Abstract

IFP Energies Nouvelles Conference: SimRace 2015: Numerical Methods and High Performance Computing for Industrial Fluid Flows

Highlights

  • The novelties emerging in the different fields involved in scientific computing, for instance graphics accelerators or manycore processors, programming languages, Domain Specific Languages (DSL; high level languages dedicated to an application domain), haute performance (HPC) libraries, mesh refinement, error estimates, high order numerical methods, ever finer models, . . . are often not so easy to combine

  • Among the topics raised by the conference, we have identified four main themes developed in the articles selected in this special issue of OGST

  • Getting further into the numerical field, we deal with numerical scheme derivations, and again we propose to link it with HPC issues

Read more

Summary

FOR INDUSTRIAL FLUID FLOWS

The different topics it relies on (scientific modelling, applied mathematics, High Performance Computing (HPC) and computer science) are continuously evolving, and not necessarily in compatible ways. One part of the challenge of building the 21st century’s ultimate numerical platform will be to create connections and joint works between these spheres so that they can feed each other and develop benefiting synergies Following this trend, IFP Energies nouvelles organized, at the end of 2015, an international conference entitled SimRace, devoted to numerical methods and high performance computing. It presents an embedded language (i.e. embedded in a host standard language, here C++) dedicated to the implementation of numerical methods written in a variational form The idea of this DSL is to provide a high level expressive language and to internally handle (thanks to statically generated code) performance and hardware compliance issues. By allowing the user to graphically describe (using, for instance, class diagrams similar to the Unified Modelling Language (UML) standard) the components of the code and their relationships, Modane provides an overview of the overall simulator architecture, facilitates a rapid integration of new collaborators and allows following of the architecture’s evolution over time

SCALABLE LINEAR SOLVERS
NUMERICAL SCHEMES ON GENERAL MESHES FOR COMPLEX FLOWS
CONCLUSION
MODÈLES DE PROGRAMMATION HPC POUR LES ARCHITECTURES ÉMERGEANTES
Findings
SCHÉMAS NUMÉRIQUES POUR ÉCOULEMENTS COMPLEXES SUR MAILLAGES GÉNÉRAUX
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call