Parallel matrix multiplication analysis / Hazrin Mohammad

• Main Author: Mohammad, Hazrin
• Format: Thesis
• Language: English
• Published: 2007
• Subjects: Electronic Computers. Computer Science
• Online Access: https://ir.uitm.edu.my/id/eprint/803/2/803.pdf; https://ir.uitm.edu.my/id/eprint/803/

The first computer was meant to solve equations sequentially and was programmed to do sequential equations. However, as computations are getting more complex and the demand of having more powerful processors doubles, it seems like the move to solve problems and equations through parallelization is the best solution we have in the world of computing today. Therefore, various parallel computing solutions had been introduced in terms of hardware and software. Since then, computers are getting more advanced than they were before thus enabling software developers and programmers to produce with vivid quality software’s, real time simulations, and more advanced modeling’s. However, the cost of having a parallel supercomputer is way beyond the limits or budget restrains of some small and medium organizations. Therefore, a low cost parallel computing solution had been introduced by developers which involve a few personal computers or PC’s networked together to imitate the producing power of a supercomputer. This research project was conducted to compare the time difference between the sample parallel matrix calculation programming and the sequential matrix calculation programming that was developed by the author in C language. The parallel matrix multiplication program will be run using 3 Ubuntu Linux PC’s connected together with communications protocol, SSH, been configured using RSA key via 100Mbps Ethernet either through the peer-to-peer connections and star topology connections and it is expected that the parallel matrix multiplication programming will run faster than the sequential matrix multiplication programming.