AI HPC Cluster

• Term: Funded in 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 528288777

The Julius-Maximilians-University of Würzburg (JMU) is a node in the AI network of the Free State of Bavaria within the framework of the Hightech Agenda Bavaria with the newly founded Center for Artificial Intelligence and Data Science (CAIDAS). CAIDAS conducts basic AI research and data science; of the planned 26 professorships, 16 professorships have already been filled; furthermore, five procedures are currently underway, the remaining five are in preparation; the goal is to fill all chairs by the end of 2023. In order to meet the resulting current and rapidly increasing demand for computing resources, especially in the GPU area, an AI-HPC cluster with AI accelerator cards is to be set up at JMU. In addition to CAIDAS, this system will also be available to all other researchers at JMU and will replace the predecessor cluster "Julia". The GPU orientation of the system, which has changed compared to the predecessor system, will additionally accommodate the AI methods now used in many disciplines with a focus on machine learning, thus leveraging further synergies. With the users from CAIDAS, the system is expected to be used by over 600 users university-wide. To handle the enormous amounts of data involved in AI research, the system relies on a high-performance, distributed parallel storage system and fast network connectivity. The central installation and support of the cluster at the computer center will relieve the burden on research groups and leverage synergy effects through interdisciplinary exchange. The use of standard server hardware ensures that the system can be expanded as easily as possible.

• DFG Programme: Major Research Instrumentation
• Major Instrumentation: KI-HPC-Cluster
• Instrumentation Group: 7000 Datenverarbeitungsanlagen, zentrale Rechenanlagen
• Applicant Institution: Julius-Maximilians-Universität Würzburg
• Leader: Matthias Funken