Journal cover Journal topic
Geoscientific Model Development An interactive open-access journal of the European Geosciences Union
doi:10.5194/gmd-2016-247
© Author(s) 2016. This work is distributed
under the Creative Commons Attribution 3.0 License.
Development and technical paper
11 Nov 2016
Review status
A revision of this discussion paper was accepted for the journal Geoscientific Model Development (GMD) and is expected to appear here in due course.
rpe v5: An emulator for reduced floating-point precision in large numerical simulations
Andrew Dawson and Peter Düben Atmospheric, Oceanic & Planetary Physics, Department of Physics, University of Oxford, Oxford UK
Abstract. This paper describes the rpe library which has the capability to emulate the use of arbitrary reduced floating-point precision within large numerical models written in Fortran. The rpe software allows model developers to test how reduced floating-point precision affects the result of their simulations without having to make extensive code changes or port the model onto specialised hardware. The software can be used to identify parts of a program that are problematic for numerical precision and to guide changes to the program to allow a stronger reduction in precision.

The development of rpe was motivated by the strong demand for more computing power. If numerical precision can be reduced for an application under consideration while still achieving results of acceptable quality, computational cost can be reduced, since a reduction in numerical precision may allow an increase in performance or a reduction in power consumption. For simulations with weather and climate models, savings due to a reduction in precision could be reinvested to allow model simulations at higher spatial resolution or complexity, or to increase the number of ensemble members to improve predictions. rpe was developed with particular focus on the community of weather and climate modelling, but the software could be used with numerical simulations from other domains.


Citation: Dawson, A. and Düben, P.: rpe v5: An emulator for reduced floating-point precision in large numerical simulations, Geosci. Model Dev. Discuss., doi:10.5194/gmd-2016-247, in review, 2016.
Andrew Dawson and Peter Düben
Interactive discussionStatus: closed
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
Printer-friendly Version - Printer-friendly version      Supplement - Supplement
 
RC1: 'x', Anonymous Referee #1, 29 Nov 2016 Printer-friendly Version 
 
RC2: 'review', Anonymous Referee #2, 23 Feb 2017 Printer-friendly Version 
AC1: 'short comment in reply to reviewer 2', Andrew Dawson, 13 Mar 2017 Printer-friendly Version 
Andrew Dawson and Peter Düben
Andrew Dawson and Peter Düben

Viewed

Total article views: 296 (including HTML, PDF, and XML)

HTML PDF XML Total BibTeX EndNote
216 65 15 296 6 16

Views and downloads (calculated since 11 Nov 2016)

Cumulative views and downloads (calculated since 11 Nov 2016)

Viewed (geographical distribution)

Total article views: 296 (including HTML, PDF, and XML)

Thereof 296 with geography defined and 0 with unknown origin.

Country # Views %
  • 1

Saved

Discussed

Latest update: 23 May 2017
Publications Copernicus
Download
Short summary
Weather and climate models must become more efficient if they are to continue growing in complexity. One option for reducing computational cost is to reduce numerical precision. We present a tool that allows users to study how models perform with reduced numerical precision. The tool is applied to a geophysical use case, where precision is heavily reduced while maintaining suitable accuracy. The tool can be applied to other models to determine if they can be made more computationally efficient.
Weather and climate models must become more efficient if they are to continue growing in...
Share