A distributed processing system for general purpose computing

Abstract

This paper presents a conceptual design of a distributed processing system aimed at providing general purpose computing and very high tolerances of failures while taking advantage of future large scale integration techniques. In addition it was desirable for the system to be easily expandable or contractable so that it could be flexibly applied to a variety of applications. The system was designed for a space-borne application; however the structure should have general applicability to a variety of systems. A method of analyzing the parallelism inherent in the computations to be carried out on the system is also presented. The distributed processor can thus be efficiently organized for the composition of general purpose (or special purpose) computations for a given application. Thought has also been directed toward the design of a system executive program and failure detection and reconfiguration methods; however, these features are still in development and only a brief discussion of the system executive is given in this paper.