Meeting Banner
Abstract #4480

A GPU-based parallel computing framework for accelerating graph theoretical analyses

Tsang-Chu Yu 1 , Yi-Ping Chao 1 , Li-Wei Kuo 2 , Chung-Chih Lin 1 , Shih-Yen Lin 2,3 , Hengtai Jan 2 , Claudia Metzler-Baddeley 4 , and Derek Jones 4

1 Department of Computer Science and Information Engineering, Chang Gung University, Taoyuan, Taiwan, 2 Institute of Biomedical Engineering and Nanomedicine, National Health Research Institutes, Miaoli, Taiwan, 3 Department of Computer Science, National Chiao Tung University, Hsinchu, Taiwan, 4 School of Psychology, Cardiff University, Cardiff, United Kingdom

The main purpose of this study is to develop a graphics processing unit based framework for brain network analysis that permit networks comprising much larger numbers of nodes and provide the acceleration for processing. From the results, our implementation for the calculation of all pair shortest paths could reduce half of time with brain connectivity toolbox (BCT) and 638x speedup with Gretna in simulation random network with larger number of nodes (>8k). Moreover, our algorithm also shows better performance in human brain data with 1.37x and 21x speedup in comparison with BCT and Gretna respectively.

This abstract and the presentation materials are available to members only; a login is required.

Join Here