A GPU/CPU Programming Model for CFD Simulation

Abstract:

In recent years, the highly parallel graphics processing unit (GPU) is rapidly gaining maturity as a powerful engine for high performance computer. More and more researchers try to port the computational fluid dynamics (CFD) simulations into heterogeneous computers. However, most researchers focus on exploring the computational capability of GPU, while ignore the computational capability of CPU. In order to utilize the computational capability of CPU and GPU, we propose a hybrid CUDA/OpenMP parallel programming model. And we proposed an adaptive load balancing scheme to distribute the workload among CPUs and GPUs. With this programming model, we implement a high-order CFD program on “Tianhe-1A” supercomputer system. The performance results validate the workload distribution scheme.

• Periodical: Advanced Materials Research (Volumes 712-715)
• Pages: 2538-2541
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.712-715.2538
• Online since: June 2013
• Authors: Cao Wei, Zheng Hua Wang, Chuan Fu Xu
• Keywords: CUDA, Graphic Processing Unit (GPU), Heterogeneous Computer, OpenMP

References

[1] NVIDIA: http://www.nvidia.com/CUDA. (2007)

[2] Yang, X.-J., Liao, X.-K., Lu, K., Hu, Q.-F., Song, J.-Q., Su, J.-S: J. Comput. Sci. Technol. Vol. 26(2011), pp.344-351

[3] Le, H., Cambier, J.L.: ArXiv e-prints (2011)

[4] Tutkun, B., Edis, F.O.: Comput. Fluids. Vol 55 (2012), pp.29-35

[5] Esfahanian, V., Darian, H.M., Gohari, S.I.: Comput. Fluids (2012). In press

[6] Geveler, M., Ribbrock, D., Goddeke, D., Zajac, P., Turek, S.: Comput. Fluids (2012). In press

[7] J.A. Martínez, E.M. Garzón, A. Plaza, etc.: J Supercomput. Vol 58 (2011), p.151–159