The GWDG has a long history of providing supercomputing resources to the researchers and students of the University of
Göttingen and of the Max Planck Institutes. The Scientific Compute Cluster (SCC) is constantly
extended and upgraded through diverse funding rounds and sub-systems of research groups.
In addition, the GWDG hosts two supercomputers for different user groups. Since 2018 the GWDG operates the Tier-2 system
Emmy (HLRN-IV). Since October 2020, the second phase has been in operation, which made it necessary to move to a modular
data center. Since 2021 GWDG and the University of Göttingen form one of eight Centers for National High Performance
Computing (NHR Centers) nationwide, making Emmy an NHR system, being operated for the
northern German states of the HLRN.
The NHR systems were augmented at the new Göttingen datacenter site by the GPU cluster Grete and a third phase for Emmy in April 2023 and May 2024, respectively.
In June 2018, the University of Göttingen, the German Aerospace Center ("Deutsches Zentrum für Luft- und Raumfahrt" -
DLR) and the GWDG signed a declaration of intent regarding a closer cooperation in compute-
and data-have intensive research. The DLR HPC resources are now distributed over two sites: one in Dresden, one in
Göttingen.
Our local system with an integrative, heterogeneous structure. This system is ideal for HPC beginners and small-scale applications as well as for advanced users, tuning their applications for our Tier-2 system.
Features
- 18,376 CPU Cores
- 56 Nvidia RTX5000
- 12 Nvidia GTX1080
- 20 Nvidia GTX980
- 16 Nvidia V100 32GB
- 99 TB RAM
- 1.4 TiB GPU RAM
- 5.2 PiB Storage
Users
- Researchers of the Max Plank Society
- Researchers of the University of Göttingen
Documentation
The NHR-supercomputer, operated for the northern states of the HLRN, offers the users 1423 compute nodes with a total compute power of 5,95 PetaFLOP/s. Advanced users and large-scale (GPU-) applications can take full advantage of this power horse.
Features
- 5.95 PetaFlop/s (Phase 1+2)
- 478 TB RAM (P1+2), 168 TB RAM (P3)
- 111,464 (P1+2), 36,288 (P3) CPU cores
Users
- HLRN- and NHR-user community.
Documentation
The NHR-supercomputer, operated for the northern states of the HLRN, offers the users 106 compute nodes with a total compute power of 5,46 PetaFLOP/s. Advanced users and large-scale (GPU-) applications can take full advantage of this power horse.
Features
- 5,46 PetaFlop/s
- 144 Nvidia A100 40GB
- 276 Nvidia A100 80GB
- 12 Nvidia V100 32GB
- 27.5 TiB GPU RAM
Users
- HLRN- and NHR-user community.
Documentation
The DLR-supercomputer CARO offers 1364 compute nodes with a total compute power of 3.46 PetaFLOP/s. The system is dedicated to DLR- applications and thus highly optimized to fit their large-scale and CPU-intensive needs.
(Imagerights CARO: DLR (CC BY-NC-ND 3.0))
Features
- 175,744 CPU Cores
- 364 TB RAM
- 3.46 PetaFlop/s
Users
Documentation