SETI@Home can now take advantage of CUDA, the parallel processing engine at the heart of NVIDIA GPUs. This means that the GPU’s computations are added to those of the normal CPUs, resulting in faster processing of computation units, which is very handy for churning through the challenging Astropulse calculations.
I now have SETI@Home running on Ubuntu on my Dell XPS m1330. It’s not particularly difficult to set up, but if you want to try it yourself you may find these steps useful.
Ensure that you have the latest NVIDIA drivers installed.
sudo apt-get install envyng-core envyng-gtk envyng-qt
Then run Applications – System Tools – EnvyNG; install the latest NVIDIA drivers.
Download the latest BOINC Linux installation. (Note that at the time I tried it, the BOINC package available from the Ubuntu repos was very out of date, and did not support CUDA. If you have an old version installed, ensure that it is not running, otherwise the new installation will just attach to it. I uninstalled my old version just to be sure.)
When you have downloaded the BOINC client package, execute it. (You may have to ‘chmod +x’ it first.)
A directory called ‘BOINC’ will be created. Move this directory to a suitable location.
Then simply execute the run_manager script. It will take you through the process of attaching to your existing project settings, or setting up a new account.
In the BOINC Manager client, go to the Advanced view. In the Messages tab, you should see a couple of lines saying something like this:
CUDA devices found Coprocessor: GeForce 8400M GS(1)
This means that the GPU has been found and will be used for processing the SETI@Home work units.
You should see some speed improvement as the GPU is contributing to the computation.
Recent Comments