Connecting to the M3 HPC cluster

We will use M3 “Strudel” Remote Desktops (docs) today, so everyone gets their own GPU on a compute node. We will run everything on that compute node, rather than submitting additional jobs to the SLURM queue.

• Login at https://m3-desktop.erc.monash.edu/
  • Choose “M3 with AAF login”
    • If your M3 account is associated with an institutional email address, use the orange AAF login button
    • If your M3 account is associated with a Google/gmail address, use the Google login button
• Select VS Code Server in the sidebar.
• Select T4, 4 hours.
• Edit the “Advanced Options” and change the custom SBATCh command to: sbatch --qos=desktopq --partition=desktop --gres=gpu:T4:1 --mem=100G --ntasks=6 --constraint=r9 --reservation=oct27workshop --account=pd27.
  • (In particular, make sure you have --reservation=oct27workshop, --account=pd27 and --gres=gpu:T4:1)
• Press Launch.
• Wait a few seconds, then press “Connect” under the “Pending/Running VS Code Servers” section.
  • Watch out for popup window blocking from the browser - allow the popup if it was blocked

Once you have the VS Code session running, click on the ‘two bits of paper’ icon in the sidebar, click Open Folder, and enter /scratch2/pd27/.

You can open a new Terminal with Ctrl+Shift+` (backtick), or find “Terminal -> New Terminal” in the hamburger menu.

Now let’s create your user directory where you’ll work today. In the Terminal, run:

mkdir -p /scratch2/pd27/users/${USER}
cd /scratch2/pd27/users/${USER}

Install these VS Code extensions (the ‘four squares’ icon in the sidebar, or press Ctrl+Shift+X):

• Protein Viewer ( ArianJamasb.protein-viewer ) - Molstar inside vscode
• “CSV” ( ReprEng.csv )
• Rainbow CSV ( mechatroner.rainbow-csv )
• Live Preview ( ms-vscode.live-server )

Suggested vscode extensions

---