ParaView Installations on Tetralith & Sigma

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 ParaViews 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 exascale size as well as on laptops for smaller data.

Installed ParaView Versions

Version NSC Module Version Info
5.4.1 ParaView/5.4.1-nsc1-gcc-2018a-eb regular, interactive version with GUI
5.4.1 ParaView/5.4.1-osmesa-nsc1-gcc-2018a-eb special version for batch jobs, without GUI

How to run

Load the paraview module corresponding to the version you want to use, e.g

module load ParaView/5.4.1-nsc1-gcc-2018a-eb

This will add the paraview application to your search path.

To start Paraview in the standard, serial way: paraview

When running the ParaView GUI, NSC recommends using ThinLinc to access Tetralith.
For more information on how to use ThinLinc, please see: Running graphical applications using ThinLinc

How to run Paraview in parallel

Paraview supports parallel rendering. This is recommended if your data sets are very large. It is based on a client-server model. The client is the Paraview graphical interface, where the screen output is rendered. The heavy computation is done on the parallel server. If the size of the data exceeds the memory capacity of a regular compute node, you may want to use a fat node.

The client and server are setup as follows:

  1. Login to a compute node

  2. Load the required Paraview module, e.g.

    module load ParaView/5.4.1-nsc1-gcc-2018a-eb
  3. Start the Paraview client in the background

    paraview &
  4. Start the parallel server

    mpprun pvserver --use-offscreen-rendering

    You should get a text output showing which node and port Paraview is using, e.g. node=n290 and port=11111

    Waiting for client...
    Connection URL: cs://n290:11111
    Accepting connection(s): n290:11111

    11111 is the default port

  5. Connect the client with the server

    • Open the server window in Paraview. *Menu: File->Connect*
    • Click *Add Server*
    • This opens the window *Edit Server Configuration* _ Name: Assign a name to the new configuration _ Server Type: Client/Server (default) * Host: The node, where you are running the parallel server. If you run the client and the server on the same node, then the default setting is fine: localhost

      • Port: 11111 (default)
      • Click Configure
      • Then select Startup Type: Manual and click Save
      • Now choose the server you just edited and click Connect

Your client should now connect to the pvserver we set up and you should be able to do parallel rendering.

To exit paraview, disconnect the client from the server

Disconnect client from server. Menu: File->Disconnect

How to check if you run a parallel server

The Pipeline Browser (top left windows in the Paraview GUI) shows if you are running in serial or parallel mode:

  • Standard serial mode: builtin
  • Parallel server mode: cs://localhost:11111

Paraview settings

One should change the following setting in Paraview, when using a parallel server:

  • Menu: Edit->Setting
  • Render View: Set Remote Render Threshold = 0

The Remote Render Threshold option tells Paraview to always use parallel rendering. This option is important to obtain parallel scalability.

How to run ParaView in parallel for large datasets

A regular compute node has a maximum memory of 96GB, including memory for the system. For large datasets, this memory may not be sufficient. In this case, you can allocate e.g. two compute nodes and distribute the number of parallel processes among the two nodes to obtain more memory. In some cases, the structure of the dataset may only allow a certain number of parallel processes to read the data. For example, if your dataset can only be read by a maximum number of 12 parallel processes, you can allocate two nodes with 6 processes on each node:

interactive -N2 --tasks-per-node=6

Then you continue as explained under "How to run Paraview in parallel"

How to run Paraview in batch mode without GUI

For complex tasks, ParaView offers the possibility to run python scripts. If you want to run python scripts in batch mode, without running the graphical user interface, you have to use a special ParaView version. This version is denoted as "osmesa". OSMesa means "off-screen mesa". This version does not support an interactive GUI, but it is intended for batch mode operations. The regular ParaView version does not work in batch mode, as it expects an interactive graphical user interface.

Load the correct ParaView module for batch jobs:

module load ParaView/5.4.1-osmesa-nsc1-gcc-2018a-eb

Call your python script, using the pvbatch command:

mpprun pvbatch <your python script.py>

In this case, you do not need the option --use-offscreen-rendering, as it is a special ParaView version.


User Area

User support

Guides, documentation and FAQ.

Getting access

Applying for projects and login accounts.

System status

Everything OK!

No reported problems

Self-service

SUPR
NSC Express