Challenge toward the prediction of typhoon behaviour and down pour

24th IUPAP Conference on Computational Physics

物理学;计算机科学

Takahashi, K.^1 ; Onishi, R.^1 ; Baba, Y.^1 ; Kida, S.^1 ; Matsuda, K.^1 ; Goto, K.^2 ; Fuchigami, H.^3

Earth Simulator Center, JAMSTC, 3173-25 Showa-machi, Kanazawa-ku, 236-0001 Yokohama, Japan^1

NEC Corporation, 1-10 Nisshin-cho, Fuchu-shi, 183-5801, Tokyo, Japan^2

NEC Informatec Systems LTD, 3-2-1 Sakato, Takatsu-ku, Kawasaki-shi, Kanagawa, Japan^3

关键词: Atmosphere-ocean modeling;    Code optimization;    Core performance;    Horizontal resolution;    Multiphysics phenomena;    Peak performance;    Performance evaluation metrics;    System capabilities;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/454/1/012072/pdf
DOI  :  10.1088/1742-6596/454/1/012072

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

Mechanisms of interactions among different scale phenomena play important roles for forecasting of weather and climate. Multi-scale Simulator for the Geoenvironment (MSSG), which deals with multi-scale multi-physics phenomena, is a coupled non-hydrostatic atmosphere-ocean model designed to be run efficiently on the Earth Simulator. We present simulation results with the world-highest 1.9km horizontal resolution for the entire globe and regional heavy rain with 1km horizontal resolution and 5m horizontal/vertical resolution for urban area simulation. To gain high performance by exploiting the system capabilities, we propose novel performance evaluation metrics introduced in previous studies that incorporate the effects of the data caching mechanism between CPU and memory. With a useful code optimization guideline based on such metrics, we demonstrate that MSSG can achieve an excellent peak performance ratio of 32.2% on the Earth Simulator with the single-core performance found to be a key to a reduced time-to-solution.