Effects of Parallel Processing Implementation on Balanced Load-Division Depending on Distributed Memory Systems

Abstract

Complex problems need long time to be solved, with low efficiency and performance. Therefore, to overcome these drawbacks, the studies went toward the approaches of breaking the problem into independent parts, and treating each part individually in the way that each processing element can execute its part of the problem simultaneously with the others.Parallel processors are computer systems that consist of multiple processing units connected via some interconnection network and the software needed to make the processing units work together. Parallel processing is divided into three types; Shared, Distributed and Hybrid memory systems.In this paper, distributed memory systems addressed depending on client/servers principles, the network can contain any number of nodes; one of them is a client and the others are servers. The algorithms used here are capable of calculating the (Started, Terminated, Consumed -CPU and Total Execution- times and CPU usage) of servers and the Client's -CPU and total execution- times. This work addresses an improved approach for problem subdivision in balanced form and design flexible algorithms to communicate efficiently between client-side and servers-side in the way to overcome the problems of hardware networking components and message passing problems. We addressed Matrix-Algebra case-study to display the effect of balance loaddivision for this approach. The obtained results are checked and monitored by special programming-checkingsubroutines through many testing-iterations and proved a high degree of accuracy. All of these algorithms implemented using Java Language