BOINC and CUDA: Distributed High-Performance Computing for Bioinformatics String Matching Problems

Publication Year: 
2010

Abstract:

Research has shown that applications ported to utilize Graphics card hardware typicially show at least an order of magnitude performance increase. This is typically achieved due to suitability and scale of the problems to these processors. String matching problems, prevalent in the bioinformatics field, have a low computation to data ratio and often a huge scale that limits them to classical supercomputers. We explore an option to try and scale classical GPGPU computing to large scales through the use of desktop grids. 

Institution: 
University of Witwatersrand/University of Johannesburg