Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Follow these steps to submit a MATLAB Batch Jobbatch job:

  1. Create the SBATCH script. You can create this script on your workstation and then upload it to the Aloe supercomputer. You can also create this script through the shell. To do this:

    1. Connect to the web portal and log on with your credentials.

    2. Select “System“ in the navigation bar then select “Aloe Shell Access“. You will be prompted for your credentials and a DUO push will be sent to you.

    3. In the command prompt, use the following commands to create a SBATCH script called “MATLAB_test_script.sh“.

      Code Block
      nano MATLAB_test_script.sh
    4. In the file editor, specify the resources, MATLAB version, and the MATLAB file needed for your computation. Here is an example of a MATLAB SBATCH script:

      Code Block
      #!/bin/bash
      #SBATCH -c 16     # number of TASKS
      #SBATCH -N 1     # keep all tasks on the same node
      #SBATCH --mem=120G     # request 120 GB of memory
      #SBATCH -t 0-4     # request 4 hours of walltime
      
      module load matlab/latest # load in the lastest version of MATLAB
      matlab –batch hello.m # run the MATLAB file


      This job will allocate 16 cores and 120 GB of memory on one node for four hours. It will also execute the “hello.m“ file with the latest version of MATLAB on the Aloe supercomputer.

    5. Save the SBATCH script and exit the file editor.

  2. In the command prompt, use the following commands to submit the batch job:

    Code Block
    sbatch MATLAB_test_script.sh
  3. Done. Your job has been submitted. Here are some helpful commands to manage your job:

    Code Block
    myjobs # This will show the status for all of your jobs.
    Code Block
    scancel JOBID # This will cancel one job by using its JOBID.
    Code Block
    scancel –u $USER # This will cancel all of your jobs. 
    Code Block
    seff JOBID # This will give the efficiency report for one job by using its JOBID.

Note

How will I know how many and which resources to use?

Learning to use the supercomputer, like learning how to use other new technologies in your life, will take trial and error. A good starting point is to start with one core for two hours in the the interactive MATLAB graphical application as you get familiar with the supercomputer. You can also request resources to replicate your current workstation to create a baseline. Through this trial and error, you will identify what resources can better improve your computation’s performance. See the “Requesting Resources on Aloe“ guide for additional tips.

...