Accelerating the Smith-Waterman Algorithm for Bio-sequence Matching on GPU
Electronic Engineering College, Naval University of Engineering, Wuhan, P. R. China, 430033
The 2012 International Conference on Bioinformatics and Computational Biology (BIOCOMP’12), 2012
@article{zhu2012accelerating,
title={Accelerating the Smith-Waterman Algorithm for Bio-sequence Matching on GPU},
author={Zhu, Q. and Xia, F. and Jin, G.},
year={2012}
}
Nowadays, GPU has emerged as one promising computing platform to accelerate bio-sequence analysis applications by exploiting all kinds of parallel optimization strategies. In this paper, we take a well-known algorithm in the field of pair-wise sequence alignment and database searching, the Smith-Waterman (S-W) algorithm as an example, and demonstrate approaches that fully exploit its performance potentials on GPU platform. We propose the combination of coalesced global memory accesses, shared memory tiles, and loop unfolding, achieving 50X speedups over initial S-W versions on a NVIDIA GeForce GTX 470 card. Experimental results also show that the GPU GTX 470 gains 12X speedups, instead of 100X reported by some studies, over Intel quad core CPU Q9400, under the same manufacturing technology and both with fully optimized schemes.
September 16, 2012 by hgpu
