Problems with dynamic parallelism in Ubuntu 14.04 and CUDA 6.5

are you doing proper cuda error checking on all cuda API calls and kernel calls?