The official OpenFWI tutorial was initially developed for the Google Colab website . We have made small modifications to ensure its compatibility with the Gadi ARE platform. Please follow the steps below to execute it.
Start a ARE Jupyterlab Session
- Login to are.nci.org.au with your NCI username and password.
2. Start a JupyterLab session by clicking the "JupyterLab" button.
3. Choose appropriate compute resources in multiple fields - see the section below.
4. Make sure you are requesting at least 1 gpu from 'gupvolta' queue.
Here We use 'nm05' as the default project so we fill 'nm05' in the 'Project' field and fill "gdata/up99+gdata/nm05+scratch/nm05" in the 'Storage' field.
Note: You need to replace 'nm05' with your own default project in above fields.
Choosing the appropriate compute resources
Click "advanced options" and
- Type in "/g/data/up99/modulefiles" in the "Module directories" box.
- Type in "openfwi/23.11" in the "Modules" box.
Click the "Launch" button to start a JupyterLab session.
Wait until your JupyterLab session is running and click "Open JupyterLab" button to open the JupyterLab session.
Set up OpenFWI tutorial environment
After entering your JupyterLab session, click "terminal" button to open a CMD line.
We need to build the tutorial environment via an iniltization script as below:
openfwi:23.11 >mkdir -p /scratch/nm05/rxy900/openfwi openfwi:23.11 >cd /scratch/nm05/rxy900/openfwi openfwi:23.11 >bldenv.sh Cloning into 'OpenFWI'... remote: Enumerating objects: 90, done. remote: Counting objects: 100% (90/90), done. remote: Compressing objects: 100% (52/52), done. remote: Total 90 (delta 52), reused 70 (delta 34), pack-reused 0 Receiving objects: 100% (90/90), 40.32 KiB | 1.92 MiB/s, done. Resolving deltas: 100% (52/52), done. openfwi:23.11 >
Then navigate to the left panel to open the OpenFWI tutorial notebook.
Now you are ready to run the tutorial notebook.