measure of core load

Is there any way to see which cores of gpu work (the measure of load of each cuda core)?