Running Jupyter on Snowy¶

Info

You can run Python in jupyter-notebook or jupyter-lab, which provide a web interface with possibility of inline figures and debugging.
jupyter-lab is installed in the python modules >=3.10.8.
You may also install your own version of jupyter-notebook or jupyter-lab if you have certain version requirements.

1. Connect to Rackham using ThinLinc.¶

You may download the ThinLinc client from here: https://www.cendio.com/thinlinc/download

You may also use the web browser to connect: https://rackham-gui.uppmax.uu.se You need to set up 2FA for the ThinLinc web.

2. Open a terminal in ThinLinc and ask for an interactive session to Snowy.¶

salloc -A naiss2023-22-247 -p node -N 1 -M snowy --gpus=1 --gpus-per-node=1 -t 04:00:00

Note: We have a magnetic 4-node reservation for today: naiss2023-22-247_1.

Is the GPU visible?

echo $CUDA_VISIBLE_DEVICES

3. Load the needed modules¶

module load python_ML_packages/3.9.5-gpu
pip list

And start the Jupyter notebook:

jupyter-notebook --ip 0.0.0.0 --no-browser

4. Open the notebook.¶

Solution 1:¶

Click on any of the links. This will open a Jupyter notebook in a Firefox browser on Rackham.

Solution 2:¶

Instead of opening Jupyter in a browser in ThinLinc, you may use the browser on your computer. You need to use ssh-tunnelling for this:

ssh -L 8000:s175:8888 iusan@rackham.uppmax.uu.se

If you cannot open the notebook using the links above, try changing it to: localhost:<port>/?token<tokenID> Example: localhost:8000/?token=ab8a98df7b6606316849015daebaca5d2fa6331d32a47446