Tuesday, August 03, 2010

Cuda on Slackware 13.1

This was much easier than I expected. Note that I'm running Slackware 13.1 32-bit.

Step 1: Install the latest NVidia driver (256.44 in my case). You can't use the SlackBuild from slackbuilds.org. The latest NVidia packages are quite different from 195.*. I uninstalled the existing driver (that I had earlier installed from slackbuilds.org), downloaded the latest NVidia driver, and installed it according to the instructions.

Step 2: Install the CUDA toolkit for Fedora 12. This amounted to downloading the file and running it. I accepted all the defaults.

Step 3: Install the CUDA SDK. Again, I downloaded the file, ran it, and accepted the defaults.

Step 4: Change the paths according to the CUDA instructions.

Done! This was so much easier than Ubuntu, Mint, Debian, and others. I didn't expect it to work by just downloading a few files and running them. I'm certainly not complaining. I ran make for the examples, they all compiled and ran perfectly.


boh said...

Do you think that CUDA works on Slackware 13.1_64?
I don't know whether to make a dual boot with Ubuntu_64 10.04 or just try Slackware.

Thank you! =)

lmf said...

Sorry, I don't blog very often, so I just read your comment. I guess you've resolved your issue.

For the record, I don't think I've ever installed 64-bit Slackware, so I can't really offer any advice there. I'm not sure what, if any, differences would exist.

Giovanni said...

I successfully installed CUDA 4.2 on Slackware64 13.37 by following the guidelines on this tutorial.