[VirtualGL-Users] Single application runs very slow with VirtualGL/TurboVNC while others run fine

[Reiner R. <rot...@at...>]

Hello,

currently we are evaluating latest versions of VirtualGL together with 
TurboVNC on RHEL 4U6 in order to allow engineers to access certain CAD 
software on a remote server.

As a testmachine we use a dedicated server with Intel quadcore CPU, 2GB RAM 
and a decent Nvidia grafics card.

Our first tests were with regular LAN speed 100 Mbit/s, later we used a slower 
broadband connection to a subsidiary.

Most applications did run quite well. For example Ansys Workbench [1] or Ansys 
CFX [2]. 

However one application that also uses OpenGL for 3D accelleration, Ansys ICEM 
CFD [3], was very slow and there was a huge lag even in the local area 
network, while other applications that use 3D accelleration at least in the 
same magnitude worked fine. Also tests on the real X-Window desktop showed 
that the server has no problems with either application to run fluently.

No we want to dive deeper into this issue to isolate the root cause as 
VirtualGL otherwise seems to do a perfect job.

My Questions are:

- Is there a way to further debug a VirtualGL session?
- What are the most prominent sources of error why this one application is 
still slow?
- Are there any general tipps to further increase overall performance?
- Is there any practical experience how many users may use VirtualGL 
simultanously?
- Is there a way to utilize multiple cpu cores, as rotating an object seems to 
produce 100% cpu usage on one core for a short period of time (ssh process)?
- Is there also a commercial support for VirtualGL/TurboVNC available?

Best regards,
Reiner Rottmann

[1] http://www.ansys.com/products/workbench.asp
[2] http://www.ansys.com/products/cfx.asp
[3] http://www.ansys.com/products/icemcfd.asp