Journal cover Journal topic
Geoscientific Model Development An interactive open-access journal of the European Geosciences Union
Journal topic

Journal metrics

Journal metrics

  • IF value: 4.252 IF 4.252
  • IF 5-year value: 4.890 IF 5-year 4.890
  • CiteScore value: 4.49 CiteScore 4.49
  • SNIP value: 1.539 SNIP 1.539
  • SJR value: 2.404 SJR 2.404
  • IPP value: 4.28 IPP 4.28
  • h5-index value: 40 h5-index 40
  • Scimago H index value: 51 Scimago H index 51
Volume 11, issue 4 | Copyright
Geosci. Model Dev., 11, 1665-1681, 2018
https://doi.org/10.5194/gmd-11-1665-2018
© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Development and technical paper 02 May 2018

Development and technical paper | 02 May 2018

Near-global climate simulation at 1 km resolution: establishing a performance baseline on 4888 GPUs with COSMO 5.0

Oliver Fuhrer1, Tarun Chadha2, Torsten Hoefler3, Grzegorz Kwasniewski3, Xavier Lapillonne1, David Leutwyler4, Daniel Lüthi4, Carlos Osuna1, Christoph Schär4, Thomas C. Schulthess5,6, and Hannes Vogt6 Oliver Fuhrer et al.
  • 1Federal Institute of Meteorology and Climatology, MeteoSwiss, Zurich, Switzerland
  • 2ITS Research Informatics, ETH Zurich, Switzerland
  • 3Scalable Parallel Computing Lab, ETH Zurich, Switzerland
  • 4Institute for Atmospheric and Climate Science, ETH Zurich, Switzerland
  • 5Institute for Theoretical Physics, ETH Zurich, Switzerland
  • 6Swiss National Supercomputing Centre, CSCS, Lugano, Switzerland

Abstract. The best hope for reducing long-standing global climate model biases is by increasing resolution to the kilometer scale. Here we present results from an ultrahigh-resolution non-hydrostatic climate model for a near-global setup running on the full Piz Daint supercomputer on 4888GPUs (graphics processing units). The dynamical core of the model has been completely rewritten using a domain-specific language (DSL) for performance portability across different hardware architectures. Physical parameterizations and diagnostics have been ported using compiler directives. To our knowledge this represents the first complete atmospheric model being run entirely on accelerators on this scale. At a grid spacing of 930m (1.9km), we achieve a simulation throughput of 0.043 (0.23) simulated years per day and an energy consumption of 596MWh per simulated year. Furthermore, we propose a new memory usage efficiency (MUE) metric that considers how efficiently the memory bandwidth – the dominant bottleneck of climate codes – is being used.

Publications Copernicus
Download
Short summary
The best hope for reducing long-standing uncertainties in climate projections is through increasing the horizontal resolution of climate models to the kilometer scale. We establish a baseline of what it would take to do such simulations using an atmospheric model that has been adapted to run on a supercomputer accelerated with graphics processing units. To our knowledge this represents the first production-ready atmospheric model being run entirely on accelerators on this scale.
The best hope for reducing long-standing uncertainties in climate projections is through...
Citation
Share