I'm planning to start some CUDA development, so I'm trying to use the NVIDIA card that I have installed and I haven't yet be able to use (an NVIDIA Corporation GF108M [GeForce GT 630M]) because Ubuntu (12.05 64 bits)seems to have many issues with it.
What I did is download the CUDA 5.5 installers and execute the following in that order:
NVIDIA-Linux-x86_64-319.37.run cuda-linux64-rel-5.5.22-16488124.run cuda-samples-linux-5.5.22-16488124.run
To make the NVIDIA driver installation to work, I have to uninstall Bumblebee first. The installer wanted to override
xorg.conf but I choose not to because some previous disastrous installations (I only get a black screen).
I used this Blog Post instructions to validate my installation. I first executed this:
cptanalatriste@cptanalatriste-Inspiron-5420:/usr/local/cuda-5.5/lib64$ cat /proc/driver/nvidia/version cat: /proc/driver/nvidia/version: No such file or directory
And I get that error message. After compiling and executing the CUDA samples. I'm getting this:
cptanalatriste@cptanalatriste-Inspiron-5420:/usr/local/cuda-5.5/samples/1_Utilities/deviceQuery$ sudo ./deviceQuery ./deviceQuery Starting... CUDA Device Query (Runtime API) version (CUDART static linking) WARNING: Not loading blacklisted module nvidia_304 FATAL: Module nvidia not found. cudaGetDeviceCount returned 38 -> no CUDA-capable device is detected Result = FAIL
Any ideas? I suspect the Driver installation didn't go that well.