A Massively Parallel Architecture for Bioinformatics

Abstract. Today’s general purpose computers lack in meeting the requirements on computing performance for standard applications in bioinformatics like DNA sequence alignment, error correction for assembly, or TFBS finding. The size of DNA sequence databases doubles twice a year. On the other hand the advance in computing performance per unit cost only doubles every 2 years. Hence, ingenious approaches have been developed for putting this discrepancy in perspective by use of special purpose computing architectures like ASICs, GPUs, multicore CPUs or CPU Clusters. These approaches suffer either from being too application specific (ASIC and GPU) or too general (CPU-Cluster and multicore CPUs). An alternative is the FPGA, which outperforms the solutions mentioned above in case of bioinformatic applications with respect to cost and power efficiency, flexibility and communication bandwidths. For making maximal use of the advantages, a new massively parallel architecture consisting of l...

• Type: Conference
• Year: 2009
• Where: ICCS
• Authors: Gerd Pfeiffer, Stefan Baumgart, Jan Schröder, Manfred Schimmler
• Keywords: Applied Computing, Computing Performance, DNA Sequence, ICCS 2009, Multicore Cpus