Preparing scientific application software for exascale computing

  • Authors:

  • J. A. Åström;A. Carter;J. Hetherington;K. Ioakimidis;E. Lindahl;G. Mozdzynski;R. W. Nash;P. Schlatter;A. Signell;J. Westerholm

  • Affiliations:

  • CSC --- IT Center for Science, Espoo, Finland;EPCC, The University of Edinburgh, Edinburgh, UK;Dept of Chemistry, Faculty of Maths & Physical Sciences, University College London, London, UK;Institute of Fluid Mechanics and Hydraulic Machinery, University of Stuttgart, Stuttgart, Germany;Science for Life Laboratory, KTH & Stockholm University, Solna, Sweden;European Centre for Medium-Range Weather Forecasts, Reading, UK;Dept of Chemistry, Faculty of Maths & Physical Sciences, University College London, London, UK;Department of Mechanics, KTH, Stockholm, Sweden;Institutionen för Informationsteknologi, Datateknik/ högeffektiva datorberäkningar, Åbo Akademi, Turku, Finland;Institutionen för Informationsteknologi, Datateknik/ högeffektiva datorberäkningar, Åbo Akademi, Turku, Finland

  • Venue:
  • PARA'12 Proceedings of the 11th international conference on Applied Parallel and Scientific Computing
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Many of the most widely used scientifc application software of today were developed largely during a time when the typical amount of compute cores was calculated in tens or hundreds. Within a not too distant future the number of cores will be calculated in at least hundreds of thousands or even millions. A European collaboration group CRESTA has recently been working on a set of renowned scientific software to investigate and develop these codes towards the realm of exascale computing. The codes are ELMFIRE, GROMACS, IFS, HemeLB, NEK5000, and OpenFOAM. This paper contains a summary of the strategies for their development towards exascale and results achieved during the first year of the collaboration project.