high performance computing on graphics processing units: hgpu.org

This paper introduces an Enhanced Multi Objective Genetic Algorithm (E-MOGA) running on Compute Unified Device Architecture (CUDA) hardware, as a general purpose tool that can solve conflict optimization problems. The tool demonstrates significant speed gains using affordable, scalable and commercially available hardware. The objectives of this research are: to enhance the general purpose Multi Objective Genetic Algorithm (MOGA) to be ready to execute on CUDA by parallelizing the time consuming parts, to test the performance of the enhanced MOGA on many testing cases, to test the quality of the result of the enhanced MOGA and to study the effect of the number of objectives on performance. The implemented system works under Matlab environment. It was tested on GPU GeForce GTX 9500 in CUDA platform, showed an average Speedup with more than 28X. The quality, measured using the error with respect to optimal solution, outperform the sequential MOGA.