An app on UCloud is an application that runs within SDU’s secure infrastructure. Apps can be either interactive (e.g. Chat UI, Coder, or Transcriber) or batch-based.
When you start an app, you create a compute job on Interactive HPC. Each job requires compute resources, which are aquired through the resource application process. For more information, see Apply for Resources.
For the technically curious: an app is a Docker container running in a Kubernetes cluster.
A job is a running instance of an application requiring its own resources. See Apply for Resources.
When you launch an app, UCloud allocates CPU and optionally GPU resources for you. You can monitor, stop, or restart your jobs from the Jobs page.
You can switch between Workspaces in UCloud using the Workspace menu in the upper-right corner. When you switch workspaces, the UCloud interface updates to show the relevant resource allocations, job history, and collaboration features for the selected workspace.
My Workspace is your personal workspace in UCloud. It provides a limited annual allocation of CPU hours. This workspace is intended solely for personal experiments and small-scale learning activities.
A Project is a shared workspace associated with a specific research or teaching project. It is created as part of the resource application process.
A project enables multiple members to collaborate, share storage, and use the allocated compute resources within a common project environment.
The system has a finite number of CPU and GPU cores available. As a result, it uses a job queue to ensure that jobs wait until sufficient resources become available, at which point they are started automatically.
In general, the fewer resources you allocate to your app, the shorter the queue time is likely to be.
Most apps offer a range of configuration options.
You can:
You can also launch a Linux-based environment and build your own setup from the ground up.
You need to apply for GPU resources for a project. When they are granted the option to select a machine type with GPUs will become available.
Closing your browser does not stop your job.
Interactive apps continue running until you either stop them manually or the allocated wall time expires.
Exception: If you run a command directly in a terminal without using a tool such as tmux or screen, the process will terminate when the connection is closed.
Remember to stop jobs you are no longer using. This helps conserve your allocated resources and makes shared compute resources available to other users.
All applications run in SDU’s data centre under institutional governance and in accordance with a data processing agreements between SDU and Aarhus University.
Data processed in UCloud:
This makes UCloud well suited for research and teaching involving sensitive or confidential data.
Students cannot apply for UCloud resources themselves. However, staff members can apply for resources to support activities such as bachelor’s projects, master’s theses, or courses.
For more information, see Apply for Resources.
Data analysis tools
Machine learning frameworks
AI language model interfaces
Development environments
Linux-based custom environments
Traditional batch compute applications
A full indexed list is available on the The Applications page.
Documentation is available directly from:
Each app’s dedicated UCloud documentation page
You’ll find guides on launching apps, submitting jobs, monitoring resources, and best practices.
Tutorials are available here: Tutorials – Interactive HPC
