https://github.com/jdoerfert approved this pull request.
> We can't expect to have regular OpenMP code working in the same way as OpenMP > offloading code when targeting a GPU meanwhile the code is not wrapped into > target region or declare target The way I see this is: If the target is a GPU, and OpenMP is enabled (-fopenmp), what we should do is codegen it targeting the GPU OpenMP runtime. This is what this patch does and I don't think it violates anything. What we define as the *internal* OpenMP runtime could reasonably be different on every target, or even depending on the day of the week ;). No runtime would also be an option if we codegen all the semantics, though we really don't want that. That said, this makes sense to me. LG. P.S. I'd suggest discussing the device runtime in the actual follow up. https://github.com/llvm/llvm-project/pull/122149 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
