Paraview is a multicore visualisation toolkit. The computational work can be done on Artemis and visualisation can be done on your local Paraview client. Your local client version must match the server version you are running remotely!

Visit https://www.paraview.org/ for detailed information and https://www.paraview.org/Wiki/Setting_up_a_ParaView_Server for more information about running a remote server.

Example Paraview PBS script

This assumes you have installed your own version of Paraview in the example /project/PANDORA folder.

The example multi core Paraview PBS script is below:

#!/bin/bash
#PBS -P PANDORA 
#PBS -N paraTest 
#PBS -l select=4:ncpus=4:mpiprocs=4:mem=8gb
#PBS -l walltime=00:30:00
#PBS -q default-Q
#PBS -m b
#PBS -M abcd1234@sydney.edu.au

# Load the libraries and software needed
echo $HOSTNAME
module load openmpi-gcc
mpirun -np 16 /project/PANDORA/PARAVIEW/ParaView-5.4.1-Qt5-OpenGL2-MPI-Linux-64bit/bin/pvserver > outputMPI.txt

 This runs a 16 core, 32gb paraview server spread over 4 nodes (4cpus and 8gb per node).

The output file in this example is outputMPI.txt. 

The output file will be produced when the job executes and will say what Artemis compute node your Paraview server is running on. You will need to check it to complete the next steps.

On your local computer

You must have the same version of Paraview installed on your local machine and on Artemis.

Create an "ssh tunnel" to the compute node on Artemis, in a local terminal run:

ssh -v -L11111:hpc000:11111 -N abcd1234@hpc.sydney.edu.au

Replace the project name (PANDORA), the example UniKey (abcd1234), and the node number (hpc000), with your own project name, UniKey, and whatever node number is assigned to your job.

Finally launch Paraview on your local machine and use "File > Connect..." 

Click "Add Server" and set up as follows:

 Now use Paraview as normal. Data can be accessed locally or on Artemis (if you copy it to Artemis first).