Networked or distributed environment of workstation has been used mainly for communication purposes such as file services, database sharing and electronic mailing. In this thesis the network of workstations is viewed a huge parallel computer in which each workstation acts as a single processing element. Distributed systems consisting of powerful workstations connected with a high-speed network represent a large amount of computing power and parallel processing capcability, especially when many workstations are idling most of the time. Limited system software support, however, requires an application user to be an expert in system programming and parallel processing theory to execute his codes on the network of workstations.
We present a computational model and software environment called the distributed scientific computing(DSC) system for scientific applications. The DSC system aimes at providing a framework that liberates users from the low-level communication programming while achiveving high computational speed through the use of multiple workstations. More specifically, DSC provides an environment for parallel programming, execution and scheduling. We test the DSC system for two computationally intensive applications: the computation of CGH(Computer Generated Hologram) and simulation of a telecommunication problem. The obtained results on their speed-up are quite encouraging.
