PAGCM: A scalable parallel spectral‐based atmospheric general circulation model

Abstract

The Atmospheric General Circulation Model (AGCM) as one of the most important components of Climate System Model (CSM), has been proved to be an effective way for weather forecasting and climate prediction. Although lots of efforts have been conducted to improve the computing efficiency of AGCMs, such as exploit parallel algorithms, migrating codes, and even redesigning systems to adapt to the emerging computer architectures, it is not enough to match the real requirement, due to the limited scalability of the parallel algorithms themselves. Therefore, we design and implement a scalable parallel spectral-based atmospheric circulation mode called PAGCM in this paper. Specifically, we first analyze the data dependencies of the dimensions in different spaces according to the calculation characteristics of spectral models, and based on which we propose a two-dimensional decomposition algorithm in PAGCM to effectively increase the involving cores for the parallel computing, and thus reduce the overall computing time.Furthermore, to adapt to the novel data decomposition in each computing stage of dynamic framework, we propose three-dimensional data transposition algorithms and data collection algorithms correspondingly, by considering of load balancing and communication optimization.Extensive experiments are conducted on Tianhe-2 to validate the effectiveness and scalability of our proposals. KEYWORDS atmospheric general circulation model, parallel computing, scalability, spectral model INTRODUCTIONGlobal climate and environmental changes under global warming, have been one of the great challenges facing human societies in recent decades.There are various factors lead to these changes, which comprehensively reflect the complex interactions among atmosphere, hydrosphere, cryosphere, lithosphere, and biosphere within the climate system. The Climate System Model (CSM) as an effective tool for establishing numerical models for simulating the interactions and exploring the nature for the climate change, has received increasing attention with the rapid development of computers in recent decades. 1,2 After many years of efforts by countless researchers, now there exist many CSMs and their variations for climate prediction. Particulary, most existing CSMs are the loosely coupled systems consisting of several components. The Atmospheric General Circulation Model (AGCM) as one of the most important components, has been proved to be an effective way for weather forecasting and climate prediction 3-5 and has been a major focus of research in Meteorology and Climatology community.Generally, AGCMs usually require integration of mass data over several years or even decades. 3,6,7 Such large-scale calculations needs powerful computing capability, which forces us to enhance the capability of high performance computing, as well as to exploit more efficient algorithms to improve the computing efficiency. As far as we know, the trend of research on AGCMs mainly lies in improving the accuracy and computing efficiency. To improve the accura...

Parallelization of a meteorological model using message passing interface and CUDA: A case study with the inversion estimation algorithm

“…They were the first to apply numerical testing of the randomized incremental optimal technique (RIOT). Ren et al attempted to increase the performance of a weather forecasting and climate prediction model called AGCM, by implementing a novel parallelization technique 27 . They aimed to effectively increase the number of cores by considering load balancing and communication optimization.…”