Journal cover Journal topic
Geoscientific Model Development An interactive open-access journal of the European Geosciences Union
Geosci. Model Dev., 8, 2815-2827, 2015
https://doi.org/10.5194/gmd-8-2815-2015
© Author(s) 2015. This work is distributed under
the Creative Commons Attribution 3.0 License.
Development and technical paper
09 Sep 2015
POM.gpu-v1.0: a GPU-based Princeton Ocean Model
S. Xu1, X. Huang1, L.-Y. Oey2,3, F. Xu1, H. Fu1, Y. Zhang1, and G. Yang1 1Ministry of Education Key Laboratory for Earth System Modeling, Center for Earth System Science, Tsinghua University, 100084, and Joint Center for Global Change Studies, Beijing, 100875, China
2Institute of Hydrological & Oceanic Sciences, National Central University, Jhongli, Taiwan
3Program in Atmospheric & Oceanic Sciences, Princeton University, Princeton, New Jersey, USA
Abstract. Graphics processing units (GPUs) are an attractive solution in many scientific applications due to their high performance. However, most existing GPU conversions of climate models use GPUs for only a few computationally intensive regions. In the present study, we redesign the mpiPOM (a parallel version of the Princeton Ocean Model) with GPUs. Specifically, we first convert the model from its original Fortran form to a new Compute Unified Device Architecture C (CUDA-C) code, then we optimize the code on each of the GPUs, the communications between the GPUs, and the I / O between the GPUs and the central processing units (CPUs). We show that the performance of the new model on a workstation containing four GPUs is comparable to that on a powerful cluster with 408 standard CPU cores, and it reduces the energy consumption by a factor of 6.8.

Citation: Xu, S., Huang, X., Oey, L.-Y., Xu, F., Fu, H., Zhang, Y., and Yang, G.: POM.gpu-v1.0: a GPU-based Princeton Ocean Model, Geosci. Model Dev., 8, 2815-2827, https://doi.org/10.5194/gmd-8-2815-2015, 2015.
Publications Copernicus
Download
Short summary
In this paper, we redesign the mpiPOM with GPUs. Specifically, we first convert the model from its original Fortran form to a new CUDA-C version, POM.gpu-v1.0. Then we optimize the code on each of the GPUs, the communications between the GPUs, and the I/O between the GPUs and the CPUs. We show that the performance of the new model on a workstation containing 4 GPUs is comparable to that on a powerful cluster with 408 standard CPU cores, and it reduces the energy consumption by a factor of 6.8.
In this paper, we redesign the mpiPOM with GPUs. Specifically, we first convert the model from...
Share