Documentation:Paraview

ParaView is an open-source, multi-platform data analysis and visualization application. ParaView users can quickly build visualizations to analyze their data using qualitative and quantitative techniques. The data exploration can be done interactively in 3D or programmatically using ParaView's batch processing capabilities.

ParaView was developed to analyze extremely large datasets using distributed memory computing resources. It can be run on supercomputers to analyze datasets of terascale as well as on laptops for smaller data.

This documentation is valid for

Paraview can work in a client-server mode, that's how we will use it on the clusters. This will only work from within the EPFL network. If you're at home, you should use the VPN.

The Client
Download the Paraview binary installer for your machine. The binary must match the version installed on the cluster. You can use Paraview as a stand-alone program or use it as a client to the server that is installed on the machines. If you want to use it as a client, you have to transfer (possibly huge) files from the clusters. Coupling it with the server lets you visualize and process your data directly form the compute nodes. In order to do so:


 * launch the client
 * click File -> connect
 * Add server: enter the name of the cluster (for example bellatrix.epfl.ch) and choose Server Type Client / Server (reverse connection)
 * Choose Startup Type: Manual

The Server
The pvserver has been installed on the clusters for which this documentation is valid. Once you instruct your client to connect to the cluster via a reverse connection, you can submit the following script to the queuing system:

# # # PVCLIENT=128.178.183.227 module load paraview/3.98.0-gcc_mvapich2-4.6.3_1.8.1 cd ${PBS_O_WORKDIR} cat ${PBS_NODEFILE} NUM_PROC=`cat ${PBS_NODEFILE} | wc -l` mpirun -np ${NUM_PROC} pvserver --use-offscreen-rendering -rc -ch=${PVCLIENT} #
 * 1) !/bin/bash
 * 1) PBS script for launching ParaView in server mode
 * 2) with offscreen rendering
 * 3) do this if you're using ParaView on a cluster
 * 1) PBS -l walltime=00:10:01
 * 2) PBS -l select=1:ncpus=16:mpiprocs=16
 * 3) PBS -N ParaView
 * 1) get the number of processors and nodes
 * 1) mpi program launch

You have to change the PVCLIENT and set it to your IP address. Change the walltime and number of nodes to something reasonable for your case and then launch with qsub. Once your job is running, the client will wake up and you're good to select your output files that reside on the cluster.