Home

This is the user documentation for the AI Cloud at Aalborg University, Denmark

What is AI Cloud?

The AI Cloud can be thought of as a small-scale supercomputer. Technically, it is a high-performance computing (HPC) cluster equipped with general-purpose graphics processing units (GPUs).

The AI Cloud is a facility consisting of several servers designed to be ideal for training deep learning algorithms. Deep learning is in many cases branded as artificial intelligence (AI) - hence the name AI Cloud. This also makes the facility good for a wide range of computationally intensive work such as numerical simulations and high-performance data analysis (HPDA). See "Overview" in the menu above for more details on what the AI Cloud consists of.

The AI Cloud is a facility at Aalborg University managed by the CLAAUDIA team.

Who can use the AI Cloud?

All researchers at AAU can use the AI Cloud. For further information on how to obtain access to AI Cloud, see CLAAUDIA's homepage.

Students at Aalborg University can also use the AI Cloud. As a student, you are granted access to the AI Cloud for one semester at a time in relation to your semester project, if the project requires the AI Cloud's resources. This requires approval from your supervisor.

What can I use the AI Cloud for?

The AI Cloud can be used to run basically any Linux-based application that you wish. The application must, however, be able to utilise the GPUs.

If your application does not require a GPU, other platforms are better suited instead. Feel free to contact CLAAUDIA for guidance on more suitable alternatives.

Typical examples of applications that you can run in AI Cloud are: deep learning applications using TensorFlow, PyTorch, or other deep learning frameworks; applications built using NVIDIA CUDA or higher-level libraries such as cuDNN, cuBLAS etc.; any numerical simulation software that can utilise GPUs for computing. See also "Additional examples" in the menu above for examples.

Currently AI Cloud is suitable for working with data classified as level 0 and 1 on the data classification model. Data classified as level 2 or 3 is therefore not supported as of now, but we plan on doing so in the future. If you need GPU-ressources to process data on levels 2 and 3, please reach out to the CLAAUDIA-team and we will help you find a solution. This could involve setting up a separate drive for your data, or to help you get acccess to one of the larger HPC-facilities outside of AAU.