https://hgpu.org/?p=4190
Highly efficient mapping of the Smith-Waterman algorithm on CUDA-compatible GPUs