AMD Radeon GCN Offloading Support For OpenMP/OpenACC On The Way For GCC 10

Merged for the GCC 9 compiler release that launched earlier this year was the preliminary AMD Radeon "GCN" GPU compiler back-end. In that initial release it wasn't particularly useful as the GPU offloading bits for the popular programming APIs/models wasn't supported so for now could just run some basic single-threaded programs. But now those interesting GPU offloading bits are pending for GCC 10.

Julian Brown of CodeSourcery, the company contracted by AMD to work on this GCC compiler offloading, has got around to prepping the patches for the mainline kernel. Up to now these patches were sitting within their internal tree.

The set of eight now-published patches provide basic GPU offloading support for OpenMP and OpenACC for the AMD GCN target. Though there's a warning the OpenACC code in particular is likely to be buggy at this point but fixes will be coming up.

The patches are around eight thousand lines of new code for the GNU Compiler Collection. That offloading support should be merged soon to Git/SVN while hopefully for next year's GCC 10 stable release (GCC 10.1) there will be good Radeon OpenMP/OpenACC support in order and mature.