Abstract
This paper studies the impact of varying the population's size and the problem's dimensionality in a parallel implementation, for an NVIDIA GPU, of a canonical GA. The results show that there is an effective gain in the data parallel model provided by modern GPU's and enhanced by high level languages such as OpenCL. In the reported experiments it was possible to obtain a speedup higher than 140 thousand times for a population's size of 262 144 individuals. © 2011 Springer-Verlag.
Author supplied keywords
Cite
CITATION STYLE
Prata, P., Fazendeiro, P., & Sequeira, P. (2011). Towards cost-effective bio-inspired optimization: A prospective study on the GPU architecture. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7077 LNCS, pp. 63–70). https://doi.org/10.1007/978-3-642-27242-4_8
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
