(196a) Graphics Processing Units for High-Performance Computing in Bioinformatics

Bioinformatics emerged under the need to analyze proteomics and DNA data in the early 1980's. Although computers could deal efficiently with the amount of sequence data that was available until the 1990's, the amount of bioinformatics data currently available overwhelms existing computing power and algorithms for tasks such as protein alignment, motif finding, and structure calculations. Supercomputers can and have been used to reduce computing times in bioinformatics but their high cost makes them inaccessible to the average researcher.

The main goal of this work is to demonstrate that bioinformatics calculations stand to benefits tremendously from parallel algorithms executed on a GPU-enabled desktop workstation with over one TFLOPS of peak-performance and a cost under $3000. In particular, we will show how the scientific-computing-based C870 Tesla GPU [1] can be used on bioinformatics optimization algorithms, such as the Smith-Waterman dynamic programming algorithm for sequence alignment [2, 3], that lend themselves to parallelization. Although GPUs have been available for a couple of decades, they have been utilized extensively for general-purpose computing only in the last few years. This trend was boosted by the introduction of CUDA [4, 5] that facilitates the use of GPUs for parallel scientific calculations. We will demonstrate the software development process using CUDA, and the emerging challenges of parallelizing an algorithm to run efficiently on a GPU. Comparative studies and experimental performance results will demonstrate how users of bioinformatics software benefit from these emerging computing environments.

[1] www.nvidia.com

[2] M. Farrar, “Striped Smith-Waterman Speeds Database Searches Six Times over Other SIMD Implementations,” Bioinformatics, pp. 156–161, Jan. 2007.

[3] S. A Manavski and G. Valle, “CUDA Compatible GPU Cards as Efficient Hardware Accelerators for Smith-Waterman Sequence Alignment,” BMC Bioinformatics, March 2008.

[4] J. Nickolls, I. Buck, M. Garland, K. Skardon, “Scalable Parallel Programming with CUDA,” ACM Queue Magazine, pp. 41–53, March/April 2008.

[5] D. Blythe, “Rise of the Graphics Processor,” IEEE Proceedings, Vol. 96, No. 5, pp. 761–778, May 2008.