Figure 4From: MEGADOCK 3.0: a high-performance protein-protein interaction prediction software using hybrid parallel computing for petascale supercomputing environmentsScalability of thread parallelization using OpenMP on (a) K computer (8 cores/node) and (b) TSUBAME (12 cores/node, hyper threading enabled). 1ACB chain E and 1ACB chain I was used for docking. Elapsed time was measured from the mean of 30 docking processes. The right area of the dashed line shows speedup by activating hyper threading.Back to article page