The best reference is the VirtualGL User’s Guide and the TurboVNC User’s Guide. The basic strategy is:
- Install TurboVNC first, using the TurboVNC User’s Guide instructions, and verify that you can launch and connect to a TurboVNC session. (It won’t have GPU acceleration yet.)
- Install VirtualGL, using the VirtualGL User’s Guide instructions, and verify that you can get GPU acceleration in a TurboVNC session when you execute ‘vglrun /opt/VirtualGL/bin/glxspheres64’. (The OpenGL renderer that GLXspheres reports should indicate that it using the nVidia GPU.) Note that this assumes that the nVidia drivers have already been installed.
- (Optionally) Enable VirtualGL by default for all TurboVNC sessions on the machine, by setting ‘$useVGL = 1;’ in /etc/turbovncserver.conf. If you don’t do that, users will still be able to use VirtualGL, but they will have to explicitly type ‘vglrun {application}’ to launch {application} with OpenGL GPU acceleration.
The User’s Guide instructions involve installing VGL and TurboVNC from discrete packages, but we also have YUM repositories. See VirtualGL.org and TurboVNC.org for YUM repository info.
Note that, when using nVidia’s drivers, Vulkan applications will be GPU-accelerated automatically when running in a TurboVNC session. VirtualGL is only needed for OpenGL applications.
Let me know if you run into specific problems.