Computational performance of a parallelized three-dimensional high-order spectral element toolbox

In this paper, a comprehensive performance review of an MPI-based high-order three-dimensional spectral element method C++ toolbox is presented. The focus is put on the performance evaluation of several aspects with a particular emphasis on the parallel efficiency. The performance evaluation is analyzed with the help of a time prediction model based on a parameterization of the application and the hardware resources. Two tailor-made benchmark cases in computational fluid dynamics (CFD) are introduced and used to carry out this review, stressing the particular interest for clusters with up to thousands of cores. Some problems in the parallel implementation have been detected and corrected. The theoretical complexities with respect to the number of elements, to the polynomial degree, and to communication needs are correctly reproduced. It is concluded that this type of code has a nearly perfect speedup on machines with thousands of cores, and is ready to make the step to next-generation petaFLOP machines. (C) 2010 Elsevier Ltd. All rights reserved.

À propos de ce résultat

Cette page est générée automatiquement et peut contenir des informations qui ne sont pas correctes, complètes, à jour ou pertinentes par rapport à votre recherche. Il en va de même pour toutes les autres pages de ce site. Veillez à vérifier les informations auprès des sources officielles de l'EPFL.

Concepts associés

Chargement

Publications associées

Chargement

Christoph Bosshard, Roland Bouffanais, Michel Deville, Jonas Marcel Lätt
2011
Article

Résumé

Official source

https://infoscience.epfl.ch/record/171544?ln=fr

À propos de ce résultat

Concepts associés (10)

Concepts associés

Chargement

Publications associées (4)

Publications associées

Chargement

Computational Performance of a Parallelized Three-Dimensional High-Order Spectral Element Toolbox

Christoph Bosshard, Roland Bouffanais, Christian Clémençon, Michel Deville, Nicolas Fietier, Sohrab Jairo Kehtari, Vincent Keller, Jonas Marcel Lätt

In this paper, a comprehensive performance review of all MPI-based high-order three-dimensional spectral element method C++ toolbox is presented. The focus is put on the performance evaluation of aspects with a particular emphasis on the parallel efficiency. The performance evaluation is analyzed with help of a time prediction model based on a parameterization of the application and the hardware resources. A tailor-made CFD computation benchmark case is introduced and used to carry out this review, stressing the particular interest for clusters with up to 8192 cores. Some problems in the parallel implementation have been detected and corrected. The theoretical complexities with respect, to the number of elements, to the polynomial degree, and to communication needs are correctly reproduced. It is concluded that this type of code has a nearly perfect speed up on machines with thousands of crores, and is ready to make the step to next-generation petaflop machines

Springer-Verlag New York, Ms Ingrid Cunningham, 175 Fifth Ave, New York, Ny 10010 Usa2009

Chargement

Evaluating modeling uncertainties in the simulation of airflow in cities

Ian Smith, Didier Vernay

Even though computational fluid-dynamics (CFD) has been used to simulate airflow in cities, accurate results are not always obtained because of difficulties related to defining the correct set of values of parameters, including boundary conditions. This paper uses a multi-model system-identification approach for improving the accuracy of airflow modeling. In this approach, multiple model predictions are generated through varying values of parameters that are not known precisely. Measurements are then carried out in order to falsify predictions that are not consistent with the measurements. A rejection criterion is defined through consideration of measurement and inherent modeling uncertainties. This paper proposes a methodology for evaluating modeling uncertainties within a system-identification framework. The methodology has been tested on an experimental facility of the Future Cities Laboratory, called “BubbleZERO”. It is shown that reliable predictions at locations where there are no sensors can be performed through implementation of the model-falsification approach.

2013

Chargement

A model-based data-interpretation framework for improving wind predictions around buildings

Ian Smith, Didier Vernay

Although Computational Fluid Dynamics (CFD) simulations are often used to assess wind conditions around buildings, the accuracy of such simulations is often unknown. This paper proposes a data-interpretation framework that uses multiple simulations in combination with measurement data to improve the accuracy of wind predictions. Multiple simulations are generated through varying sets of parameter values. Sets of parameter values are falsified and thus not used for predictions if differences between measurement data and simulation predictions, for any measurement location, are larger than an estimate of uncertainty bounds. The bounds are defined by combining measurement and modeling uncertainties at sensor locations. The framework accounts for time-dependent and spatially-distributed modeling uncertainties that are present in CFD simulations of wind. The framework is applied to the case study of the CREATE Tower located at the National University of Singapore. Values for time-dependent inlet conditions, as well as values for the roughness of surrounding buildings, are identified with measurements carried out around the CREATE Tower. Results show that, on average, ranges of horizontal wind-speed predictions at an unmeasured location have been decreased by 65% when measurement data are used. (C) 2015 Elsevier Ltd. All rights reserved.

Elsevier2015

Chargement

Mécanique des fluides numérique

La mécanique des fluides numérique (MFN), plus souvent désignée par le terme anglais computational fluid dynamics (CFD), consiste à étudier les mouvements d'un fluide, ou leurs effets, par la résolu

Mise en œuvre

La mise en œuvre est le fait de mettre en place un projet. Ingénierie et informatique En ingénierie et plus particulièrement en informatique, la mise en œuvre désigne la création d’un pr

Théorie de la complexité (informatique théorique)

vignette|Quelques classes de complexité étudiées dans le domaine de la théorie de la complexité. Par exemple, P est la classe des problèmes décidés en temps polynomial par une machine de Turing déterm