JUWELS
Modular supercomputer at Jülich with an AMD and Nvidia booster module
JUWELS is a modular supercomputer built by Atos and run at the Jülich Supercomputing Centre, part of Forschungszentrum Jülich in Germany. It went into service in 2020 as the successor to JUQUEEN, and its booster module reached seventh place on the November 2020 TOP500 ranking.
- Built by Atos, hosted at the Jülich Supercomputing Centre in Germany
- Booster module theoretical peak of 70.980 petaflops
- Seventh on the November 2020 TOP500 list; cluster module ranked 44th
- Uses AMD Epyc processors with Nvidia A100 GPUs for acceleration
- Top 15 on the Green500 list of energy-efficient supercomputers
What it is
JUWELS, short for Jülich Wizard for European Leadership Science, is a supercomputer hosted at the Jülich Supercomputing Centre within Forschungszentrum Jülich in Germany. Atos developed the machine, which entered service in 2020 and took over from the retired JUQUEEN system. Its design is modular: a Cluster Module based on Intel Xeon processors and a separate Booster Module that adds GPU acceleration. The booster reaches a theoretical peak of 70.980 petaflops. On the November 2020 TOP500 list the booster placed seventh worldwide, while the cluster module ranked 44th as a separate entry. The system supports research work across several scientific fields and also runs AI training benchmarks.
How it came about
JUWELS was built by Atos and placed at Jülich in 2020, replacing JUQUEEN, which had gone out of operation. Its debut on the November 2020 TOP500 list was strong: the booster module entered at seventh place globally and the Xeon-based cluster module at 44th. In 2021 the booster took part in MLPerf HPC, a training benchmark run by a consortium drawn from universities, research labs and industry that measures how well hardware, software and services handle AI training and inference. The machine also appeared in the top 15 of the Green500 ranking for energy efficiency.
The compute
The booster module pairs AMD Epyc processors with Nvidia A100 GPUs, which handle the accelerated calculations. A second module, the JUWELS Cluster Module, is built around Intel Xeon chips and is listed separately in rankings. Together the two form one modular system, with a theoretical peak of 70.980 petaflops quoted for the booster. The machine is used for AI work as well as simulation, having joined the MLPerf HPC training benchmark in 2021 alongside other large systems.
Where the GPUs are
JUWELS sits at the Jülich Supercomputing Centre, part of Forschungszentrum Jülich in Jülich, Germany. The centre operates the machine and runs research programmes on it, including work by its Simulation and Data Laboratory for Climate Science.
Customers and deals
The University of Edinburgh signed an agreement to use JUWELS for studies in particle physics, astronomy, cosmology and nuclear physics. Forschungszentrum Jülich's own climate science lab also runs workloads on the system.
Its place in AI
JUWELS is a European leadership-class system and the successor to JUQUEEN at Jülich. Its booster module reached seventh position on the November 2020 TOP500 list, with the cluster module at 44th, and the machine placed within the top 15 of the Green500 efficiency list. It has served as a platform for AI benchmarking through MLPerf HPC and for climate research, including tracking gravity waves in the atmosphere using radiance measurements pulled from NASA data servers.
Questions people ask about JUWELS
What is JUWELS?
It is a modular supercomputer at the Jülich Supercomputing Centre, part of Forschungszentrum Jülich in Germany. Atos built it and it started work in 2020, replacing the older JUQUEEN machine. The name stands for Jülich Wizard for European Leadership Science.
How is the machine structured?
JUWELS has two parts that are ranked separately. A Cluster Module uses Intel Xeon processors, while a Booster Module adds acceleration using AMD Epyc processors with Nvidia A100 GPUs. The booster alone has a theoretical peak of 70.980 petaflops.
How fast is it?
The booster module reaches a theoretical peak of 70.980 petaflops. When it first appeared on the November 2020 TOP500 list it ranked seventh in the world, and the cluster module ranked 44th on the same list.
Who uses JUWELS?
The University of Edinburgh contracted access for research in particle physics, astronomy, cosmology and nuclear physics. Jülich's Simulation and Data Laboratory for Climate Science also uses it to detect gravity waves in the atmosphere by computing on radiance measurements from NASA data servers.
Good reading and sources
The facts on this page come from the Wikipedia article "JUWELS", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryNeocloud rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.