HPC-researchcluster

• Term: Funded in 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 455622343

The specialization of HPC systems and the bundling of common requirements of the various research groups have led to the consolidation of previously independent efforts and hardware operation. This not only exploits economies of scale, but also the fact that the effort required for the design, procurement and operation of a cluster inversely correlates with it's size. It also allows a much more efficient use of resources, since idle times that usually occur in individual systems are filled by other users. The scheduling result also typically improves with an increasing number of different types of jobs. This generates added value, which makes it possible to acquire a few significantly larger systems delivering higher performance at the same cost as many smaller systems with significantly lower overall performance. Additionally, mechanisms such as "Fairshare" make it possible to provide virtual shares of the overall system in almost any denomination. This lowers the entry threshold especially for smaller groups and junior researchers. For large teams applying for traditional large scale research systems grants of the DFG, the added value lies in a rapid start to their research projects, a potentially greater resource than individual procurement without the disadvantages of the requirement of an in-house system administration team. Staggered procurement and regular renewal cycles ensure that the attractiveness of the resources is maintained in the long term and the participation of new user groups can be ensured. These developments changes the landscape of large research infrastructures in the long term.The objective of the present proposal is the continuation of a computer system jointly used by researchers of the state of Baden-Württemberg in the fields of elementary particle physics, neuroscience, microsystems engineering and materials science (ENM) with significantly expanded, demand-oriented capacity and adaptation to current technological developments. This includes new capabilities in the field of processors as well as the use of computational accelerators. Current storage technologies are of particular importance against the background of the BaWü data federation which is currently being set up and the Storage-for-Science project implemented in cooperation with the computer center of the University of Tübingen.The ENM community in Baden-Württemberg will continue to require computing resources in the years to come, as demonstrated by the numerous supplementary sheets from the research groups.

• DFG Programme: Major Research Instrumentation
• Major Instrumentation: bwForCluster NEMO2 HPC-Forschungscluster für Neurowissenschaften, etc
• Instrumentation Group: 7040 Vektorrechner
• Applicant Institution: Albert-Ludwigs-Universität Freiburg
• Leader: Professor Dr. Gerhard Schneider