Automatic Pipeline Construction Focused on Similarity of Rate Law Functions for an FPGA-based Biochemical Simulator
Hideki Yamada, Yui Ogawa, Tomonori Ooya, Tomoya Ishimori, Yasunori Osana, Masato Yoshimi, Yuri Nishikawa, Akira Funahashi, Noriko F. Hiroi, Hideharu Amano, Yuichiro Shibata, Kiyoshi Oguri · IPSJ Transactions on System and LSI Design Methodology · 2010
For FPGA-based scientific simulation systems, hardware design technique that can reduce required amount of hardware resources is a key issue, since the size of simulation target is often limited by the size of the FPGA. Focusing on FPGA-based biochemical simulation, this paper proposes hardware design methodology which finds and combines common datapath for similar rate law functions appeared in simulation target models, so as to generate area-effective pipelined hardware modules. In addition, similarity-based clustering techniques of rate law functions are also presented in order to alleviate negative effects on performance for combined pipelines. Empirical evaluation with a practical biochemical model reveals that our method enables the simulation with 66% of the original hardware resources at a reasonable cost of 20% performance overhead.