There's an interesting project for CUDA too, Milkyway@Home (http://milkyway.cs.rpi.edu/milkyway/) but I can't tell how well it does because it only works for higher end nvidia gpu's, and my GeForce 9600GT isn't capable of running it.

You should also consider running FreeHAL@Home (http://freehal.net/freehal_at_home/) because it takes very little cpu usage and so boinc allows it to run in parallel with other cpu intensive projects. And in the preferences you can set how many instances (WU's) you want to have running at once, between 1 and 25, but beware that each of them takes around 30Mb of RAM.