Getting started on the HPC

Basics of working on the HPC

This page is intended as a first stop for new users to the CQUniversity HPC (High Performance Computer) and HPC’s in general. Here we will explain the basic workflow and get you started on using the HPC system, though the use of an interactive session.  To begin using the HPC resources correctly, you must understand how to start an interactive session or submit non-interactive job to ensure you are utilising the HPC resources available to CQUniversity Students and Researchers.

The HPC is a shared resource that is managed by a scheduler (Slurm), in which you will need to submit jobs to computational nodes for processing to ensure that all users are able to access the HPC as needed.  It should be noted that connecting to an interactive session actually submits a job to the HPC scheduler on your behalf.

There are two kinds of ways to work on the HPC, the most universally used method is through non-interactive job submission, where a batch script is written to submit a job, request resources and execute commands. The benefit of this allows you to set up several jobs to run concurrently to help improve workflow and results.

The other is through an interactive session where you reserve resources for a period of time, in which can make use of the HPC remotely to run and view results in real time. This allows for more instant feedback for smaller jobs, help troubleshoot and testing purposes.  Most HPC jobs should be considered running as non-interactive jobs, but there are some cases where interactive sessions are more suitable.  If you need help in converting your HPC workloads to non-interactive jobs, it may be worth reaching out to the eResearch support team to see what benefits not-interactive jobs can have for your workflow and help getting scripts made to run them.

Moving files onto the HPC can be achieved by mapping the storage drive and moving the files needed to the HPC. Files and results on the HPC aren’t backed up so please ensure that any important files are moved to a secure location such as your Reseach Data Share.

When submitting a non-interactive job to the HPC, there are 3 partitions to choose from, depending on the resources needed will determine which partition you should run your job on.

For running interactive sessions, the H100’s are not available.

Node	Partition name	Use case
Login	N/A	Used to log into the HPC and submit job to the scheduler or to check on the process of currently running jobs. This node is where most of the administration and setting up of jobs should happen. viewing files and results, moving and organising things as needed and minor testing. This node resources are shared between all users so running large jobs on it will be slow and also slow down the works of other users.
CPU only	workq	CPU nodes are used in the majority of the uses and have the most nodes available for use, they’re able to process large amounts of data and non-specialised but significant computational tasks.
L40s GPU	gpuq	the L40s nodes are limited but allow for graphical rendering of specialised programs such as pytorch, ansys and tensorflow. These nodes can also be run interactively for experimenting with Large Language Models (LLM) also known as AI
H100 GPU	gpucomputeq	the H100 nodes are best suited for very intensive computational tasks and are ideal for LLM use or any similarly large task. These nodes don’t allow for interactive sessions to be run do to being limited and the demand for LLM processing. Jobs need to be submitted to these nodes non-interactively to these nodes.

First steps

 The first step to getting started on the HPC is requesting to be added as a HPC user by contacting eresearch@cqu.edu.au.

After getting a HPC account you’ll need to access the CQUniversity network either by being on campus or via VPN, if you need assistance with getting the VPN set up contact TaSAC.  

After connecting to the CQUniversity VPN you can map the HPC storage drive to your PC to allow you to transfer the files to the HPC and the results back to your PC.

Now that your files are on the HPC you can access Open on Demand (OOD) at https://adalovelace-gui.cqu.edu.au to access the HPC resources.

From OOD you can establish a graphical connection to the HPC in to start an interactive session on either the login node, CPU nodes, or the L40s nodes.

 

Getting started once connected to the HPC

Once you have your files on the HPC the next step is to connect to the HPC with either SSH or by starting a graphical connection. For new users we recommend starting a graphical session via open on demand and load up the program needed. The instruction below will give you the basics of how to get started once you’re on the desktop and the recommended resources needed to run the program.

Once logged into the HPC you’ll need to open up a Terminal window by clicking the button located on the top bar.

This window is how you’ll load the launch the programs you want to use on the HPC using the commands below for the respective programs. There are recommended resources for running each of the programs but depending on your work load you may need more.
With the commands below between each line you’ll need to hit enter to run the command before starting on the next line and after the last command wait for the program to launch. The commands are case sensitive so if commands aren’t working double check the spelling and capital placement.

Frequently Used Programs

The following commands are written for the new HPC Ada Lovelace, Some of the software versions may not be present on the current HPC.

MATLAB

Commands to put into Terminal
module load MATLAB/2024b
matlab


R 

Commands to put into Terminal
module load R
module load rstudio
rstudio

PYTHON

Commands to put into Terminal
module load Anaconda3/2024.06-1
python


ABAQUS

Commands to put into Terminal
module load abaqus/2025 
abaqus cae -mesa

ANSYS

Commands to put into Terminal
module load ANSYS/2025R1
runwb2 -oglmesa


For more information about using the scheduler (Slurm) to submit non-interactive jobs there are more resources available here.