The GPU Enhanced Parallel Computing for Large Scale Data Clustering
Oak Ridge National Laboratory, Oak Ridge, TN 37831
International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery (CyberC), 2011
@article{cui2011gpu,
title={The GPU Enhanced Parallel Computing for Large Scale Data Clustering},
author={Cui, X. and Beaver, J. and Charles, J.S. and Potok, T.},
year={2011}
}
Analyzing and clustering large scale data set is a complex problem. One explored method of solving this problem borrows from nature, imitating the flocking behavior of birds. One limitation of this method of data clustering is its complexity O(n^2). As the number of data and feature dimensions grows, it becomes increasingly difficult to generate results in a reasonable amount of time. In the last few years, the graphics processing unit (GPU) has received attention for its ability to solve highly-parallel and semi-parallel problems much faster than the traditional sequential processor. In this chapter, we have conducted research to exploit this architecture and apply its strengths to the flocking based data clustering problem. Using the CUDA platform from NVIDIA, we developed a Multiple Species Data Flocking implementation to be run on the NVIDIA GPU. Performance gains ranged from 30 to 60 times improvement of the GPU over the CPU implementation.
January 24, 2012 by hgpu
