OpenMP | EPFL Graph Search

Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?

Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.

Découvrez-en plus sur les applications Graph.

OpenMP (Open Multi-Processing) est une interface de programmation pour le calcul parallèle sur architecture à mémoire partagée. Cette API est prise en charge par de nombreuses plateformes, incluant GNU/Linux, OS X et Windows, pour les langages de programmation C, C++ et Fortran. Il se présente sous la forme d'un ensemble de directives, d'une bibliothèque logicielle et de variables d'environnement. OpenMP est portable et dimensionnable. Il permet de développer rapidement des applications parallèles à petite granularité en restant proche du code séquentiel. La programmation parallèle hybride peut être réalisée par exemple en utilisant à la fois OpenMP et MPI. Le développement de la spécification OpenMP est géré par le consortium OpenMP Architecture Review Board. Historique OpenMP 1.0 pour Fortran a été publié en octobre 1997. En octobre 1998, la spécification pour le C/C++ a été publiée. La version 2.0 pour Fortran a été disponible en 2000. La version 2.0 pour C/C++ en 2

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.

Publications associées (9)

Publications associées

Chargement

Personnes associées (1)

Personnes associées

Chargement

Unités associées (2)

Unités associées

Chargement

Concepts associés (56)

Concepts associés

Chargement

Cours associés (7)

Cours associés

Chargement

Séances de cours associées (20)

Séances de cours associées

Chargement

Hybrid OpenMP/MPI parallelization of the charge deposition step in the global gyrokinetic Particle-In-Cell code ORB5

Stephan Brunner, Claudio Gheller, Emmanuel Lanti, Noé Thomas Elie Ohana, Aaron Lewis Scheinberg, Laurent Villard

Gyrokinetic simulations are computationally extremely demanding due to the high dimensionality of the physical phase space and the interplay between plasma particles and electromagnetic fields. It is thus essential to make full use of the available numerical resources to be able to simulate more complex physical problems. With the aim of optimizing the gyrokinetic Particle-In-Cell code ORB5 towards exascale computing, a particle sorting method is implemented to increase data locality. Furthermore, different algorithms are used to improve vectorization, and the MPI parallelization is complemented with OpenMP. More specifically, we shall focus on the particle to grid operations involved in the PIC charge deposition step. The latter is critical to parallelize using a shared memory paradigm due to the scatter operations involved. We will present the different algorithms and parallelization schemes implemented in the ORB5 charge deposition step and how they affect the speedup compared to the base MPI case.

2017

Chargement

Can a Helmholtz solver run on a cluster?

Marc Anthony David Habisreutinger, Vincent Keller, Emmanuel Leriche

Our vectorized Helmholtz solver runs at 85% efficiency on a NEC SX-5. The most time-consuming parts have been ported on SMP, NUMA, and cluster architectures. It is shown that an OpenMP version can deliver similar performance when running it on a 16 processor SGI Altix. A partial parallelisation using MPI is made to validate the Gamma model to predict application behaviours on parallel machines. This model is then applied to simulate the behaviour of a hypothetical full MPI version on different distributed memory machines. It is found that only the Cray XT3 with its very fast internode communication network will be able to deliver the performance of a NEC SX-8 with the advantage that bigger case could be handled

IEEE2006

Chargement

Towards the optimization of a gyrokinetic Particle-In-Cell (PIC) code on large-scale hybrid architectures

Stephan Brunner, Claudio Gheller, Emmanuel Lanti, Noé Thomas Elie Ohana, Trach-Minh Tran, Laurent Villard

With the aim of enabling state-of-the-art gyrokinetic PIC codes to benefit from the performance of recent multithreaded devices, we developed an application from a platform called the "PIC-engine" [1, 2, 3] embedding simplified basic features of the PIC method. The application solves the gyrokinetic equations in a sheared plasma slab using B-spline finite elements up to fourth order to represent the self-consistent electrostatic field. Preliminary studies of the so-called Particle-In-Fourier (PIF) approach, which uses Fourier modes as basis functions in the periodic dimensions of the system instead of the real-space grid, show that this method can be faster than PIC for simulations with a small number of Fourier modes. Similarly to the PIC-engine, multiple levels of parallelism have been implemented using MPI+OpenMP [2] and MPI+OpenACC [1], the latter exploiting the computational power of GPUs without requiring complete code rewriting. It is shown that sorting particles [3] can lead to performance improvement by increasing data locality and vectorizing grid memory access. Weak scalability tests have been successfully run on the GPU-equipped Cray XC30 Piz Daint (at CSCS) up to 4,096 nodes. The reduced time-to-solution will enable more realistic and thus more computationally intensive simulations of turbulent transport in magnetic fusion devices.

2016

Chargement

CS-307: Introduction to multiprocessor architecture

Multiprocessors are a core component in all types of computing infrastructure, from phones to datacenters. This course will build on the prerequisites of processor design and concurrency to introduce the essential technologies required to combine multiple processing elements into a single computer.

MATH-454: Parallel and high-performance computing

This course provides insight into a broad variety of High Performance Computing (HPC) concepts and the majority of modern HPC architectures. Moreover, the student will learn to have a feeling about what architectures are suited for several types of algorithms.

PHYS-743: Parallel programming

Learn the concepts, tools and API's that are needed to debug, test, optimize and parallelize a scientific application on a cluster from an existing code or from scratch. Both OpenMP (shared memory) and MPI (distributed memory) paradigms are presented and experimented.

Parallélisme (informatique)

vignette|upright=1|Un des éléments de Blue Gene L cabinet, un des supercalculateurs massivement parallèles les plus rapides des années 2000. En informatique, le parallélisme consiste à mettre en œuvr

Microprocesseur multi-cœur

vignette|Un processeur quad-core AMD Opteron. vignette|L’Intel Core 2 Duo E6300 est un processeur double cœur. Un microprocesseur multi-cœur (multi-core en anglais) est un microprocesseur possédant p

Fortran

vignette|Supercalculateur IBM Blue Gene/P.|240x240px vignette|300x300px|Simulation en Fortran de l'accrétion autour d'un trou noir (www.bhac.science). Densité à gauche et densité d'énergie magnétique