Choosing Efficient Processor Configuration in Star Network via Simulation

I. Pozniak-Koszalka (Poland)


Simulation, star network, divisible load, knowledge-based decision making.


In the paper, the problem of parallel processor configuration design with taking into account processing and transmission costs is considered. An idea of knowledge-based decision making system for choosing such a configuration is presented. The choice of configuration consists in determining the number of processors engaged, their localization in a star network and partition workload among processors. The objective is to minimize the cost of performing the workload. The decision is to be made on the basis of knowledge acquitted with utilizing experimentation system. The proposed experimentation system allows investigating relations between network parameters and optimal decisions. The analysis of results of simulation made with using this system are presented.

Important Links:

Go Back