Performance Prediction Modelling of Multicomputers

In order to effectively program Multicomputers users must be able to evaluate how well the system performs for a given application. In this paper we present an efficient execution model that can be used for tree structured computations. We provide a general framework for analyzing the performance of this type of computation for any given topology. This framework is used to derive models for two widely used parallel programming strategies: Processor Farms and Divide and Conquer. These models were validated on a large multicomputer and the accuracy of the model is i,uch that they can be used to predict the performance of applications that use these strategies. We ctiscuss how these models can be used to evaluate performance and how they could be used to restructur the application to improve performance.