A CUDA-Based Multi-Channel Particle Swarm Algorithm

Abstract

In order to improve the efficiency and accuracy of the particle swarm optimization algorithm in optimization, the parallel CUDA-based particle swarm algorithm is proposed and developed. With the Compute Unified Device Architecture (CUDA) technology, the parallel data structure is defined, and the mechanism of computing tasks mapping to CUDA is described. From the optimization experiments results of 4 benchmark functions it shows that the CUDA-based parallel algorithm can greatly save computing time and improve computing accuracy. This new¿@PSO algorithm is more suitable for the relevant application of the particle swarm algorithm.