Accurate Profiling and Acceleration Evaluation of the Smith-Waterman Algorithm Using the Molen Platform
In this paper, the authors present an accurate method to evaluate the amount of acceleration gained by the hardware implementation of the Smith-Waterman (SW) algorithm. This is done using the MOLEN processor prototype (MOLEN platform), where algorithms can be executed both as software or as reconfigurable hardware. By profiling the algorithm, they identify a specific function that consumes 78% of the total runtime. Implementing this function in hardware results in a speedup of 2.16 in comparison with a software-only implementation. Since the hardware footprint needed for this implementation is rather small, this speedup is highly efficient in terms of resource utilization.