A DIFFERENTIAL EVOLUTION ALGORITHM PARALLEL IMPLEMENTATION IN A GPU

• Published in: Journal of Theoretical and Applied Information Technology Vol. 86; no. 2; p. 184
• Main Authors: Laguna-Sánchez, G A, Olguín-Carbajal, M, Cruz-Cortés, N, Barrón-Fernández, R, Martínez, R Cadena
• Format: Journal Article
• Language: English
• Published: Islamabad Journal of Theoretical and Applied Information 01.04.2016
• Subjects: Algorithms; Computation; Convergence; Evolutionary algorithms; Heuristic; Information technology; Optimization; Parallel programming
• ISSN: 1817-3195

The computational power of a Graphics Processing Unit (GPU), relative to a single CPU, presents a promising alternative to write parallel codes in an efficient and economical way. Differential Evolution (DE) algorithm is a global optimization based on bio-inspired heuristic. DE has a good performance, low computational complexity and need few parameters. This article presents parallel implementation of this population-based heuristic, implemented on a NVIDIA GPU device with multi-thread support and using CUDA as the model of parallel programming for these case. Our goal is to give some insights about GPU's parallel programming by a simple and almost straightforward parallel code, and compare the performance of DE algorithm running on a multithreading GPU. This work shows that with a parallel code and a NVIDIA GPU not only the execution time is reduced but also the convergence behavior to the global optimum may be changed in a significant manner with respect the original sequential code.