Accession Number : ADA116762
Title : Models and Measurements of Parallelism for a Distributed Computer System.
Descriptive Note : Master's thesis,
Corporate Author : CONNECTICUT UNIV STORRS LAB FOR COMPUTER SCIENCE RESEARCH
Personal Author(s) : Lane,Debra S
PDF Url : ADA116762
Report Date : Jan 1982
Pagination or Media Count : 81
Abstract : This thesis performed two tasks: the formulation of a model of parallelism for a distributed computer system, and the instrumentation of that system to validate the model. The model of parallelism is a Markov process, where the states of the model correspond to the states of the system. A model exists for two and three processors. The two processor model has four states and the three processor model has eight states. It has been determined that a four processor model would have nineteen states and a five processor model, twenty-nine states. These models were not formulated due to the large number of states and the results of validating the two and three processor models indicated that not much information could be gained from them. Validation of the two and three processor models revealed that they weren't very accurate. The relative error was 20% and less for two processors and 30% and less (with the exception of one state whose error was 55%) for three processors.
Descriptors : *Parallel processing, *Distributed data processing, *Message processing, Programming languages, Optimization, Decentralization, Algorithms, Synchronization(Electronics), Trade off analysis, Performance tests, Markov processes, Multiprocessors, Theses
Subject Categories : Computer Programming and Software
Distribution Statement : APPROVED FOR PUBLIC RELEASE