Failed to initialize NVML, No CUDA Device

Hi there,

I am trying to set up a server with three Tesla M2070Q graphic cards and one ATI ES 1000 (primary display device).
When I try to set them in TCC mode for using CUDA I receive the following error:

Failed to initialize NVML: Unknown Error

I installed the latest CUDA SDK (4.128 x64), Toolkit (4.128 x64) and display driver (developer 286.119).
Bandwidth test says:

No CUDA-capable device is detected.

To connect to the server I use teamviewer. I dont know what to do anymore. After searching a solution for the last 3 hours I am finally asking my question here.
I also tried to use the latest Tesla Drivers (295.73), but nothing helped.

Thank you for helping me :)

Ok, I finally found a workaround for the problem on the IBM website.

Here is how it works:

Here is the link: IBM