Skip to main content

Matlab

Matlab is a numerical computing environment from MathWorks. On the DSRI it runs as a container with a desktop UI accessible directly in your web browser.

info

We are not experts in Matlab - if you have issues with the official Docker image, feel free to contact MathWorks support directly (login with your UM account first). Since Matlab is closed source, we cannot fix issues in the image itself.

Deploy​

Find the Matlab template in the DSRI Catalog (make sure the Templates checkbox is checked) and instantiate it with the following parameters:

Once the pod is running, you can access Matlab through two routes visible when clicking on the MATLAB node in the Topology view:

  • The matlab route - access the Matlab desktop UI directly in your browser. This is the recommended option.
  • The matlab-vnc route - access Matlab using a VNC client. Only use this if you know what you are doing.

Persistent storage​

Use the /home/matlab/persistent folder to store data that should be preserved across pod restarts.

To navigate to the persistent volume from the terminal:

cd /home/matlab/persistent

Particularities​

Access the terminal​

To open a terminal inside your Matlab pod, either go to the pod page in the DSRI web UI, or connect from your local terminal:

oc rsh <matlab-pod-name>

Type bash when first accessing the terminal for a better experience. Then type matlab to start MATLAB from the terminal.

Use Matlab in Jupyter​

To run Matlab in a Jupyter notebook you first need to install MATLAB inside your Jupyter pod. Contact the DSRI team if you need assistance with this.

Deploy on GPU​

The Matlab template in the DSRI Catalog can also be used to deploy a pre-built Nvidia Matlab Deep Learning Container on GPU nodes. See the official MathWorks documentation for more details about this image.

caution

GPU access is not enabled by default. You need to make a reservation first - see the GPU documentation for details.