JUPITER Supercomputer to Power AI Research in New Gauss AI Compute Competition
Jan. 17, 2025 -- The German Federal Ministry of Education and Research (BMBF), the Ministry of Culture and Science of the State of North Rhine-Westphalia (MKW NRW), the Ministry of Science, Research and Arts Baden-Württemberg and the Bavarian State Ministry of Science and the Arts are launching a competition in the field of artificial intelligence (AI) for research groups and companies on the first European exascale supercomputer JUPITER.
The competition is being organized by the Gauss Centre for Supercomputing (GCS) together with the Jülich Supercomputing Centre (JSC). The aim is to advance the development of new generative models with artificial intelligence (AI). The winning projects will be allowed to use the future exascale supercomputer JUPITER at the Jülich Supercomputing Centre (JSC). The call for submissions starts on January 17, 2025 and ends on March 17, 2025 at 5:00 p.m. CET.
JUPITER is the first European exascale supercomputer to be launched at JSC at Forschungszentrum Jülich. It is the first system in Europe to exceed the computing power of one quintillion computing operations per second. JUPITER is expected to achieve one exaFLOP/s with double precision (64 bits), which is typically required for scientific simulation calculations. For training AI models, JUPITER is expected to achieve even more than 70 ExaFLOP/s – for calculations that require lower precision (8 bits). This will make JUPITER one of the fastest AI computers in the world and help to strengthen Germany's leading role in the international development and application of AI technologies.
As part of the ‘Gauss AI Compute Competition,’ researchers will have exclusive access to JUPITER for a limited period of time. A total of around 15 million GPU hours will be available for this purpose once the exascale computer begins operations. Researchers who are successful in the competition can use these resources to train new fundamental models for AI applications that contribute to important research areas. In particular, those projects that use generative AI models as part of their research and development activities can benefit.
The Gauss AI Compute Competition is designed to promote projects that not only make effective use of JUPITER's powerful technology but also demonstrate a high level of societal impact to ensure knowledge transfer and practical implementation.
Interested scientists working at German universities, colleges or other research institutions are invited to participate. Commercial companies that work with German educational and research institutions can also participate, as can German federal, state and local administrations and other social organizations. The sponsors of the competition encourage both experienced HPC researchers and those new to the field to apply. To help applicants prepare their proposals, GCS offers test access to their HPC systems.
The competition supports the development and expansion of the German AI ecosystem, strengthens technological sovereignty in Germany and Europe, and increases efficiency in science and industry. It thus creates sustainable added value for the German and European research landscape and for society as a whole. The competition is also an important building block for implementing the next phase of the BMBF program ‘High-performance computing for the digital age: Research and investment for HPC.’
JUPITER is funded half by the European supercomputing initiative EuroHPC JU and a quarter each by the Federal Ministry of Education and Research (BMBF) and the Ministry of Culture and Science of the State of North Rhine-Westphalia (MKW NRW) via the Gauss Centre for Supercomputing (GCS).
The opportunity to submit project proposals starts on January 17, 2025 and ends on March 17, 2025, 17:00 CET. Applications for participation can be submitted here:
https://jards.gauss-centre.eu/gaussaicomputecompetition
For more information about the competition, please contact the GCS Coordination Office at JSC at: coordination-office@gauss-centre.eu.
Source: HLRS and GCS
