Optimizing High-Resolution Community Earth System Model on a Heterogeneous Many-Core Supercomputing Platform

Dr. Shaoqing Zhang
Ocean University of China
邀请人： 段晚锁 研究员
2020年7月9日（星期四）10:00

腾讯会议 (会议ID: 341 335 177)
会议链接: https://meeting.tencent.com/s/3J7mIQepR1Ni 
会议直播: https://meeting.tencent.com/l/wjzxilF1cPq2

报告摘要:        
With the semi-conductor technology approaching its physical and heat limits, supercomputers have adopted major architectural changes to increase the performance with more power-efficient heterogeneous many-core structure such as Sunway TaihuLight and Summit GPUs. Meanwhile, high-resolution Earth system models developed on traditional homogeneous multi-core systems desperately require more computing power, and refactoring and optimizing such models for new architectures become a key challenge in taking advantage of greener and faster supercomputers, better supporting global weather-climate research and predictions. We report the efforts of a large group in the International Laboratory for High-Resolution Earth System Prediction (iHESP) established by the cooperation of Qingdao Pilot National Laboratory for Marine Science and Technology (QNLM), Texas A & M University and the National Center for Atmospheric Research (NCAR), with the goal of enabling highly efficient simulations of the high-resolution (25-km atmosphere and 10-km ocean) Community Earth System Model (CESM-HR) on Sunway TaihuLight. The refactoring and optimizing makes the CESM-HR from 1 simulation years per day (SYPD) to 3.4 SYPD and completed over 500 hundred years of pi_ctl simulations. The success of CESM-HR on the Sunway system opens a door for higher-resolution Earth system modelling with utilization of greener heterogeneous many-core systems.

报告人简介:
Shaoqing Zhang, Ph.D. (Florida State University, USA, 2000), is a Distinguished Professor in Key Laboratory of Physical Oceanography, Ministry of Education, Ocean University of China (POL/OUC), an Outstanding Scientist in Qingdao Pilot National Laboratory for Marine Science and Technology (QNLM), and the Leading PI of High-Resolution Earth System Modeling and Prediction Initiative at QNLM. His major research interests include coupled Earth system modeling, seamless weather-climate predictability, coupled model data assimilation and parameter estimation. Currently one of his major missions is leading a national-wide large group to develop an ultra-high resolution earth system model on Chinese home-growth heterogeneous HPC systems.