Figure 2: A better practical partitioning of the network for multithreaded SLPA.