A Highly Scalable Solution of an NP-Complete Problem Using CUDA