The Alpha Centauri HPC cluster with a total of 272 powerful GPUs is available for ScaDS.AI researchers. The cluster consists of 34 nodes, each equipped with
- 8 Nvidia A100 GPUs,
- 2 AMD Epyc 7352 processors,
- 1 TB RAM and
- 3.5 TB local NVMe storage.
Machine Learning frameworks, such as TensorFlow and PyTorch, are already installed on the system. The cluster is operated by the Center for Information Services and High Performance Computing (ZIH) at TU Dresden.
Access to the Alpha Centauri HPC Cluster
For ScaDS.AI members, access to the system is granted in a straightforward process. A simplified HPC project application is required, which includes essentially a brief description of the planned research to justify the resource need and the required amount of hardware resources, including storage. In contrast to standard HPC projects at ZIH, a detailed project description is not required for ScaDS.AI projects. For application, please use the HPC Project Application Portal.
After logging in to the portal, choose New Project Application and then select SCADS Project. Once your HPC project is approved, users assigned to it can access the cluster via SSH or JupyterHub. Users that do not have a ZIH login yet should ask TU Dresden service desk for a ZIH login.
More Information about the HPC Cluster