SP7 High Performance Analytics and Computing Platform
The High Performance Analytics and Computing (HPAC) Platform provides supercomputing, storage, and visualisation resources for the simulation of complex models and for analyzng huge neuroscientific datasets.
Because of the incredible complexity of the human brain, analyses and simulations of the brain require massive computational power, and hence incredibly powerful computers, as well as immense storage capacity. Our experts support researchers in preparing their software to use these systems efficiently. Our partner institutions provide supercomputing, storage and visualization capabilities as well as the necessary software to facilitate and enable modelling, simulation, and data analysis on supercomputers. Some of our key areas of research include new methods for brain simulation, linking extreme scale data processing challenges to the exploitation of scalable compute resources, using accelerator technologies to address computational challenges, novel visualization methods, and innovative approaches for dynamic resource management on supercomputers.
We built and released a federated High Performance Computing (HPC) infrastructure comprising six supercomputers in Germany, Italy, Spain and Switzerland. New supercomputer architectures have been developed by HPC vendors in the context of a competitive Pre-Commercial Procurement focusing on dense memory integration, scalable visualisation and dynamic resource management. The two resulting pilot systems JULIA and JURON are available to HBP scientists as integral parts of the HPAC Platform. Co-development with HPAC Platform users has resulted in algorithms and tools enabling HBP-specific workloads and workflows on supercomputers. Research on numerical methods to improve brain simulations, programming models enabling efficient use of new HPC architectures, and visualization and data management methods and tools, resulted in the release of several software tools and libraries.
In SP7, members constantly research new technologies:
- Data intensive supercomputing: links extreme scale data processing challenges to the exploitations of scalable computational resources
- Interactive visualization: developing tools to visually analyse brain simulation results
- Dynamic resource management: novel approaches to manage computing resources in a supercomputer across applications
- New HPC architectures: based on the specific requirements of neuroscience simulations and analysis methods
In the current phase of the Project (SGA1), we will extend the HPAC Platform with federated data services enabling users to easily upload large data to the Platform, transfer it between sites, and share simulation and analysis results with others inside and outside the HBP.
The High Performance Analytics & Computing Platform Guidebook: https://hbp-hpc-platform.fz-juelich.de
Our Twitter page (@HBPHighPerfComp): https://twitter.com/HBPHighPerfComp
Our YouTube channel (HBP HighPerfComp)
SP Leader: Thomas LIPPERT
Deputy SP Leader: Thomas SCHULTHESS
Work Package Leaders:
- WP7.1 Simulation Technology: Markus DIESMANN, Hans Ekkehard PLESSER
- WP7.2 Data-Intensive Supercomputing: Dirk PLEITER
- WP7.3 Interactive Visualisation: Torsten KUHLEN, Benjamin WEYERS
- WP7.4 Dynamic Resource Management: Raül SIRVENT, Julita CORBALÁN
- WP7.5 High Performance Analytics & Computing Platform: Thomas SCHULTHESS, Colin McMURTRIE
- WP7.6 Management and Coordination: Thomas LIPPERT, Boris ORTH
Hahne, J., Helias, M., Kunkel, S., Igarashi, J., Bolten, M., Frommer, A., & Diesmann, M. (2015). A unified framework for spiking and gap-junction interactions in distributed neuronal network simulations. Frontiers in Neuroinformatics, 9, 22.
Tejedor, E., Becerra, Y., Alomar, G., Queralt, A., Badia, R., Torres, J., Cortes, T., & Labarta, J. (2015). PyCOMPSs: Parallel computational workflows in Python. International Journal of High Performance Computing Applications, doi: 10.1177/1094342015594678.
Nowke, C., and Zielasko, D., Weyers, B., Peyser, A. Hentschel, B., & Kuhlen, T.W. (2015). Integrating Visualizations into Modeling NEST Simulations. Frontiers in Neuroinformatics, 9:29.
Zacharatou, E.T., Tauheed, F., Heinis, T., & Ailamaki, A. (2015). RUBIK: efficient threshold queries on massive time series. Proceedings of the 27th International Conference on Scientific and Statistical Database Management:18.