Effect of parallel distributions system on computer efficiencyand performance

Abstract EN

In this paper, the distribution parallel system tool MPI (Message Passing Interface) to solve squared matrixes product was used.

In order to make parallel tasks, the work will be implemented at least by using two computers.

Also, several tasks will be included in this project to estimate the Speedup of the processor (p), Efficiency of the processor (p), fraction f of sequential work and fraction of parallel part of the work using Amdahl’s law.

Moreover, this paper will estimate many approaches which are: running the application on 1 processor for 3 different values of matrix elements N (for example 1000 ×1000, 2000 × 2000, and 3000 × 3000), running the application on 2, 3 or more processors for the same values of N, estimating fraction f of sequential part, estimate f for 3 different values of N by having measured times, finding out whether this fraction depends on the problem size N, and more operation to complete the task required.