I don't want to read the result from the GPGPU mem, but only want to know if the calculation has reach a point or not in the kernel. Can I control the kernel to return an specified value? Or get an event about the status after the kernel execute finished? Can anyone give me suggestions on it?