You don't need to manually load any modules to use the Linux desktop, instead this software is automatically loaded for you when you start an Open OnDemand desktop session.
See Open OnDemand and Interactive Applications for full details.
We discourage the use of SSH + X11 which was the only mechanism to display output on Rocket and earlier HPC facility, as this is limited to connecting to the HPC head nodes. Please consider using a Open OnDemand virtual desktop session if you want to run graphical applications.
Log in to the Comet OOD dashboard:
From the Interactive Apps menu, select VNC Desktop Session and complete the form which is shown:
Ensure that you choose the correct Slurm partition for your VNC Desktop instance. Consult our Comet HPC Resources & Partitions page to understand the difference between the available Slurm partitions.
Once you submit the form the system will schedule your request and shortly create a new dedicated VNC Desktop application for you, this will open in your browser, and you will be able to run any and all software installed on Comet, including from your $HOME or /nobackup directories. Full access to all the regular Comet filesystems is possible, and you can request up to 256 CPU cores for your VNC Desktop session.
$HOME
/nobackup
All normal desktop applications should work without issue. If you want to run applications which make use of 3D rendering / OpenGL or similar, please read the options below for working with OpenGL on both compute and GPU nodes.
When you are finished running the desktop session, logout by click the Applications menu in the upper left, and then clicking Logout the session will then automatically close within a few seconds and your browser tab will show the session closing.
If you start a Linux X11 Desktop session on a CPU compute node, then you will be able to make use of 3D rendering using the MESA OpenGL library.
In a terminal window in the virtual desktop session, load the following modules:
$ module load Mesa $ module load mesa-utils
Now, any applications started from the same terminal window will use the MESA OpenGL driver. Test it by running a simple demo:
$ glxgears
If you start a Linux X11 Desktop session on a GPU compute node, then you will be able to make use of 3D rendering using the hardware accelerated Nvidia OpenGL library.
Please use VirtualGL to start any applications which require 3D graphics hardware for visualisation output - this allows the application to create accelerated graphics using the Nvidia hardware, and then render the image back on your desktop. Applications requiring GPU for compute will work normally as long as you run the session on a GPU node - you do not need to load VirtualGL if your application is only making use of the GPU for compute (e.g. CUDA).
$ module load virtualgl $ virtualgl glxspheres64
You must allocate at least one GPU card resource to your Open OnDemand desktop session in order to use the Nvidia OpenGL library, or access CUDA functionality.
Requests for new/alternate versions of this package should be submitted via the NUIT ITService website using the HPC Software - Installation category.
Software requests will be collated, prioritized and submitted to our HPC support vendor.
Back to Software List
Table of Contents
Main Content Sections
Documentation Tools