high performance computing on graphics processing units: hgpu.org

Now high-throughput sequencing technologies can yield a huge volume of sequence data with affordable price, the explosion rate of sequence data is outpacing the performance improvement of CPU, which means trivial sequence analysis task may challenge existing serial programs running purely on CPU. This paper depicts a highly parallel sequence translation program running on a commodity multiprocessor GPU with compute unified device architecture (CUDA) platform. The parallel kernel code runs 10 times faster than any serial implementation on CPU. The experiment demonstrates that low-cost GPU solution will provide a new alternative for tackling a broad set of bioinformatics problems with high performance, low implementation cost and great scalability.