Difference between EGL and OSMESA

I need to start a session of pvserver on a Linux workstation where only remote login is possibile. No external clients can connect to the graphical X server session (it is always xhost -).
Because of that, it is impossible to connect to display 0.0, 1.0 etc. and to have access to remote rendering. I think the simplest solution is to use the paraview headless builds, but certain aspects are not clear to me. I must premise that I want to utilise the GPU for remote rendering :

  1. What are the differences between EGL and OSMESA pre-build versions?
  2. How do I check if the GPU is being utilized, for example when running osmesa prebuild pvserver?
  3. How to force EGL or OSMESa pvserver versions to utilise the GPU?

EGL version uses the GPU - this was only tested with NVidia graphic cards.
OSMesa does software rendering.

So, if you have an NVidia graphics card (and proprietary driver installed) you should use the EGL version, otherwise use the OSMesa version.


1 Like

Thank you very much!