NHR@ZIB
Berlin University Alliance
Berlin
The Zuse Institute Berlin (ZIB) is an interdisciplinary research institute for applied mathematics, data-intensive high-performance computing and artificial intelligence. Its research focuses on modeling, simulation and optimization with scientific cooperation partners from academia and industry.
Key Expertise and Profile
- Life Sciences, Earth System Sciences, Chemistry, Physics, Engineering, and Social Sciences
- Scientific computing, especially Artificial Intelligence, Machine Learning and Performance Engineering
- Expertise in reliable operation of large-scale HPC systems with heterogeneous architectures
- NHR Atomistic Simulation Center, together with NHR@FAU and PC2.
NHR@ZIB is engaged in several nationwide and international research alliances, like e.g. the Cluster of Excellence MATH+, the Forschungscampus MODAL, the CECAM node “Mathematics and Computation in Molecular Simulation“, and Labs together with the HPC hardware vendors.
Technical Specifications
| Center Name | NHR@ZIB |
|---|---|
| Total Resources: | |
| # CPU-Cores | 128 448 |
| # GPUs | 256 |
| Additional Information | |
| max. job run time [h] | 48 |
| jupyter notebook available [y/n] | n |
| CPU Resources: | |
| System 1 | CLX |
| CPU-Type | Intel Cascade Lake 9242 |
| # nodes | 838 |
| # Cores per node | 96 |
| memory per node [GB] | 384 - 1536 |
| System 2 | GENOA |
| CPU-Type | AMD Genoa 9654 |
| # nodes | 250 |
| # Cores per node | 192 |
| memory per node [GB] | 1536 |
| GPU Resources | |
| System 1 | A100 |
| GPUs per node | 4 x NVIDIA A100 |
| # nodes | 42 |
| memory per GPU [GB] | 80 |
| System 2 | H200 |
| GPUs per node | 4 x NVIDIA H200 |
| # nodes | 14 |
| memory per GPU [GB] | 141 |
| System 3 | PVC |
| GPUs per node | 4 x Intel Data Center |
| # nodes | 8 |
| memory per GPU [GB] | 128 |