Failed Cuda Driver and Runtime version may be mismatched Cuda installation fails on Ubuntu 10.4 x86_

Accelerated Computing CUDA CUDA Programming and Performance
1

I installed the cuda sdk and the cuda drivers. Version 3.2, both from the same page.

I compiled the sample code, which does compile successfully.

But then I get this error when running deviceQuery

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

cudaGetDeviceCount FAILED CUDA Driver and Runtime version may be mismatched.

FAILED

Press to Quit…
----------------------------------------end output------------------------------------------------------

The card is a Tesla

I’ve blacklisted some modules that are mentioned in other forums
blacklist vga16fb
blacklist nouveau
blacklist rivafb
blacklist nvidiafb
blacklist rivatv

I’ve purge all nvidia packages and started from scratch . No improvment.

The card does have an issue recognizing monitors, which is also mention on other forums. But this is not a problem in need of solving since the machine will be a monitor-less server; nevertheless, this bit of info must be a good clue as to what is going on.

Any help is appreciated.

I’ll install version 3.0 while I wait for a response.

Most of the troubleshooting steps where taken from:
[url=“http://www.ubuntugeek.com/howto-install-nvidia-drivers-manually-on-ubuntu-10-04-lucid-lynx.html”]http://www.ubuntugeek.com/howto-install-nv...lucid-lynx.html[/url]

2

I installed the cuda sdk and the cuda drivers. Version 3.2, both from the same page.

I compiled the sample code, which does compile successfully.

But then I get this error when running deviceQuery

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

cudaGetDeviceCount FAILED CUDA Driver and Runtime version may be mismatched.

FAILED

Press to Quit…
----------------------------------------end output------------------------------------------------------

The card is a Tesla

I’ve blacklisted some modules that are mentioned in other forums
blacklist vga16fb
blacklist nouveau
blacklist rivafb
blacklist nvidiafb
blacklist rivatv

I’ve purge all nvidia packages and started from scratch . No improvment.

The card does have an issue recognizing monitors, which is also mention on other forums. But this is not a problem in need of solving since the machine will be a monitor-less server; nevertheless, this bit of info must be a good clue as to what is going on.

Any help is appreciated.

I’ll install version 3.0 while I wait for a response.

Most of the troubleshooting steps where taken from:
[url=“http://www.ubuntugeek.com/howto-install-nvidia-drivers-manually-on-ubuntu-10-04-lucid-lynx.html”]http://www.ubuntugeek.com/howto-install-nv...lucid-lynx.html[/url]

3

An update. I tried 3.1 and the latest 3.2 RC .14 release, but still get the same result.

4

An update. I tried 3.1 and the latest 3.2 RC .14 release, but still get the same result.

5

I fixed it by manually loading the modules.

I used a script from

http://forums.nvidia.com/index.php?showtop…mp;#entry272085

I only had to change the path to lspci to

/usr/bin/lspci

6

I fixed it by manually loading the modules.

I used a script from

http://forums.nvidia.com/index.php?showtop…mp;#entry272085

I only had to change the path to lspci to

/usr/bin/lspci

7

Thank you for posting a followup to your question. I came here to post the exact same question, and the script you linked to will serve as an acceptable solution.

I created a new topic inquiring if anyone knew why this was necessary (my system runs X, so the /dev entries should be automatically created). If you know the answer, or are interested in following the discussion, the thread is here:

[url=“The Official NVIDIA Forums | NVIDIA”]http://forums.nvidia.com/index.php?showtopic=184282[/url]

8

Thank you for posting a followup to your question. I came here to post the exact same question, and the script you linked to will serve as an acceptable solution.

I created a new topic inquiring if anyone knew why this was necessary (my system runs X, so the /dev entries should be automatically created). If you know the answer, or are interested in following the discussion, the thread is here:

[url=“The Official NVIDIA Forums | NVIDIA”]http://forums.nvidia.com/index.php?showtopic=184282[/url]

9

Thanks for your mention of Centos 5.x success story. After spending roughly 30 hours trying to get what was a properly working F13 w 256.xx driver based system running with any and all 260.xx drivers (every combo of direct from nividia and rpmfusion sourced driver & settings), I will have to seriously consider switching the OS.

10

Thanks for your mention of Centos 5.x success story. After spending roughly 30 hours trying to get what was a properly working F13 w 256.xx driver based system running with any and all 260.xx drivers (every combo of direct from nividia and rpmfusion sourced driver & settings), I will have to seriously consider switching the OS.

11

Any tips, tricks to installing devdriver 260.19.xx on centos 5.5? Have performed a clean install (direct from nvidia) and looked thru the xorg.conf file but startx halts @ NV-GLX loading. TIA, V.

12

Any tips, tricks to installing devdriver 260.19.xx on centos 5.5? Have performed a clean install (direct from nvidia) and looked thru the xorg.conf file but startx halts @ NV-GLX loading. TIA, V.

13

Try to install the newest driver and a lower version of toolkit. I had the same problem and found the driver provided is not always sufficient for the runtime for some devices

14

Try to install the newest driver and a lower version of toolkit. I had the same problem and found the driver provided is not always sufficient for the runtime for some devices