm/llvm-project/pull/126143 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
