En poursuivant votre navigation sur ce site, vous acceptez le dépôt de cookies dans votre navigateur. (En savoir plus)
Portail > Offres > Offre UMR5574-BENCOM-009 - Postdoc en calcul haute performance (H/F)

High Performance Computing Postdoc (M/F)

This offer is available in the following languages:
- Français-- Anglais

Date Limite Candidature : lundi 26 février 2024

Assurez-vous que votre profil candidat soit correctement renseigné avant de postuler

Informations générales

Intitulé de l'offre : High Performance Computing Postdoc (M/F) (H/F)
Référence : UMR5574-BENCOM-009
Nombre de Postes : 1
Lieu de travail : LYON 07
Date de publication : lundi 5 février 2024
Type de contrat : CDD Scientifique
Durée du contrat : 24 mois
Date d'embauche prévue : 1 avril 2024
Quotité de travail : Temps complet
Rémunération : min 2934 euros brut monthly salary (commensurate with experience)
Niveau d'études souhaité : Niveau 8 - (Doctorat)
Expérience souhaitée : Indifférent
Section(s) CN : Solar system and distant universe

Missions

A postdoctoral position in high performance computing is open at the Centre de Recherche Astrophysique de Lyon (CRAL) for a period of two years. This position is financed within the framework of the Centre of Excellence (CoE) EuroHPC SPACE (https://www.space-coe.eu/) which gathers various European partners, both academic and industrial. The objective is to develop flagship applications in numerical astrophysics for exascale architectures. CRAL is leading the developments around the adaptive mesh refinement code RAMSES, known for its wide range of applications in astrophysical fluid dynamics and structure formation. RAMSES is currently written in Fortran and uses a distributed memory parallelisation with the MPI library. The aim is to pursue different avenues of development, starting with a simplified version (miniramses) using different strategies (load-balancing, MPI communications, ordering). The objectives are 1/ run on hybrid CPU and GPU architectures; 2/ improve the performances on ARM type architectures with a high number of computing cores 3/ published the new developments.

Activités

From mini app software, different tracks will be explored to optimize both the structure of the code (octs tree, hydrodynamic kernels, load balancing) and the parallelism (MPI-3, OpenACC, OpenMP, etc...). Beyond the daily interactions with the researchers and engineers at CRAL, the work will involve regular interactions with the teams of INAF in Italy, Eviden in France and Princeton in USA. An involvement in the scientific animation around HPC developments at CRAL (~10 people) will be necessary: knowledge transfer to students and researchers involved in the project, participation in meetings and scientific life and supervision of trainees. Regular reporting will also be required as part of the work plan established for the SPACE CoE.

This project could involve the development of large-volume simulations (cosmological box or star formation in the interstellar medium using sink particles).

Compétences

Programming on CPU and GPU (OpenACC, OpenMP and MPI). Ideally Fortran. SYCL, CUDA, other for GPU preferred.
Mastery of version management and code profiling tools
Experience in algorithmic and code optimization.
Knowledge in astrophysics appreciated but not mandatory.
The successful candidate should have a PhD degree or equivalent.

Contexte de travail

The CRAL is located on two sites (ENS de Lyon in Gerland and Observatoire de Lyon in Saint Genis Laval). The successful candidate will be hosted at the ENS of Lyon. He/she will be required to travel in France and in Europe to meet the different partners of the European project and to participate in the progress meetings. CRAL has access to the resources of the Pôle Scientifique de Modélisation Numérique and the Centre Blaise Pascal at the ENS de Lyon (15,000 cores, technical platform with 67 different GPU models). The CRAL teams also have exclusive access to a Nvidia DGX station equipped with a 64-core AMD Epyc CPU and 4 NVIDIA A100 GPUs linked in NVlink. Access to state-of-the-art EuroHPC technologies.