https://hgpu.org/?p=10830
Communication Optimization for Multi GPU Implementation of Smith-Waterman Algorithm