Computational performance of WRF on Earth Simulator

Earth Simulator Center
Information Systems Department

1. Introduction

WRF (Weather Research and Forecasting Model) is a mesoscale meteorological simulation code which has been developed under the collaboration among US institutions, including NCAR (National Center for Atmospheric Research) and NCEP (National Centers for Environmental Prediction). JAMSTEC has optimized WRFV2 on the Earth Simulator (ES2) renewed in 2009 with the measurement of computational performance. As a result, we successfully demonstrated that WRFV2 can run on the ES2 with outstanding performance and the sustained performance.

2. Computational performance

The numerical meteorological simulation was conducted by using WRF on the Earth Simulator for the earth's hemisphere with the Nature Run model condition [1]. The model spatial resolution is 4486 by 4486 horizontally with the grid spacing of 5km and 101 levels vertically. Mostly adiabatic conditions were applied with the time integration step of 6 seconds.

3. Conclusions

We achieved a very high performance on the Earth Simulator for high-resolution WRF. While the number of CPU cores used is only 1% as compared to the world fastest class system Jaguar (CRAY XT5) at Oak Ridge National Laboratory, the sustained performance obtained on the Earth Simulator is almost 50% of that measured on the Jaguar system. The peak performance ratio on the Earth Simulator is also record-high 22.2%.

"Earth Simulator"

We appreciate NEC assisting the code porting and optimization of WRF.

Reference :

J. Michalakes, J. Hacker, M. O. McCracken, R. Loft, and A. Snavely et al. WRF nature run.
In Proceedings of the 2007 ACM/IEEE conference on Supercomputing, Nov. 2007.