ParaView

Personal computer

Available for Windows, OS X, and Linux

Windows and OS X

Download latest version from ParaView website

  • MPI means in can use multiple CPUs (non-MPI is available on Windows so you can use it and not have to install the separate MS-MPI package unless you want to)

  • AMD64 and x86_64 are synonyms that both means 64bit version for AMD or Intel chips (AMD is just a reference to the fact that AMD developed the x86_64 instruction set)

  • arm64 means 64bit version for ARM chips (currently just the Mac M1)

  • older version may work if newer fails on older OS or graphics card

Linux

Install with your package manager (can also download as above)

  • Debian and Ubuntu: sudo apt-get install paraview

  • Fedora: sudo yum install paraview

Clusters

Available in the Compute Canada StdEnv (StdEnv/2020 also requires loading the gcc module)

$ module load paraview

Then you can just run it

$ paraview

VirtualGL

VirtualGL (GPU accelerated OpenGL) is enabled by default on nibi if an appropriate GPU was allocated. Recent versions of ParaView will give the error ISPCRT Error (1): Fail to load libispcrt_device_cpu.so library. Working around this requires disabling the VirtualGL libdlfaker.so (dynamic linker faker) LD_PRELOAD when starting it

$ LD_PRELOAD=${LD_PRELOAD/libdlfaker.so/} paraview

JupyterLab desktop does not enable VirtualGL by default. If you have allocated an appropriate GPU you can manually enable it by running it as

$ vglrun -d egl paraview

If you get the aforementioned error ISPCRT Error (1): Fail to load libispcrt_device_cpu.so library, then you need to tell vglrun to disable the VirtualGL libdlfaker.so

$ vglrun -nodl -d egl paraview