guest ::
| Home > Publications database > NIC Symposium 2025 Proceedings |
| Home > Publications database > NIC Symposium 2025 Proceedings |
| Book/Proceedings | FZJ-2025-01965 |
; ;
2025
Forschungszentrum Jülich GmbH Zentralbibliothek, Verlag
Jülich
ISBN: 978-3-95806-793-6
This record in other databases:
Please use a persistent id in citations: doi:10.34734/FZJ-2025-01965
Abstract: In a longstanding tradition, the John von Neumann Institute for Computing (NIC) holds biennial symposia, accompanied by proceedings volumes – illustrating the broad range of modern computational science and the advances in high performance and data-intensive computing. Symposium and proceedings thus provide a glimpse into supercomputingbased research at its best and make it accessible both to the general public and to computational scientists across disciplinary boundaries. As such they foster exchange between different fields of natural science and engineering with respect to modern algorithms and computational strategies. To this end, on March 6th and 7th, 2025, computational scientists will again convene in Jülich for the 12th NIC symposium. We are very pleased that this time it is again possible to showcase the breadth of high-performance computing research supported by the NIC with contributions from astrophysics, elementary particle physics, and statistical physics of hard and soft condensed matter, computational chemistry and materials science, as well as computer science, fluid mechanics, and earth system modelling – covering both fundamental research and projects with a strong application orientation. We are also delighted to extend a very warm welcome to our colleagues from the Goethe University Frankfurt who have joined the NIC in 2024 as a new member institution. Together, we will further strengthen research in the field of computational science in Germany and Europe. The NIC continuously provides the scientific community with essential high-performance computing resources and training. Within the framework of the Gauss Centre for Supercomputing (GCS), the Jülich Supercomputing Centre (JSC) has been operating the modular supercomputer JUWELS (Jülich Wizard for European Leadership Science) since 2020, which is composed of a CPU-based cluster and a GPU-based booster module. Thanks to the excellent training and user support by the technical experts from the JSC, the JUWELS architecture has been widely adopted across disciplines and communities. In particular, porting codes to the booster module and adapting algorithms to the GPU architecture has been fundamental in getting the disciplines ready for the next generation of GPU-based exascale computing. After the decision by the European High Performance Computing Joint Undertaking (EuroHPC JU) that the Forschungszentrum Jülich is to operate the first exascale supercomputer in Europe, the JSC and the GCS have been preparing for JUPITER (Joint Undertaking Pioneer for Innovative and Transformative Exascale Research). The new system will become available in 2025. To optimally prepare applications and users for JUPITER and to facilitate the transition from current petascale and pre-exascale supercomputers to actual exascale computing, the JSC has launched JUREAP, the JUPITER Research and Early Access Program. In the first phase of 2024, users have participated in the Scalability and Performance Evaluation Phase (SPEP), an open call to test and demonstrate the performance and the scaling of the applications on test architectures. In September 2024, the GCS Exascale Pioneer Call has been initiated with two objectives: the successful projects are given early access to JUPITER during build-up, approximately from January 2025 onwards, and the call distributes JUPITER resources for the time period after the machine is officially operational until the end of October 2025 – thus enabling groundbreaking computational research for the German scientific community. A more detailed overview on JUPITER, the new opportunities that exascale computing opens up to all scientific communities, and in particular the shifts driven by the wave of developments in AI technologies and large foundation models are provided in the introductory article of the proceedings by Thomas Lippert and coauthors "Paradigm Change or Riding the Wave? Exascale-Computers to Train Foundation Models”.
; 
|
The record appears in these collections: |
PDF