Hi Joseph, Josef Melcr wrote:
Thank you for the detailed explanation :) It should be fairly easy to implement if ipa-cp acts before the tables are dumped. Otherwise it would probably require a bit of work, but if we implement some reorganizing logic for the tables, then we might be able to drop the noclone attribute altogether (if I understand it correctly). I will tinker around with it in the future, unfortunately I can't get the output_offload_tables to run, so I will have to figure that out. But the application sounds really exciting.
I think you only get this when you build with offloading support – either for Nvidia GPUs (nvptx) or AMD GPUs (gcn) or for both. https://gcc.gnu.org/wiki/Offloading#Building_and_Obtaining_GCC describes how you do this. However, the simplest should be the following: Only compiled the host compiler but configure with (pick one or both, as you like): --enable-offload-targets=nvptx-none,amdgcn-amdhsa --enable-offload-defaulted The former sets ENABLE_OFFLOADING to true and enabled in the driver support for the listed offload driver. The latter permits that the offloading compiler is not actually installed - that makes sense for distros, where to make it optional for the user to install it – but it also permits here to test it without actually having to compile and install an offloading compiler. Tobias
