Case Study in Large Scale Climate Simulations: Optimizing the Speedup/Efficiency Balance in Supercomputing Environments
Muhammad Asif, Andrés Cencerrado, Oriol Mula Valls, Domingo Manubens, Ana Cortés, Francisco Doblas Reyes · 2014
In this work we present the EC-Earth coupled climate model, which is a seamless Earth System Model (ESM) used to carry out climate research in 24 academic institutions and meteorological services from 11 countries in Europe. This model couples several components and it is continuously under development. As a coupled model, EC-Earth consists of several different sub models, each of them presenting different degrees of performance because of their different features and complexities. This represents an important disadvantage at the time of consuming computing time in supercomputing environments, which usually is granted by the corresponding host institution provided that the efficient usage of the computing resources is demonstrated. For this reason, a study to determine the best distribution of computing processors between components was carried out to assess empirically the different performance of the model depending on the amount of computing processors allocated to each component. This experimentation allowed us to optimize the speedup/efficiency balance of the model. Moreover, the obtained results highlight the important drawback caused by the low scalability of one of the EC-Earth components.