Abstract
The Global ‐ Regional Integrated forecast SysTem (GRIST) is the next- generation weather and climate integrated model dynamic framework developed by Chinese Academy of Meteorological Sciences. In this paper, we present several changes made to the global nonhydrostatic dynamical (GND) core, which is part of the ongoing prototype of GRIST. The changes leveraging MPI and PnetCDF techniques were targeted at the parallelization and performance optimization to the original serial GND core. Meanwhile, some sophisticated data structures and interfaces were designed to adjust flexibly the size of boundary and halo domains according to the variable accuracy in parallel context. In addition, the I/O performance of PnetCDF decreases as the number of MPI processes increases in our experimental environment. Especially when the number exceeds 6000, it caused system-wide outages (SWO). Thus, a grouping solution was proposed to overcome that issue. Several experiments were carried out on the supercomputing platform based on Intel x86 CPUs in the National Supercomputing Center in Wuxi. The results demonstrated that the parallel GND core based on grouping solution achieves good strong scalability and improves the performance significantly, as well as avoiding the SWOs.
Highlights
Nowadays, several dynamical core models, such as DYNAMICO [Dubos, Dubey, Tort et al (2015)], FVM [Smolarkiewicz, Kühnlein and Grabowski (2017)], GEM [Girard, Plante, Desgagné et al (2014)] and ICON [Zängl, Reinert, Rípodas et al (2015)], have been developed as a fundamental component of global atmospheric modeling systems
A prototype of Global-Regional Integrated forecast System (GRIST) is being developed by Chinese Academy of Meteorological Sciences and composed of two parts: one part is a global shallow water modeling (SWM) framework [Zhang (2018)], which was developed on an unstructured Voronoi-Delaunay grid and tested against various two-dimensional (2D) benchmarks by isolating most horizontal components of a 3D model; the other part is a new global nonhydrostatic dynamical (GND) core [Zhang, Li, Yu et al (2019)] for supporting multiscale modeling of the atmosphere and enabling resolve more fine-scale fluid structures without a uniform increase in the global resolution
This study has presented the parallel implementation and optimization to a new global nonhydrostatic dynamical (GND) core running on the commercial auxiliary computing system of national supercomputing center in Wuxi, which is a supercomputing platform based on Intel x86 CPUs
Summary
Several dynamical core models, such as DYNAMICO [Dubos, Dubey, Tort et al (2015)], FVM [Smolarkiewicz, Kühnlein and Grabowski (2017)], GEM [Girard, Plante, Desgagné et al (2014)] and ICON [Zängl, Reinert, Rípodas et al (2015)], have been developed as a fundamental component of global atmospheric modeling systems. The Global-Regional Integrated forecast System (GRIST) is the next-generation weather and climate integrated model dynamic framework and aiming. A prototype of GRIST is being developed by Chinese Academy of Meteorological Sciences and composed of two parts: one part is a global shallow water modeling (SWM) framework [Zhang (2018)], which was developed on an unstructured Voronoi-Delaunay grid and tested against various two-dimensional (2D) benchmarks by isolating most horizontal components of a 3D model; the other part is a new global nonhydrostatic dynamical (GND) core [Zhang, Li, Yu et al (2019)] for supporting multiscale modeling of the atmosphere and enabling resolve more fine-scale fluid structures without a uniform increase in the global resolution. We focus on the parallelization and performance optimization to GND core as the simulation to 3D model is more significant than 2D model in a weather and climate system.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
