Large-Scale Modeling of Epileptic Seizures: Scaling Properties of Two Parallel Neuronal Network Simulation Algorithms

Figure 5

Run time (wall clock time here, not total time as in the previous plot) for a 100 K-cell network on 9 nodes (216 cores) of Beagle. The number of MPI processes was varied while the number of OpenMP threads was adjusted so that each run used all 216 available cores. The leftmost point corresponds to 3 MPI processes per node with 8 OpenMP threads per process.