...
| Info |
|---|
We are currently working on an experimental JupyterHub which will allow you to log on to the login nodes and run notebooks directly from the browser. If you want to test this out, you can try here: http://albedo0.dmawi.de:8000 Note that VPN is required! Currently access is provided only by request, please open a ticket on hpc@awi.de to get on the list during the testing phase. You will be presented with a login page, and after login, with a selection of job profiles. You can either run your notebook on a login node (not recommended), a compute node, or a GPU node. In case of a compute node or GPU node, yu need to specify which computing account SLURM should use. A list of available computing accounts is provided for you. Additionally, in the case of a GPU node, you need to specify which type of GPU you want to use (A40 or A100) and how many GPUs you wish to use. On the JupyterHub page (the interface you are provided after SLURM launches your job), you can select the Python3 kernel (bare-bones Python only) or the Analysis Toolbox kernel (most common scientific analysis and plotting packages). If you want to install your own kernel, please follow the instructions in HowtouseyourowncondaenvironmentinJupyterHub/Lab?%E2%86%92addaJupyterKernel? Add a Jupyter Kernel SLURM-enabled Jupyterhub jobs are restricted to 12 hours. |
...