A Parallel Genetic Algorithm for the Graphs Mapping Problem

The application of generic algorithm to the mapping of the graph of a parallel program on the structure of multiprocessor system is considered. The parallel algorithm based on simulation of evolutionary process and heuristics accelerating of usual genetic algorithm based on the local search on neighbourhood are offered. The results of some experiments of the algorithm conducted on multiprocessor system MBC-100 are presented. The proposed algorithms are included in the library of algorithms for Java-based system TOPAS for investigation, visualization and animation of mapping algorithms