Impact of New Sub- Classes of Hypercube Topology on Execution Time of Matrix Multiplication

Abstract

Today’s advanced research areas such as DNA computing, different branches of nanotechnology, immune cell system and optical computing require extensive data processing. Therefore, parallel processing systems with sophisticated hardware and software platforms are widely used. Furthermore, implementing the right algorithm which affects the overall execution time is a challenging task. This paper presents the principle of a massively parallel processing system based on Master-Slave Super-Super Hypercube 4-Cube (MS3H4-Cube) topology which could be easily implemented by using SGI products. Moreover, it is verified that the execution time of a matrix product is shorter when is applied on Master- Slave Super-Hypercube (MSSHP) compared with Hypercube (HP) topology.