Linux High Performance Compute Cluster - Athene III

• Term: Funded in 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 566433262

The demand for computing power has been steadily increasing since the simulation of natural processes was established as a recognized methodology in many scientific disciplines. This is reinforced by the increased knowledge gained from AI or ML approaches, which are becoming increasingly established as interdisciplinary methods. An HPC system is therefore an important component of modern research infrastructure and the renewal of the HPC cluster Athene is of central importance for the scientific competitiveness of the university. The applications of the new HPC system are spread across many research groups: starting with the physics of elementary particles, where quantum chromodynamics (QCD) with the theory of strong interactions is being investigated in particular, continuing to the theory of condensed matter with spintronics, but also going beyond this to molecular electronics more generally, whose functional units could be specially synthesized molecules. Statistical bioinformatics focuses on the modeling of biological networks, genomic data integration and the analysis of high-dimensional data. In structural bioinformatics, numerous computer-aided methods are used to gain cross-scale, time-resolved insights into the structure and dynamics of molecular machines. To this end, algorithms of molecular mechanical simulations, quantum chemical calculations and artificial intelligence are combined with data from biophysical and biochemical experiments. Molecular dynamics deals with the real-time modeling of photochemical reactions, as well as with time-resolved spectroscopic experiments and in particular with photobiological systems. Machine learning is used to predict molecular properties with a focus on photochemical properties and vibronic spectra. Projects are also planned in the field of organometallic chemistry and homogeneous catalysis. Theoretical ecology develops statistical methods to develop theories about the organizational principles of ecosystems and to test them with data. Further applications arise from genetic epidemiology with the identification and characterization of risk factors for complex diseases. A wide variety of software packages and algorithms are used. The breakdown of GPU and CPU performance, including InfiniBand connectivity, reflects these demanding requirements of "classic" and AI and ML applications, while requiring high-performance access to large amounts of data, too. The requirements range from very IOPs-heavy applications, for which a central NVMe SSD scratch area via Infiniband is indispensable, to large amounts of data on HDD storage media accessed in streaming mode, to infrequent access to data for backup and archiving.

• DFG Programme: Major Research Instrumentation
• Major Instrumentation: Linux High Performance Compute Cluster - Athene III
• Instrumentation Group: 7030 Dedizierte, dezentrale Rechenanlagen, Prozeßrechner
• Applicant Institution: Universität Regensburg
• Leader: Dr. Christoph Bauer