Fig. 5
From: Parallel algorithms for large-scale biological sequence alignment on Xeon-Phi based clusters
Xeon Phi vectorized implementation of pairwise alignment according to Definition 2 by dynamic programming using 25 core instructions. The variables in these instructions can be divided into two classes. One class includes v H A , vE, vF, and vS which are used in the Smith-Waterman algorithm. Another class contains v N A , v N E , v N F and v N S which are defined in Definition 2. Here v N A is the target vector and v N S is the value n i d(S i ,S j )