To offer the results of EXA2CT to the wider community open-source proto-applications are being build. These help enable exascale machine/software co-design and are a basis for exascale applications.
Download the current applications below:
- Mini-FEM a proto-application reproducing the assembly step of FEM applications working on 3D unstructured meshes.
- Source code: https://github.com/EXAPARS/Mini-FEM
- A CFD Proxy kernel, aimed at ghost cell exchanges in unstructured meshes.
- Source code: https://github.com/PGAS-community-benchmarks/CFD-Proxy
- Documentation: https://github.com/PGAS-community-benchmarks/CFD-Proxy/blob/master/documentation/CFD-Proxy.pdf
- BPMF A big data and machine learning proto application. BPMF is aimed at showing how to do big data and machine learning at scale.
- Source code: https://github.com/ExaScience/bpmf
- Background info: https://www.cs.toronto.edu/~amnih/papers/bpmf.pdf
- Industrial application: The machine learning framework MACAU. https://github.com/jaak-s/BayesianDataFusion.jl
We are also developing proto-applications based on these industrial codes:
- YALES2 – CFD and combustion code from CORIA, France http://www.coria-cfd.fr/index.php/YALES2
- MUPHY – Multi-physics code for biomedical applications developed by Roma, Italy http://dx.doi.org/10.1016/j.cpc.2009.04.001
- OASIS3-MCT – Coupling code by CERFACS, France, developed primarily for climate applications https://verc.enes.org/oasis