guest ::
| Home > Publications database > EoCoE Performance Benchmarking Methodology for Renewable Energy Applications |
| Home > Publications database > EoCoE Performance Benchmarking Methodology for Renewable Energy Applications |
| Poster (Other) | FZJ-2018-00744 |
; ; ; ; ; ; ; ; ; ; ; ;
2017
Please use a persistent id in citations: http://hdl.handle.net/2128/16692
Abstract: The global transition away from fossil fuels towards a sustainable, decarbonized energy ecosystem will rely heavily on digitization to drive necessary innovations in production and storage technologies, mitigate power source intermittency and manage its distribution via a complex grid hierarchy. At the same time, supercomputing is also experiencing a major paradigm shift: future exascale technologies will open up unprecedented opportunities to tackle complex physical problems – such as the design of wind farms or smart materials for photovoltaics and batteries – but will demand major restructuring of application software, numerical algorithms and programming models. These challenges motivated the creation of the Energy Oriented Centre of Excellence (EoCoE) two years ago; an EU-funded consortium twenty-one partners across eight countries with strong engagements in both the HPC and energy fields.This poster presents an optimisation strategy developed by the Energy Oriented Centre of Excellence (EoCoE) for computational models used in a variety of renewable energy domains. It is found that typical applications in this comparatively new sector cover the widest possible range of HPC maturity, from simple parallelization needs to near-exascale readiness. A key part of this process has therefore been the quantitative, reproducible performance assessment of applications consolidated by follow-up actions by code-teams comprising members of both developer groups and HPC centres involved with the EoCoE consortium. Examples of early successes achieved with this practice are given, together with an outlook on challenges faced for energy applications with next-generation, pre-exascale architectures.
|
The record appears in these collections: |
PDF
PDF (PDFA)