Loughborough University
Leicestershire, UK
LE11 3TU
+44 (0)1509 222222
Loughborough University

IT Services : High Performance Computing

Visualisation: Using

Getting Work to the Node

Using Sbatch

You should ideally use sbatch to send jobs to the required node. For live visualisation this works best for systems that can use a client-server mode, such as paraview .

Node Selection

To use the dedicated visualiation nodes ensure that any SLURM commands run reference --partition=visual for full hardware acceleration. For other visualisation tasks that do not require hardware acceleration any compute partition (compute-12 or compute-12) can be used.

For more information on partitions see Partitions .

Hardware Acceleration and Displays

To get the best from the visualisation nodes you need to use the hardware acceleration of the nodes. For some packages this means using the X Windows server on the node by setting the option -display :0 on the commands run to make it use the hardware accelerated X Windows display that is running on the node. For applications such as paraview this then allows the client-server operation to silently use the hardware acceleration then relay the information back to the user in real-time. This is as it has been set up to do so without the need for special system-level permissions to use the hardware, but allowing the current user and only the current user with a job running on a node access to the X Windows system for the duration of the job.

If for some reason you need to render on a node without hardware acceleration then you can start up a dummy X Windows server to use by prepending commands with xvfb-run.