jhuber6 wrote: > Ok, but that still doesn't change the fact that the Clang driver will search > for a system-wide CUDA installation unless passed `--cuda-path`...
We also do this with the GCC toolchain, the issue is whether or not there's an error if it didn't find it. Doing `clang -v` will always print out if it found CUDA or HIP even on standard compilation AFAIK. Could you go ahead and add `-nogpulib -nogpuinc` to your tests. You can keep the `--cuda-path=` for now since I don't expect you to wait until my other fix lands. Thanks for bringing that to my attention, I knew there was something off if you were having problems. https://github.com/llvm/llvm-project/pull/84008 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
