Debunking the 100X GPU vs. CPU myth: an evaluation of throughput computing on CPU and GPU