deviceQuery fails CUDA Driver and Runtime version may be mismatched.

Hi there,

I’m having a major problem getting cuda working on my new computer. I have the Fermi 2050 card. I’m running scientific Linux 5.5 and have installed the NVIDIA-Linux-x86_64-260.19.36 driver. I’ve downloaded the cuda3.2 toolkit and every thing seems to be okay until I run the deviceQuery SDK and it fails with this output:

deviceQuery Starting…

CUDA Device Query (Runtime API) version (CUDART static linking)

FATAL: Module nvidia not found.
cudaGetDeviceCount FAILED CUDA Driver and Runtime version may be mismatched.


Press to Quit…

Not too sure where to go from here. Would really appreciate any help/suggestions/advice.


can you post the output of the following two commands?

$ cat /proc/driver/nvidia/version 

NVRM version: NVIDIA UNIX x86_64 Kernel Module  260.19.21  Thu Nov  4 21:16:27 PDT 2010

GCC version:  gcc version 4.3.3 (Ubuntu 4.3.3-5ubuntu4) 

$ ls /dev/nv*

/dev/nvidia0  /dev/nvidia1  /dev/nvidiactl

The error you have suggests the driver isn’t loaded at all. Are you running X11 or not?

Okay, nvidia isn’t in my driver directory and I’m running X11. What do I need to do to load them? I followed the installation instructions exactly but clearly somethings gone wrong somewhere.


If you are running X11, what display card and driver are you using?

Page 6 of the Linux getting started guide contains some instructions for loading the driver and creating the necessary device files when the NVIDIA X11 driver is not used, but whether that works might depend on the answer to the above question…

I’ve reinstalled my drivers are everything is now working fantastically. Cheers for your help avidday!