Compiler generated systolic arrays for wavefront algorithm acceleration on FPGAs

Wavefront algorithms, such as the Smith-Waterman algorithm, are commonly used in bioinformatics for exact local and global sequence alignment. These algorithms are highly computationally intensive and are therefore excellent candidates for FPGA-based code acceleration. However, there is no standard form of these algorithms, they are used in a wide variety of situations with various constraints. It is therefore not practical to have a standard kernel that can be mapped to an FPGA, hence the importance of being able to compile such codes from a high level language. ROCCC is a C to VHDL compiler, which optimizes and parallelizes the most frequently executed kernel loops in applications such as in multimedia, scientific and high-performance computing. In this paper we describe the transformations performed by ROCCC, which transformed the kernel of the Smith-Waterman algorithm into a hardware systolic array that is mapped onto the FPGA on the SGI Altix RASC blade. We report a throughput increase by over 3,000times over a 2.8 GHz Xeon.