An In-Storage Implementation of Smith-Waterman in Resistive CAM

An in-storage implementation of the Smith-Waterman sequence alignment algorithm on a resistive content addressable memory (ReCAM) based storage is proposed and is simulated and compared to state-of-the-art systolic arrays and GPU-based solutions. An in-storage implementation of the Smith-Waterman sequence alignment algorithm on a resistive content addressable memory (ReCAM) based storage is proposed. The ReCAM native compare operation is used to find matching basepairs in a fixed number of cycles, regardless of the sequence length. Our in-storage implementation is simulated and compared to state-of-the-art systolic arrays and GPU-based solutions. We show a 3.4x higher throughput and an order of magnitude lower