The latency of user-to-user, kernel-to-kernel and
interrupt-to-interrupt level communication