Gpu-based personal supercomputing

The complexity of today´s applications has increased exponentially with the growing demand for unlimited computational resources. Cluster based supercomputing systems and high performance data centers have traditionally been the ideal assets to fulfill the ever increase in computing demands. Such computing resources require multi-millions dollars investment for building the system. The cost increases with the amount of power needed for operating and cooling the system along with the maintenance. In this paper we investigate the Personal SuperComputing (PSC) as an emerging concept in high performance computing that provides an opportunity to overcome most of aforementioned problems. We explore and evaluate the Gpu-based personal supercomputing system, and we compare it with conventional high performance cluster-based computing system. Our evaluations show promising opportunities in using GPU-based clusters for high performance computing applications.