Remote accelerated graphics with VirtualGL and TurboVNC

In this short blog I would like to do something useful. My supervisor Gábor encouraged me to write a short Quick Start Guide to remote connections. The task we’ll discuss is pretty common – you want to connect to another computer (assuming *NIX server) and see a nice graphical output e.g. form 3D visualization app. A typical scenario in HPC is running visualizations on specialized nodes equipped with high performance graphics hardware and direct connection to a supercomputer’s storage.

There are many options. I will just briefly mention some commercial solutions, like Microsoft Remote Desktop Protocol which works great for connecting to Windows machines (form almost any platform), but lacks accelerated graphics on forks for other platforms servers. Well known TeamViewer doesn’t support 3D graphics at all, Chrome Remote Desktop still needs development and my favorite NoMachine uses protocol that I will describe below.

We will concentrate on session forwarding using VirtualGL and TurboVNC.

How it works:

VirtualGL transportation scheme from WikiPedia https://en.wikipedia.org/wiki/VirtualGL

Normal X-forwarding sends all X rendering instructions and the client does all the rendering on a CPU. VirtualGL overrides this protocol and renders 3D graphics at the host’s GPU and then send only the rendered 2D images to the client (the scheme is much more complicated).

Or you can utilize this and render everything at host and transfer the image of the desktop via VNC protocol. The Turbo stands for using turbo fast SIMD libraries for jpeg encoding.

For more information and inspiration see links 1, 2, 3.

How to run it example:

Installing VGL on a remote server

on the remote visualization station (needs root):

1. get and install packages VirtualGL and TurboVNC from SourceForge
2. configure the VGL server and modify the X-server trough command line

   sudo vglserver_config

3. restart display manager (xdm or lightdm) or just reboot server

   sudo /etc/init.d/xdm restart

Installing on your local client

On your local client install the very same packages (needs root):

1. Get and install packages VirtualGL and TurboVNC from SourceForge .

Scenario A – Making VirtualGL connection

This is the easiest choice if you want to run only certain application remotely.

1. Establish connection (type in terminal)

   vglconnect -s user@graphics-server

   e.g. vglconnect -s jano@vserver.debrecen.hpc.niif.hu

   This will make a connection with ssh-encrypted X-forwarding and VGL Image transport. For more detail about options see VirtualGL Reference.

2. Run your OpenGL application with vglrun command

   vglrun application

   e.g. vglrun glxinfo | more to see OpenGL status or ParaView vglrun paraview

   Scenario B – Making a TurboVNC connection

   This is more a complex (and a bit faster) solution for forwarding whole X sessions. More people can collaborate on the same session and it will last even after you close your connection, so you can continue with your work.

   1. Start a VNC server session on the remote visualization station using the command

      vncserver

      You will be prompted to set (and confirm) a password. If it was successful, you will obtain something like this

      Desktop 'TurboVNC: login:1 (jano)' started on display login:1
      Starting applications specified in /home/jano/.vnc/xstartup.turbovnc
      Log file is /home/jano/.vnc/login:1.log

      the important thing here is the last number in the first line (after the colon i.e. login:1) – the display number (1 in this case), by adding 5900 to it we’ll obtain the port number of the VNC (in this case 1 + 5900 = 5901).

      You can always check this by using the command vncserver -list.

      You can disconnect now.

   2. On your local machine, create an ssh tunnel for encrypting the VNC stream by typingbashshh -N -q -L 1047:localhost:<display_number> user@serverwhere 1047 is the arbitrary free port number and <display_number> is obtained from previous the paragraph, e.g. ssh -N -q -L 1047:localhost:5902 jano@vserver.debrecen.hpc.niif.hu.
   3. Now you can start the TurboVNC client on your local machine e.g. running command

      vncviewer localhost:1047

      If you cannot find the vncviewer command, this might help

      /opt/TurboVNC/bin/vncviewer localhost:1047

      Now you just need to type your chosen password for VNC and start the X-session,

   4. Start the 3D accelerated application with vglrun command

      vglrun application

      e.g. vglrun glxinfo | more to see OpenGL status or ParaView vglrun paraview. (same as in the Scenario A)