2016/10/28 by Satya P. Jammy, Jammy, Satya P., Christian T. Jacobs +3 · 2 citations
Computer Science · Earth and Planetary Sciences · Engineering · Mathematics · Physics and Astronomy · #Computational Fluid Dynamics and Aerodynamics #Computational Physics (physics.comp-ph) #Data Structures and Algorithms (cs.DS) #Distributed #FOS: Computer and information sciences #FOS: Physical sciences #Fluid Dynamics (physics.flu-dyn) #Mathematical Software (cs.MS) #Meteorological Phenomena and Simulations #Numerical methods for differential equations #Parallel #and Cluster Computing (cs.DC) #cs.DC #cs.DS #cs.MS #physics.comp-ph #physics.flu-dyn
paper · pdf · doi:10.48550/arxiv.1610.09146
Author accepted version. Accepted for publication in Journal of Computational Science on 27 October 2016
arxiv created 2016/10/28 · openalex publication_date 2016/10/28 · arxiv updated 2016/10/31 · openalex created_date 2022/10/03 · openalex updated_date 2026/07/28
Future architectures designed to deliver exascale performance motivate the need for novel algorithmic changes in order to fully exploit their capabilities. In this paper, the performance of several numerical algorithms, characterised by varying degrees of memory and computational intensity, are evaluated in the context of finite difference methods for fluid dynamics problems. It is shown that, by storing some of the evaluated derivatives as single thread- or process-local variables in memory, or recomputing the derivatives on-the-fly, a speed-up of ~2 can be obtained compared to traditional algorithms that store all derivatives in global arrays.