Gpu Particle Swarm Optimization Applied To Travelling Salesman Problem

Recently, the Graphic Processing Unit (GPUs) are used as an exciting new hardware environment for truly parallel implementation and execution of nature and Bio-inspired algorithms thanks to their excellent price-to-power ratio. Indeed, they are represented by the software platform using compute unified device architecture from NVIDIA, and the one of particle swarm optimization (PSO) which can be executed simultaneously on GPUs to speed up complex optimization problems such as Travelling Salesman Problem (TSP). In this paper, we illustrate a novel parallel approach to run standard particle swarm optimization PSO on GPUs and applied to TSP (GPU-PSO-A-TSP). Both the developed and the previous PSO centroid algorithm are implemented on the GPUs. The achieved results show that we have obtained at least one order of magnitude difference between speed of the GPUs and a typical sequential CPU implementation for performance optimization. Results show also that running speed of GPU-PSO is four times as fast as that of CPUP-SO.