Figure 5
From: High-throughput sequence alignment using Graphics Processing Units
Suffix Tree layout. The nodes of the suffix tree are rearranged into cache blocks to optimize 2D locality. Near the root of the tree, nodes of the same depth are placed into the same "wide" block. Futher down the tree, nodes from the same subtree are placed into the same "tall" block. MUMmerGPU uses blocks of 32 × 32 nodes, but for clarity, 8 nodes cache blocks are displayed here.