Is My Quadro 4000 compatible with Cuda Toolkit 8.0?

I installed the Cuda Toolkit 8.0 from https://developer.nvidia.com/cuda-toolkit, on Ubuntu 16.04, with a Quadro 4000 graphics card.

When running a Theano code sample, I get this:

ERROR (theano.sandbox.cuda): Failed to compile cuda_ndarray.cu: libcublas.so.8.0: cannot open shared object file: No such file or directory
WARNING (theano.sandbox.cuda): CUDA is installed, but device gpu is not available (error: cuda unavailable)
[Elemwise{exp,no_inplace}(<TensorType(float32, vector)>)]
Looping 1000 times took 2.982572 seconds
Result is [ 1.23178029 1.61879337 1.52278066 …, 2.20771813 2.29967761
1.62323284]
Used the cpu

Why would the GPU not be available? I know the graphics card is old, but I can’t find a compatibility list on the website to check if it is too old or not.

It’s compatible with CUDA 8.0. It’s possible you didn’t install CUDA correctly, and there are probably a number of other possibilities.

You can validate your CUDA 8.0 install using the instructions in the linux install guide.

while installing cuda toolkit version 9.1 to my pc my quadro 4000 is not recognized. I purchased this card specifically for cuda, is cuda not available on all quadro cards?

All GPUs sold by NVIDIA in the past ten years are CUDA capable, however CUDA’s support for outdated hardware is regularly first deprecated and then eliminated.

The Quadro 4000 is a Fermi-class device with compute capability 2.0 first released in the summer of 2010. CUDA support for compute capability 2.x extended up to and including CUDA 8.0 (where it was deprecated), and was discontinued in CUDA 9.x. You can download CUDA 8.0 from the archive: https://developer.nvidia.com/cuda-toolkit-archive