Capella HPC Cluster

The Capella HPC cluster is a state-of-the-art resource designed for advanced AI research and high-performance computing (HPC) applications. Installed in 2024, it ranks #51 on the TOP500 and #5 on the GREEN500 for exceptional performance and energy efficiency. The cluster consists of 144 nodes, each equipped with 4 NVIDIA H100 GPUs, and is fully integrated into the HPC infrastructure of the Center for Information Services and High Performance Computing (ZIH). Capella incorporates state-of-the-art technologies such as NVIDIA’s Multi-Instance GPU (MIG) for scalable resource allocation and the cat filesystem, optimized for the high-performance I/O demands of AI and Machine Learning workflows. Its robust configuration and modern architecture make it an ideal platform for both exploratory research and large-scale computational projects.

Alpha Centauri HPC Cluster

The Alpha Centauri HPC cluster provides powerful computational resources tailored for ScaDS.AI Dresden/Leipzig researchers. With a total of 272 NVIDIA A100 GPUs, it is engineered to handle demanding AI and machine learning workloads.

Unified Software Stack

Both Capella and Alpha Centauri clusters share the same software stack, enabling seamless transitions between the two systems. This standardized environment supports popular machine learning frameworks such as TensorFlow and PyTorch, ensuring researchers can efficiently develop and execute their AI workflows. The shared stack ensures a consistent user experience, allowing projects to be easily moved between clusters based on computational demands.

Access to the HPC Clusters

Members of ScaDS.AI Dresden/Leipzig can access the Capella and Alpha Centauri clusters through a streamlined application process via the HPC Project Application Portal. Applicants are required to submit a brief description of their research project and resource needs.

For researchers without ZIH credentials, login details can be requested by contacting the Service Desk of TU Dresden or the IT support of ScaDS.AI Dresden/Leipzig at Leipzig University.

More Information about the HPC Cluster

• HPC documentation of ZIH including the Introduction to HPC at ZIH (slides)
• About HPC at ZIH

In case of technical questions, please contact the HPC support.