Massively parallel numerical simulations with Julia

arXiv (Cornell University)

Abstract

The Julia programming language aims to provide a modern approach to develop high-performance computing (HPC) applications. It tries to achieve this by combining a high-level, dynamic interface with just-in-time compilation to native machine code, thereby facilitating high developer productivity and native code performance at the same time. While this approach has already been shown to work well for serial applications, it is not clear if it readily translates to traditional, massively parallel HPC work loads. In this paper, we fill this gap by analyzing the parallel performance of the numerical computational fluid dynamics simulation code Trixi$.$jl, written in Julia, and compare it to the Fortran code FLUXO. We show some of the challenges of using Julia at scale and discuss possible solutions, specifically with respect to code loading and compilation at startup. Finally, we demonstrate the parallel scaling of our Julia code on up to 61440 CPU cores.

Authors 9

1. Simon Candelaresi

   University of Augsburg

   Affiliation as printed

   High-Performance Scientific Computing , Centre for Advanced Analytics and Predictive Sciences , University of Augsburg , Germany

2. Benedict Geihe

   TH Köln - University of Applied Sciences · University of Cologne

   Affiliation as printed

   Department of Mathematics and Computer Science , Division of Mathematics , University of Cologne , Germany

3. Marco Artiano

   Johannes Gutenberg University Mainz

   Affiliation as printed

   Institute of Mathematics , Johannes Gutenberg University Mainz , Germany

4. Lars Christmann Aachen Applied and Computational Mathematics

   TH Köln - University of Applied Sciences · University of Cologne · RWTH Aachen University

   Affiliation as printed

   Applied and Computational Mathematics , RWTH Aachen University , Germany

   Department of Mathematics and Computer Science , Division of Mathematics , University of Cologne , Germany

5. Valentin Churavy

   University of Augsburg · Johannes Gutenberg University Mainz

   Affiliation as printed

   High-Performance Scientific Computing , Centre for Advanced Analytics and Predictive Sciences , University of Augsburg , Germany

   Institute of Mathematics , Johannes Gutenberg University Mainz , Germany

6. Andrés Mauricio Rueda-Ramírez

   Universidad Politécnica de Madrid

   Affiliation as printed

   School of Aerospace Engineering , Universidad Polit écnica de Madrid , Spain

7. Hendrik Ranocha

   University of Augsburg · Johannes Gutenberg University Mainz

   Affiliation as printed

   High-Performance Scientific Computing , Centre for Advanced Analytics and Predictive Sciences , University of Augsburg , Germany

   Institute of Mathematics , Johannes Gutenberg University Mainz , Germany

8. Gregor J. Gassner

   Johannes Gutenberg University Mainz

   Affiliation as printed

   Institute of Mathematics , Johannes Gutenberg University Mainz , Germany

9. Michael Schlottke‐Lakemper Aachen Applied and Computational Mathematics

   University of Augsburg · RWTH Aachen University

   Affiliation as printed

   Applied and Computational Mathematics , RWTH Aachen University , Germany

   High-Performance Scientific Computing , Centre for Advanced Analytics and Predictive Sciences , University of Augsburg , Germany