The Jülich Supercomputing Centre (JSC), through the John von Neumann Institute for Computing (NIC), the Gauss Centre for Supercomputing (GCS) and the Partnership for Advanced Computing in Europe (PRACE), provides state-of-the-art supercomputing resources, IT tools, methods and know-how to researchers in more than 200 German and European projects. To ensure an optimal mapping of expertise required by users of high-end supercomputers, JSC provides advice through its Algorithms, Tools and Methods Labs (ATMLs) in the fields of mathematical methods and algorithms, performance analysis or visualisation and through community-oriented high-level research and support teams, the Simulation and Data Laboratories (SDLs), which support applications in various fields of the natural sciences, ranging from biology to physics, materials science and climate research.
Role in the project
JSC contributes to two work packages in ESiWACE3. In WP1, JSC contributes to establishing a standard software stack for Earth system modelling. Together with BSC, CSC and ATOS, JSC will use automated build tools such as EasyBuild and Spack to improve ESM-related tools and software installation and deployment at the different sites. In WP5, JSC supports the networking of ESM training activities of the ESiWACE partners and the wider community. JSC contributes to the organisation of hackathons and summer schools focusing on exploiting new exascale systems, particularly the upcoming JUPITER machine in Jülich.
List of people involved
Lars Hoffmann (WP1 co-leader), Martin Schultz, Ling Zou
Relevant infrastructure and services
JSC currently operates two powerful supercomputers in Europe - JUWELS and JURECA-DC. JUWELS is a modular supercomputer that combines CPUs and GPUs for various computing tasks. The JUWELS Booster is powered by 936 BullSequana XH2000 nodes, each equipped with four NVIDIA A100 GPUs, which are hosted by AMD EPYC Rome CPUs. The compute nodes are connected with HDR-200 InfiniBand in a DragonFly+ topology. JUWELS has a peak performance of 73 petaflops. JURECA-DC has a peak performance of 14.98 petaflops and is equipped with AMD EPYC 7742 processors, NVIDIA A100 GPUs and Mellanox HDR InfiniBand DragonFly+ network. JSC provides various support services to help researchers optimise their use of these supercomputers.