Load balance and Parallel I/O: Optimising COSA for scale

M. Sergio Campobasso, William Jackson, Jernej Drofelnik

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper describes steps taken to optimise COSA for large scale parallel simulations. Optimising the general computational structure of Fluidity, along with work to improve the data decomposition and parallel load balance enabled simulations to be run over 3x faster than with the original code, even at scale.
Original languageEnglish
Title of host publicationComputer and Fluids
Publication statusPublished - 17 May 2017
EventInternational Conference on Parallel Computational Fluid Dynamics - University of Strathclyde, Glasgow, United Kingdom
Duration: 15 May 201717 May 2017
Conference number: 29
http://www.strath.ac.uk/media/1newwebsite/departmentsubject/engineering/parallelcfd/cfd_flyer_dec_2016.pdf

Conference

ConferenceInternational Conference on Parallel Computational Fluid Dynamics
Abbreviated titleParCFD
CountryUnited Kingdom
CityGlasgow
Period15/05/1717/05/17
Internet address

Fingerprint

Dive into the research topics of 'Load balance and Parallel I/O: Optimising COSA for scale'. Together they form a unique fingerprint.

Cite this