Skip to article frontmatterSkip to article content
Simula Research Laboratory

DOLFINx and FFCx on Amazon’s Graviton3

Authors
Affiliations
Rafinex Sarl
University of Luxembourg

We discuss performance results from executing the FEniCS Project finite element software on Amazon Web Services (AWS) systems with ARM64-based Graviton3(e) processors (Habera and Hale, 2023). We show both weak scaling tests across nodes and explore the auto-vectorisation capabilities of the latest clang and GCC compilers on FFCx generated kernels.

References

Habera, M., Hale, J. S., 2023. The FEniCS Project on AWS Graviton3. Preprint. https://hdl.handle.net/10993/55560

Abstracts
Determination of Navier's slip parameter and the inflow velocity using variational data assimilation
Abstracts
Efficient and scalable solvers for a cell-by-cell dual-poroelasticity problem