Searching Genetic Databases on Splash 2

Hoang’s paper demonstrated the absolutely massive speedup potential present in FPGA acceleration for DNA sequence matching, showing that a single Splash 2 board (consisting of 17 Xilinx 4010 FPGAs) should be two orders of magnitude faster than a MasPar MP-1, a SIMD supercomputer uniquely suited to this problem. A full 16-board Splash 2 configuration would be sixteen times faster. Comparisons with a general-purpose computer are even more favorable, showing an expected four orders of magnitude improvement with a 16 board Splash 2 compared with a typical SPARC. This paper helped create a thriving industry producing custom FPGA solutions for bioinformatics, an industry that continues to demonstrate substantial benefits in compute per dollar and massive benefits in compute per Watt.