M2050 OpenGL on Windows How to access Tesla M2050 from nVidia Control Panel ?

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi all,

I recently got a machine which has 1 TESLA M2050 card along with another regular not nvidia Card. Windows 2008 Server R2. I need to access this machine remotely.

How would I access TESLA M2050 properties from NVIDIA Control Panel ? I can’t do this now since I’m getting “You are not currently using a display attached to an NVIDIA GPU” error at the console …

What are the Registry settings that NVIDIA Control Panel changes with the force EDID option ?

Has anyone done this experiments with this card ? Or with M2070Q ? Using Quadro Cards is not a feasible option.

Thanks and regards
KyoTZ

2

The problem you’re seeing is that the M2050 is using the non-WDDM driver (the TCC driver), which is not supported by the control panel. However, if you’re using a non-WDDM display card or you need to use Remote Desktop, you must use the TCC driver, or the WDDM driver won’t load. You can change what mode the driver is in with nvidia-smi (installed at C:\Program Files\NVIDIA Corporation\NVSMI).

3

Thanks tmurray,

Couple of (additional) questions:

  1. Does OpenGL needs DirectX on Windows 2008 to actually work ? (I was told that yesterday, so if M2050 doesn’t support DirectX I’m pretty much going nowhere …).

  2. So as tmurray says … I’ll need to hack something to virtually attach a display to my M2050 and use that so I could enable WDDM, so NV Ctrl Panel can launch, so I can have a happy life, right ?

Cheers

KyoTZ