Abstract—With the rapid development of the graphics processing unit (GPU), a recent GPU offers incredible resources for general purpose computing. We apply this technology to Monte Carlo simulations of the 2D and 3D lattice Ising models. By implementing the checkerboard algorithm, results are obtained up to 54, 62 and 68 times faster on the GPU than on a current CPU core for the honeycomb, square and triangular lattice respectively. For the 3D situation, however, the speedup is impacted greatly by the threads assignments on the GPU, the fastest one gives a speedup of 179, at the same time, the results of those simulations are consistent with the theoretical results for the 2D Ising model and previous results for the 3D Ising model.
Index Terms—GPU, CUDA, Ising model, monte carlo
Xing Lu, Jing Cai and Peidong Cui are postgraduate students with the Department of Physics, Jinan University, Guangzhou, P. R. China (e-mail: landstar@126.com, caijing164@163.com).
Wei Zhang is working as Assistance Professor with the Department of Physics, Jinan University, Guangzhou, P. R. China(email: twzhang@jnu.edu.cn)
Cite: Xing Lu, Jing Cai, Peidong Cui, and Wei Zhang, "Accelerating the Simulations of the Ising Model by the GPU under the CUDA Environment," International Journal of Modeling and Optimization vol. 1, no. 5, pp. 426-431, 2011.
Copyright © 2008-2024. International Journal of Modeling and Optimization. All rights reserved.
E-mail: ijmo@iacsitp.com