Development and performance-testing of FEniCS-HPC

Dnr:

SNIC 2017/5-10

Type:

SNAC Small

Principal Investigator:

Johan Jansson

Affiliation:

Kungliga Tekniska högskolan

Start Date:

2017-01-31

End Date:

2017-12-01

Primary Classification:

10105: Beräkningsmatematik

Webpage:

http://ctl.csc.kth.se

Allocation

Abstract

We plan to carry out performance testing and scalability studies on Triolith as part of the research and development of our adaptive FEM framework Unicorn/FEniCS-HPC [1, 3, 4] where a specific focus is on turbulent flow and FSI problems, We have verified optimal strong scaling on both Lindgren and Beskow up to ca. 5000 cores with a pure MPI backend using PETSc [1] and up to 12000 cores on Beskow using the experimental hybrid MPI+PGAS JANPACK backend [2, 4]. We will present our results at computational mathematics and mechanics conferences and in journals. [1] J. Hoffman, J. Jansson, R. V. de Abreu, N. C. Degirmenci, N. Jansson, K. Muller, M. Nazarov, and J. H. Spuhler. Unicorn: Parallel adaptive finite element simulation of turbulent flow and fluid-structure interaction for deforming domains and complex geometry. Computers & Fluids, 2012 [2] N. Jansson. Optimizing Sparse Matrix Assembly in Finite Element Solvers with One-sided Communication. In High Performance Computing for Computational Science { VECPAR 2012, volume 7851 of Lecture Notes in Computer Science , pages 128{139. Springer Berlin Heidelberg, 2013. [3] Towards a parameter-free method for high Reynolds number turbulent flow simulation based on adaptive finite element approximation, J Hoffman, J Jansson, N Jansson, RV De Abreu, Computer Methods in Applied Mechanics and Engineering, 2015 [4] J Hoffman, J Jansson, N Jansson, FEniCS-HPC: Automated predictive high-performance finite element computing with applications in aerodynamics, PPAM 2015