0

I have an NVIDIA Quadro FX 5800 GPU and want to use it for GPU programming. My current operating system is ubuntu 19.10. I could install "Driver Version: 340.107" from ubuntu additional drivers and now nvidia-smi runs okay.

Based on the searches that I did, I found that to install nvcc I need to install Cuda toolkit which the latest compatible with my graphic card version is Cuda toolkit 6.5. I could download this version from the NVIDIA website but I cannot install it due to its dependencies. For example, it needs gcc 4.8 which is too old. Does anybody have an idea how can I do GPU programming with this old graphic card on my relatively new Ubuntu version?

Improve this question

1 Answer 1

Reset to default
1

It took me several days and trying different methods including a docker image (which didn't work) to figure out how can I solve it:

  1. install the latest compatible driver from ubuntu->software & update->Additional Drivers-> Using NVIDIA binary driver (for my case the latest compatible was 340.107). Make sure nvidia-smi works successfully.

  2. install gcc-4.7 and g++-4.7 by adding the following lines to /etc/apt/sources.list and then running sudo apt update && sudo apt install g++-4.7 gcc-4.7

    deb http://dk.archive.ubuntu.com/ubuntu/ xenial main deb http://dk.archive.ubuntu.com/ubuntu/ xenial universe

    the above lines coud be removed from /etc/apt/sources.list after that

  3. if you have previously installed a newer version of gcc and g++, remove gcc and g++ which are symbolic links from your $PATH and create new sybolic links to newly installed gcc-4.7 and g++-4.7 (e.g sudo ln -s /usr/bin/gcc-4.7 /usr/bin/gcc)

  4. install required libraries by: apt-get install libglu1-mesa libxi-dev libxmu-dev libglu1-mesa-dev

  5. Download cuda_6.5.14_linux_64.run from NVIDIA website and extract it by: ./cuda_6.5.14_linux_64.run --tar mxvf

  6. copy InstallUtils.pm file from the extracted package and copy it to your perl-base location (for mine it was /usr/lib/x86_64-linux-gnu/perl-base)

  7. set your perl library path (e.g export PERL5LIB=/usr/lib/x86_64-linux-gnu/perl-base)

  8. then install the cude toolkit from the extracted package (sudo sh cuda-linux64-rel-6.5.14-18749181.run)

  9. after installation and setting the PATH and LD_LIBRARY_PATH you sould be able to compile and run cuda samples.

a simple example you can find in cuda-samples/0_Simple/simplePrintf and compile it by running make command. Following was the output that I got:

GPU Device 0: "Quadro FX 5800" with compute capability 1.3

Device 0: "Quadro FX 5800" with Compute 1.3 capability cuPrintf() is called. Output:

[1, 0]: Value is:10 [1, 1]: Value is:10 [1, 2]: Value is:10 [1, 3]: Value is:10 [1, 4]: Value is:10 [1, 5]: Value is:10 [1, 6]: Value is:10 [1, 7]: Value is:10 [3, 0]: Value is:10 [3, 1]: Value is:10 [3, 2]: Value is:10 [3, 3]: Value is:10 [3, 4]: Value is:10 [3, 5]: Value is:10 [3, 6]: Value is:10 [3, 7]: Value is:10 [2, 0]: Value is:10 [2, 1]: Value is:10 [2, 2]: Value is:10 [2, 3]: Value is:10 [2, 4]: Value is:10 [2, 5]: Value is:10 [2, 6]: Value is:10 [2, 7]: Value is:10 [0, 0]: Value is:10 [0, 1]: Value is:10 [0, 2]: Value is:10 [0, 3]: Value is:10 [0, 4]: Value is:10 [0, 5]: Value is:10 [0, 6]: Value is:10 [0, 7]: Value is:10

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .