Skip to content

Compute

Empa researchers have access to compute resources covering the full range, from high-performance code running on your Empa workstation to the Swiss National Supercomputing Centre. The Scientific IT team at Empa manages these compute resources and assists you in finding the optimal solution for your project’s compute needs.

Your Empa workstation and Empa’s local compute resources are often the best starting point for your project. We help you with setting up, configuring, and running high-performance code locally, from preparing a workstation for heavy computations to running parallel jobs on local hardware.

Empa has a contract with the Swiss National Supercomputing Centre (CSCS) as the main provider of large-scale compute resources. CSCS operates the Alps research infrastructure, giving every Empa laboratory access to GPU-accelerated and CPU-only supercomputing, machine learning resources, and storage.

The CSCS LLM Inference API service is a managed and sovereign service for running large language models. It offers OpenAI- and Anthropic-compatible endpoints for a curated selection of open-weight models, and your data is processed entirely within CSCS infrastructure in Switzerland. See the CSCS page for more details.

Euler is the central high-performance computing cluster of ETH Zürich. Empa currently has an allocation on specific nodes of the cluster, which is available to Empa researchers. Note that we are still in a test phase with Euler: if you would like access, please contact us.

If you would like to learn more about any of these resources, or are unsure which one fits your project best, please get in touch with us via the contact page.

CSCS

Learn about accessing the CSCS supercomputers, pricing options, available resources, and how to get started.

Euler

Learn about accessing the Euler cluster at ETH Zürich, getting an account, available resources, and how to get started.

Compute Support

Find out where to get help with compute: support from Empa Scientific IT and system-specific support from the compute providers.