Hardware Utilization of Models of Genetic Algorithms

Abstract

The paper compares the models of Genetic Algorithms (GA). It is focused on parallelized genetic algorithms, namely on the models of Master-Slave GA, Coarse-Grained GA, and Fine-Grained GA. The obtained results were compared with a serial model. A main memory usage comparison and a Central Processor Unit (CPU) comparison were performed. Furthermore, the parallelization module was verified on several workstations and the testing was evaluated. As expected, the serial model had the lowest main memory requirements. For parallelized models with increasing population size, main memory usage also increased.