Abstract: Multi-GPU-Based Calculation of Percolation Problem on the TSUBAME 2.0 Supercomputer

We present the fast multi-GPU-based calculation of percolation problem on the two-dimensional (2D) square lattice for the multiple GPUs on the large-scale open science supercomputer TSUBAME 2.0. Recently, we presented the multiple GPU computing with the common unified device architecture (CUDA) for the cluster labeling. In this paper, we adapt this cluster labeling algorithm to the percolation problem. In addition, we modify this cluster labeling algorithm in order to simplify the analysis for the percolation. As a result, we realized the large scale and rapid calculations without a decay of computational speed on the analysis for the percolation, and the calculation time for the 2D bond percolation with L=65536 is obtained as 180 milliseconds per a single realization.