https://hgpu.org/?p=6170
Improving CUDASW++, a Parallelization of Smith-Waterman for CUDA Enabled Devices