A GPU-Accelerated Algorithm for Self-Organizing Maps in a Distributed Environment
Swedish School of Library and Information Science, University of Boras, Boras, Sweden
20th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN-12), 2012
@article{wittek2012gpu,
title={A GPU-Accelerated Algorithm for Self-Organizing Maps in a Distributed Environment},
author={Wittek, P. and Dar{‘a}nyi, S.},
year={2012}
}
In this paper we introduce a MapReduce-based implementation of self-organizing maps that performs compute-bound operations on distributed GPUs. The kernels are optimized to ensure coalesced memory access and effective use of shared memory. We have performed extensive tests of our algorithms on a cluster of eight nodes with two NVidia Tesla M2050 attached to each, and we achieve a 10x speedup for self-organizing maps over a distributed CPU algorithm.
May 10, 2012 by hgpu
