Bienvenue aux Mines Paristech
Bienvenue à MINES ParisTech
Newsletter International
Website
Théorie & Pratique
Vous êtes

webTV

Lecture

Modéliser pour préparer l'expérience

Lecture

Créer de la musique à partir de gestes dans l'espace

Lecture

Electromagnetic forming process for metallic pieces

Lecture

David Ryckelynck | Entretiens de Toulouse 2015

Lecture

L'intégration du stockage dans un réseau électrique durable

+ Toutes les vidéos

Partager

Synthèse de code hybride à partir d'un programme à mémoire purement distribuée

Synthèse de code hybride à partir d'un programme à mémoire purement distribuée

(Hybrid code synthesis from a pure distributed memory program)

Proposition de thèse

SpécialitéInformatique temps réel, robotique et automatique - Fontainebleau
Ecole doctoraleSMI - Sciences des Métiers de l'Ingénieur
Directeur de thèseTADONKI Claude
Co-directeur de thèseTADONKI Claude
Unité de rechercheMathématiques et Systèmes
ContactClaude TADONKI
Site Web
Mots-clésparallélisme, mémoire
parallelism, memory
RésuméWith the advent of multicore architectures we need us to reconsider the way we organize the tasks of our parallel code and plan their cooperation. Current (and future) supercomputers are built up with multicore processors probably coupled with accelerators such as GPUs. Regarding the connectivity, latency and network bandwidth have certainly evolved significantly, but the gap between the virtual topology and physical topology greatly increases the effective cost of interprocessor communications, especially for applications involving a high connectivity like stencil computation. Therefore, it becomes crucial to really consider hybrid implementations, which consider distributed memory model on top and shared memory model on the nodes. However, several factors are likely to discourage this effort: the required skill to design and implement hybrid codes is not common; in addition, number of quite complex codes have already been written based on the distributed memory model exclusively, and it is hard to consider modifying them; moreover, the belief of a clear effective reward from a hybrid code is moderate, because getting a benefit as close as expected is indeed not trivial. Therefore, designing a quasi-systematic methodology and a corresponding framework, that allow to easily switching from a pure MPI code to an efficient corresponding MPI + OpenMP version would be a valuable contribution. In this thesis, we propose to tackle the problem quite rigorously, on the basis of a tasks graph model and a clearly quantified accounting of major hardware and network aspects. The first goal will be to understand the criteria for an efficient and scalable hybrid code, followed by a methodology for a quick and pragmatic implementation. Applications envisaged for conducting this study are: Lattice QCD; pattern matching on a large sequence of high resolution images; partial differential equations; shortest paths computation in a graph.
With the advent of multicore architectures we need us to reconsider the way we organize the tasks of our parallel code and plan their cooperation. Current (and future) supercomputers are built up with multicore processors probably coupled with accelerators such as GPUs. Regarding the connectivity, latency and network bandwidth have certainly evolved significantly, but the gap between the virtual topology and physical topology greatly increases the effective cost of interprocessor communications, especially for applications involving a high connectivity like stencil computation. Therefore, it becomes crucial to really consider hybrid implementations, which consider distributed memory model on top and shared memory model on the nodes. However, several factors are likely to discourage this effort: the required skill to design and implement hybrid codes is not common; in addition, number of quite complex codes have already been written based on the distributed memory model exclusively, and it is hard to consider modifying them; moreover, the belief of a clear effective reward from a hybrid code is moderate, because getting a benefit as close as expected is indeed not trivial. Therefore, designing a quasi-systematic methodology and a corresponding framework, that allow to easily switching from a pure MPI code to an efficient corresponding MPI + OpenMP version would be a valuable contribution. In this thesis, we propose to tackle the problem quite rigorously, on the basis of a tasks graph model and a clearly quantified accounting of major hardware and network aspects. The first goal will be to understand the criteria for an efficient and scalable hybrid code, followed by a methodology for a quick and pragmatic implementation. Applications envisaged for conducting this study are: Lattice QCD; pattern matching on a large sequence of high resolution images; partial differential equations; shortest paths computation in a graph.
ContexteParallélisme
EncadrementClaude Tadonki, CRI, Centre de recherche en informatique
Profil candidatPlease send a CV, a motivation letter, the list of grades for your last academic year and 1 or 2 recommendation letters.
ObjectifInnovation
Références-
Type financementConcours pour un contrat doctoral

Retour à la liste des propositions

actualité

FiBreMoD2017, MINES ParisTech tisse l'avenir des composites

Formation FiBreMoD2017, MINES ParisTech tisse l'avenir des… Former la prochaine génération de chercheurs européens Du 25…
> En savoir +

Retour sur le prix Pierre Laffitte

Formation Retour sur le prix Pierre Laffitte Cette médaille est créée en l'honneur du  Professeur…
> En savoir +

Employabilité des étudiants

Formation Employabilité des étudiants Dans l'édition 2017 du Global Employability University Ranking , publiée le 16 novembre…
> En savoir +

Un équilibre hommes femmes au travail

Formation Un équilibre hommes femmes au travail « La chance sourit aux esprits préparés » . C'est par cette…
> En savoir +

Hydrocontest 2017, les mineurs relèvent le défi !

Formation Hydrocontest 2017, les mineurs relèvent le défi ! Du 4 au 10 septembre 2017, en compétition pour la 4 e édition…
> En savoir +

+ Toutes les actualités

agenda

Du 25 octobre au 25 novembre 2017 Clivages

Le 22 novembre 2017 Qu'en est-il des transitions énergétiques à l'échelle…

Le 23 novembre 2017 Journée d'accueil des nouveaux doctorants

Le 24 novembre 2017 La cyber sécurité maritime

Du 2 au 10 décembre 2017 Soutenances ALEF, en Chine

Le 5 décembre 2017 Remise du Prix Entrepreneuriat MINES…

+ Tous les événements

contact

Régine MOLINS
Direction de l'Enseignement
Service du Doctorat
> envoyer un mail

Plan du site
MINES
ParisTech

60, Boulevard Saint-Michel
75272 PARIS Cedex 06
Tél. : +33 1 40 51 90 00

Implantations
Formation
Mentions légales | efil.fr | ©2012 MINES ParisTech | +33 1 40 51 90 00 |