Articles | Volume 19, issue 9
https://doi.org/10.5194/gmd-19-3783-2026
https://doi.org/10.5194/gmd-19-3783-2026
Development and technical paper
 | 
08 May 2026
Development and technical paper |  | 08 May 2026

Actionable reporting of CPU-GPU performance comparisons: insights from a CLUBB case study

Gunther Huebler, Vincent E. Larson, John M. Dennis, and Sheri A. Voelz

Model code and software

GitHub repo of CLUBB code Gunther Huebler and Vincent Larson https://github.com/larson-group/clubb_release/tree/clubb_performance_testing

Zenodo archive of CLUBB code and profiling results Gunther Huebler and Vincent Larson https://doi.org/10.5281/zenodo.17081296

Download
Short summary
Central processing units (CPUs) and graphics processing units (GPUs) are different devices that suit different kinds of work. Using a climate modeling component, we provide a clearer way to tell which device type is faster for a given task. This matters because runs usually use only one device type. Our results are actionable: they guide device choice, report performance gains fairly, highlight code areas to improve, and show how code structure and optimization can change conclusions.
Share