Author(s)

C.R. Mechoso, J.D. Farrara, L.A. Drummond, J. Spahr & J.-Y. Yu

Abstract

The top of the hierarchy of numerical models used for research and prediction of climate variability (e.g., El Nino and anthropogenic climate change) is occupied by general circulation models of the coupled atmosphere-ocean system. These are large codes characterized by heterogeneous components and diverse algorithms. This paper presents methods to overcome two major obstacles to performance efficiency when the application runs in a multi-processor environment: balancing of the computing load among processors and data transfers among model components. The power of these methods is illustrated in the context of the model developed at the University of California, Los Angeles. An example of El Nino simulated by that model is presented. 1 Introduction

Keywords