Abstract
*† Modern GPUs have proven to outperform the number of floating point operations when compared to CPUs through inherent data parallel architecture and higher bandwidth capabilities. The advent of programmable graphics hardware in recent times further provided a suitable platform for scientific computing, particularly in the field of population based optimization algorithms. Previous work by the authors demonstrated the capability of digital pheromones within PSO for searching n-dimensional design spaces with improved accuracy, efficiency and reliability in both serial and parallel CPU computing environments. Preliminary GPU implementations were also made using OpenCL and OpenGL Shading Languge by off-loading objective function evaluations to the GPU. In this paper, hardware acceleration of Particle Swarm Optimization with digital pheromones using the CUDA architecture on commodity Graphics Processing Unit (GPU) is investigated and presented. Specifically, two objectives will be attained in this work: 1) a successful implementation of PSO with digital pheromones on a high-end workstation GPU and a popular low-cost consumer level GPU and 2) results comparison between these implementations. Based on testing the algorithm with a number of unconstrained problems, recommendations will be made on the suitability of high-end or consumer level GPUs for solving population based optimization problems.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call