GCS LARGE-SCALE PROJECTS

Large-scale projects and highly scalable parallel applications are characterised by large computing time requirements, not only for short time frames but often for longer time periods. Projects are currently classified as "large-scale", if they require at least 25,000 node-h on Hunter, (100 Mcore-h on Hawk up to the 33rd call), or 45,000 EFLOP on JUWELS, or 45 Mcore-h on SuperMUC-NG. These values correspond to 2% of the systems’ annual production in terms of estimated availability. The call for GCS Large-Scale Projects is issued twice a year and approved projects start on 1 May and 1 November, respectively.

For an overview of approved GCS Large-Scale Projects, please chose from the list below.

Winner Projects of the Gauss AI Compute Competition

The Gauss Centre for Supercomputing has partnered with federal and state government funding agencies to promote the development of new generative artificial intelligence models through a competition leveraging the power of the Jülich Supercomputing Centre’s exascale supercomputer, JUPITER.

To further promote the AI uptake in German research and industry, the Gauss Centre for Supercomputing (GCS), along with the German Federal Ministry for Research, Technology and Space, the State Ministry for Culture and Science of the State of North Rhine-Westphalia, the Ministry of Science Research and the Arts of Baden-Württemberg and the Bavarian State Ministry of Science and the Arts held the “Gauss AI Compute Competition.”

Here, we present the list of the winners and approved projects of the Gauss AI Compute Competition:

  • "OneProtGPT: Integrating diverse protein modalities with scientific large language models for enhanced annotation and design"
    Dr. Alina Bazarova
    Forschungszentrum Jülich GmbH
     
  • "ReFoRMo: Reproducible open foundation reasoning models with strong generalization"
    Dr. Jenia Jitsev
    Forschungszentrum Jülich GmbH
     
  • “Advancing Open-Source Language Models through Efficient Pretraining"
    Prof. Dr. Bernhard Schölkopf
    Max-Planck-Institut für Intelligente Systeme
     
  • “Global Kilometer-scale Probabilistic Data-driven Modelling”
    Dr. Ioan Hadade
    European Centre for Medium-Range Weather Forecasts (ECMWF)
     
  • "CRESCENDO - Cell state and regulatory sequence encoding across all tree-of-life domains"
    Prof. Dr. Julien Gagneur
    Technische Universität München
     
  • "European Large-scale Omni-modal Foundation Model"
    Dr. Almut Sophia Koepke
    Technische Universität München 
     
  • "Holistic Multimodal Egocentric Video Forecasting"
    Prof. Dr. Juergen Gall
    Universität Bonn
     
  • "M3: The first Robot Foundation Model for Long-Horizon Multimodal Mobile Manipulation"
    Prof. Dr. Wolfram Burgard
    University of Technology Nürnberg
     
  • "MeshFoundation: Building a General-Purpose 3D Mesh Foundation Model via Extreme-Scale Context Transformers on JUPITER"
    Prof. Dr. Matthias Niessner
    Technische Universität München
     
  • “World Models with Multi-Scale Diffusion Transformers trained on Real World and Simulation-Generated Data for the Automotive Domain”
    Prof. Dr. Thomas Brox
    Universität Freiburg
go back