I have succeded to install drivers… at least apparently, by stopping the service “lightdm”, and launching the install cuda_5.5.22_linux_64.run. Now i can see nvidia-settings, and there everything is working fine, my screen is set to the maximum resolution: 1440x900.
But if i go to the Additional Drivers none of the 304 or 319 is set as “active”.
Moreover in synaptic none of the packages nvidia-cuda-toolkit or nvidia-cuda-gdb is resulting installed.
Moreover i’m again able to compile deviceQuery using nsight.
The questions are: - should I check as active the additional drivers?
- should I install the packages through synaptic?
If you used the CUDA installer 5.5 installer and installed the driver included with it, disregard any info on the Ubuntu synaptics packages/additional drivers… they won’t be installed / in use because the standalone NVIDIA CUDA driver installer you used does not depend on them. As long as you’re able to compile/run deviceQuery SDK sample, you should be all set.
i had exactly this problem on my lab machine, this issue is about my home machine.
Now the problem is that tomorrow at work i will receive my new card (GTX 660) for the lab machine.
Can you tell me exactly what to do in this case to uninstall all the previous drivers correctly, and then do not have the problem with the kernel update??
You’d have to use the originall installer .run file you use and see what the options are… I believe one of them in --uninstall or something similar. After that, you’d have to add the xorg-edgers repository to your ubuntu sources and install whatever driver version you want… perhaps nvidia-319 or a later one.
The installer installs an uninstall helper in /usr/bin/nvidia-uninstall, so you can just run that instead of downloading the original .run file again just to use its --uninstall option.