TAMU MATLAB Help Session

The Texas A&M University (TAMU) MATLAB help session provides a comprehensive guide to utilizing MATLAB with Slurm, a batch scheduler used for running jobs on high-performance computing systems. In this session, we will explore how to run MATLAB scripts using Slurm and the `matlabsubmit` command.

Example 2: Utilizing Matlab Workers (Single Node)

To begin, let's consider an example of utilizing MATLAB workers on a single node. This involves creating a batch script, a submission script, and a MATLAB driver code that sets the number of threads and calls a user-defined function. The output from MATLAB will be redirected to a log file, along with any errors that may occur during execution.

Directory Structure

When running `matlabsubmit`, it will create a new directory named `MatlabSubmitLOG` (where N is the matlabsubmit ID). Within this directory, you can find all relevant files generated by `matlabsubmit`, including the batch script, MATLAB driver code, and redirected output and error logs.

Checking Run Status or Killing a Job

To check the run status of your job or kill it if necessary, you can use the respective batch scheduler commands. For example, to check the run status, you can use `sbatch` followed by the `-o` option to specify the output file. To kill a job, you can use `scancel` followed by the job ID.

Matlab Output File

The MATLAB output file is stored in the `MatlabSubmitLOG1` directory and is named `matlab.log`. This file contains all output generated by MATLAB during execution, including any errors that may have occurred.

Batch Scheduler Commands

When running `matlabsubmit`, it will execute the batch scheduler commands to submit your job. These commands include options such as `-o` for specifying the output file, `-job-name` for naming the job, and `-ntasks` for specifying the number of tasks to run concurrently.

Matlab Resources Requested

When running `matlabsubmit`, it will request specific resources from MATLAB, including the number of threads and workers. These resources are specified using options such as `-cpus-per-task` for specifying the number of CPUs per task and `-mem` for specifying the amount of memory to allocate.

In this article, we have explored how to utilize `matlabsubmit` to run MATLAB scripts on high-performance computing systems using Slurm. We have discussed the directory structure generated by `matlabsubmit`, how to check the run status or kill a job, and the output files generated during execution. By following these guidelines, you can effectively use `matlabsubmit` to run your MATLAB scripts on TAMU's high-performance computing systems.

Tamu matlab help session