Comparison of Parallel CUDA and OpenMP Implementations of Particle Swarm Optimization

Since the physical constraints on micro computing devices have forced the researchers to design next generation chips, the significance of the parallelization and distributed computing grow in importance. In this study, a sequential implementation of the Particle Swarm Optimization algorithm is converted into a concurrent version, which is executed on the cores of both CPU and GPU. For this reason, CUDA and OpenMP libraries are operated on the parallel algorithm to make a concurrent execution on CPU and GPU, respectively. The aim of this study is to compare CPU and GPU implementation of the PSO algorithm as regards the average execution time of independent Monte Carlo runs and computation architecture. For this purpose, nine benchmark functions are selected as test problems, and the parallel algorithm is executed on different processing units. The results show that parallel performance of the algorithm on different architectures outperforms to some application oriented computation units.