I'd like to use a Tesla C2075's VGA (D-SUB) DVI output. I installed the latest Quadro driver (as suggested by the NVIDIA driver finder) on the 64-bit Windows 7, but it doesn't seem to be working. E.g. I can't use Aero, can't set a larger resolution than 1600x1200, etc.

Is it possible, that the card doesn't function as a graphics card? Or should I install some other drivers?

Update: the D-SUB output is actually a DVI-I output with the appropriate (passive) adapter.
The driver is obviously working, because OpenCL applications recognize the device, and they can use it.


Windows doesn't recognize the GPU, because it uses the Tesla Compute Cluster (TCC) driver model by default for maximal performance. However, Windows uses the Windows Display Driver Model (WDDM) to enable common graphic features.

The card can be switched to the WDDM driver model by the NVIDIA-SMI (System Management Interface) utility:

C:\Program Files\NVIDIA Corporation\NVSMI\nvidia-smi.exe -dm 1

You need to reboot after that.


Everything I can find about this card suggests that it is supposed to be capable of 3d hardware accelerated graphics (Aero, Direct3D, OpenGL, etc). However, are you sure you should be using the Quadro drivers? I am not confident that the Quadro drivers are the correct drivers for this GPU.

  • I used the NVIDIA website, I picked the Tesla C2075 as my product, and I got the Quadro driver. It installed without any error.
    – KovBal
    Dec 20 '12 at 15:58
  • Oh, okay. In that case, it is probably the correct driver. Sorry about that. Well, I am not sure what is going wrong, but many sources say the Tesla is a "companion GPU" and is not designed for desktop 3d rendering through the traditional APIs; so maybe it is actually not supported. They might do that for performance reasons: they're giving you a plain old VGA card as a "console" but they expect you to do your actual computation using CUDA or OpenCL? We need more information on this. :-( Dec 20 '12 at 16:06
  • Upvoted your question because I don't think I've properly answered this question; I can't find any concrete information about this and it's driving me nuts. We need a firm answer. Dec 20 '12 at 16:06
  • OK, thank you for your efforts. Although, I think it would strange to include a D-SUB output, if it had almost no support for graphics operation with the "standard" API.
    – KovBal
    Dec 20 '12 at 16:10
  • Strange? Perhaps, but it would give them more space on the GPU chip for GPGPU operations if they could eliminate all the circuitry that is used for fixed function / graphics rendering operations; for example, hardware ROPs for texture processing. If the whole GPU is just bunches of CUDA cores, they don't have to waste die space with things that are used for desktop graphics, when most people who use Teslas will have a GeForce or Quadro or NVS branded chip to use as the primary graphics output (which does support proper 3d rendering). I can see a justification for doing this. Dec 20 '12 at 16:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.