Using DataHub

DataHub is a campus-wide JupyterHub service, and there is a dedicated instance for CDSS Discovery students at https://cdss-discovery.datahub.berkeley.edu. It runs on NRP Nautilus. Our goal is to provide students with access to GPUs and more computing than they may have available on their personal devices.

Access¶

CDSS Discovery students can log into the hub without having to request access. Only those currently enrolled in the program can login.

Features¶

The hub supports JupyterLab, RStudio, and a number of other applications.

Each user has access to a GPU. The NRP cluster policies limit interactive jobs such as Jupyter servers to 16 CPU cores, 32 GB RAM, and a duration of 6 hours.

Software libraries like torch and tensorflow are already installed. You can install additional software in your active session. If you would like software to be permanently installed, you can either ask us to alter the user environment, or you can install the software into your home directory. For the latter, you will need to be familiar with the various software installers and the implications of how software in your home directory may interact with software installed on the system.

Support¶

If you need support using the hub, please contact us.