https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100573
--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> --- I didn't have the nvidia binary module loaded and cuda installed when doing the light testing, now I've installed that and see FAIL: libgomp.c/../libgomp.c-c++-common/for-3.c execution test FAIL: libgomp.c/../libgomp.c-c++-common/for-9.c execution test XPASS: libgomp.c/../libgomp.c-c++-common/pr96390.c (test for excess errors) FAIL: libgomp.c/../libgomp.c-c++-common/pr96390.c execution test FAIL: libgomp.c/../libgomp.c-c++-common/target-41.c execution test FAIL: libgomp.c/../libgomp.c-c++-common/target-42.c execution test fail. target-41.c and -42.c FAIL with the same error as for-3.c, libgomp: cuLaunchKernel error: too many resources requested for launch I'm puzzled about that message though, it really shouldn't request too many resources, it should spawn a single thread doing a very simple kernel. Maybe the __builtin_unreachable (); calls are the culprit? I didn't know if I should use __builtin_trap (), __builtin_abort () and __builtin_unreachable () is what has been used in task.c.