Opening the event, Professor Maren Bennewitz, Vice Rector for Digitalization and Information Management, said: “Marvin has been driving innovation in our research at theUniversity of Bonn for two years now. Even just over the past year, however, it has brought about further palpable changes to our academic and scientific work. Whether it’s been complex simulations, data-intensive analyses or top-level research in artificial intelligence and machine learning, Marvin is enabling our researchers to do things that they couldn’t have imagined doing two years ago.”
Referencing the University’s successful retention of its University of Excellence status and record-high eight Clusters of Excellence, she added: “It goes without saying that digital infrastructure like Marvin is playing a key role. Marvin is a shining example of our mission to combine scientific excellence with cutting-edge infrastructure. With high-performance computing, we’re laying the groundwork for growing scientific innovation even further at the highest level.”
Since it began work in 2024, 675 researchers from 183 research groups spread across the University’s seven faculties have made use of Marvin’s CPU and GPU clusters. They put the supercomputer to work in various ways, such as for AI-powered methods for developing new drugs—an approach that has already laid the foundations for a joint cancer research project worth several million euros. In addition, Marvin is opening the door to training complex neural networks, e.g. for autonomous driving, robotics and sustainable applications in agriculture and global nutrition, as well as analyzing climate and satellite data and simulating sustainable battery systems. It is also playing a key role in theoretical chemistry, physics and astrophysics, including the study of black holes.
What is more, the infrastructure forms a vital component of the University of Bonn’s Clusters of Excellence, which include PhenoRob, Color meets Flavor and Dynaverse, and of other large-scale collaborative projects making intensive use of Marvin, such as DETECT and NuMeriQS—two Collaborative Research Centers funded by the German Research Foundation (DFG).All in all, Marvin often prepares the ground for publications and new approaches in research and enables projects to happen that would simply not be possible without this infrastructure.
Talks highlight the sheer breadth of research
The researchers demonstrated just how broad the range of applications for high-performance computing is at the University of Bonn in 10 brief talks. This year’s Marvin Trophy went to the team led by Mohamad Hakam Shams Eddin and Professor Jürgen Gall for their new AI model, “RiverMamba,” which enables flood risks to be predicted more precisely than previous methods. “The event illustrated once again how the HPC infrastructure and Marvin are crucial to excellent research at the University of Bonn,” Gall said. “I really liked the transdisciplinary nature of the event; discussing potential applications and solutions with colleagues and doctoral students proved very interesting.” He plans to use his allotted time with Marvin to train his model at a higher resolution and thus make better forecasts on a local scale.
Second place went to Katrin Drysch from the Mulliken Center for Theoretical Chemistry. She has been harnessing the possibilities presented by high-performance computing to devise a workflow for creating nanoplastic particles, which can be used to study how nanoplastics interact with various environments such as the human body and the sea.
The team led by Professor Michael Hölzel from the Institute of Experimental Oncology and Privatdozent Dr. Gregor Hagelueken from the Institute for Structural Biology are using AI models to design new proteins that can be harnessed to treat cancer or fight viruses; high-performance computers like Marvin can generate thousands of theoretically suitable designs that can then be tested in the lab. These efforts secured them third place in the audience vote. Besides extra allotted time with Marvin, the prizes also included closer support from the HPC support team in some cases.
One highlight was the keynote speech by Dr. Georg Hager, Head of the Research Division at the National High Performance Computing Center, which is part of Friedrich-Alexander-Universität Erlangen-Nürnberg. He showed how a few simple approaches can be used to better understand and analyze code performance and optimize it in a targeted way. This was accompanied by several intensive workshops, opportunities for dialogue and a hackathon, where participants were able to make practical improvements to their programs with Hager’s expert assistance.
Vital support for researchers
The HPC team at University IT is responsible for the supercomputer’s technical operation and helping users to get the most out of it. Team lead Dr. Dirk Barbi stressed the importance of having powerful infrastructure: “With Marvin, we’ve created a stable central platform that allows researchers from all manner of different fields to access high-performance computing. We’re seeing a sharp rise in demand, especially where artificial intelligence and data-intensive simulations are concerned.”
The High Performance Computing and Analytics Lab (HPC/A Lab), which is based at the Digital Science Center, supports the researchers with scientific aspects and methods. Its director, Professor Petra Mutzel, explained: “Marvin is at the heart of our mission to put high-performance computing at our researchers’ fingertips. That Bonn has retained its status as a University of Excellence shows what huge potential there is here, and we’re proud to be doing our bit. The projects we’ve learned about today demonstrate in impressive fashion how much scientific potential is packed into this infrastructure.”
For instance, Marvin is playing an important role in the Modelling Transdisciplinary Research Area (TRA), a research alliance that combines mathematical and computer science methods with applications drawn from numerous other disciplines, from life sciences and medicine through to geosciences and economics. Its speaker, Professor Jürgen Gall, explained: “Many of the questions that the researchers in the TRA Modelling are tackling involve complex systems with many components interacting with one another. You need modeling, simulations and high-performance computing in order to understand these kinds of systems, so Marvin is a crucial element of many projects in the TRA.” A look back at last year’s event shows that transdisciplinary research approaches like these can secure targeted funding. For instance, the Polyglot project was given financial support by the TRA Sustainable Futures and succeeded in developing a leading language model for Portuguese thanks to its extra allotted time with Marvin.
Marvin the supercomputer: facts, figures and dates
With 320 NVIDIA GPUs, 18,400 CPU cores and a connected load of 430 kW, Marvin is among the most powerful academic supercomputers in Germany. Two thirds of its capacity has been specifically optimized for machine learning and AI applications, something of which no other university supercomputer can boast. The HPC team at University IT has already deployed Marvin to help 675 registered users from 183 research groups since it began operating on March 11, 2024, during which time the team members have processed over 1,900 support requests and held no fewer than 70 workshops (figures correct as of the end of 2025).
Ansprechpersonen für die Medien:
Prof. Dr. Petra Mutzel (HPC/A Lab) for research inquiries:
Director of the High Performance Computing and Analytics Lab
Phone.: +49 (228) 73-69917
Mail: pmutzel@uni-bonn.de
Dr. Dirk Barbi (HPC/University IT) for technical inquiries:
HPC Team Lead
Phone: +49 (0)228 73-66136
Mail: dbarbi@uni-bonn.de
Part of the Digital Strategy
The University of Bonn’s Digital Strategy sets out the measures and structures required for its digital transformation. The Marvin project formed part of the Infrastructure for Research, Teaching and Services set of objectives.
