Figure 5From: MEGADOCK 3.0: a high-performance protein-protein interaction prediction software using hybrid parallel computing for petascale supercomputing environments Scalability of parallelization among nodes by MPI on (a) K computer (6144 to 24576 nodes), 220×220 dockings of FFT size = 140 protein pairs; (b) TSUBAME (100 to 400 nodes), 44x44 dockings of FFT size = 140 protein pairs. Back to article page