Solving large processor configuration problems with the guided genetic algorithm

Abstract

The Processor Configuration Problem (PCP) is an NP hard real life problem. The goal involves designing a network of a finite set of processors, such that the maximum distance between any two processors that a parcel of data needs to travel is kept to a minimum. Since each processor has a limited number of communication channels, a carefully designed layout would assist in reducing the overhead for message switching in the entire network. The Guided Genetic Algorithm (GGA) is a hybrid of genetic algorithm and meta heuristic search algorithm: Guided Local Search. As the search progresses, GGA modifies both the fitness function and fitness templates of the candidate solutions based on feedback from the constraints. We are interested in generating processor configurations between eight and 128 processors. GGA is used as a tool to generate these configurations, and is shown to have considerable advantages over published results.