Awful how all the focus is on Windows, although I do understand it considering Windows has ~90% marketshare.
OT, sorry: Distributed computing is pretty cool IMO, but some reason for some projects go over my head. I'll admit, I'm no math whiz or anything. So what's the point of the Collatz Conjecture? I've also heard that stuff like SETI doesn't matter too much because any radio transmissions from other planets would have dissipated by the the time they got to Earth. I'm sure many of the projects have good reasons, but hell if I can understand.
It's actually a lot to do with how difficult it can be to write and use decent OpenCL compilers for Mac for the purpose of creating dynamic and
memory-effective GPU applications.
NViDIA is pretty good about saying "oh cool, you wanna write something in CUDA? for science? Absolutely, we'll help!!" while ATi kind of does their own thing.
In a way it's great that the newer Macs have the 67xxM series of Cayman cards, since it represents a huge increase in the raw DP-capabilities of GPU-accelerated Mac computing. It's sad that only the Mac Pro currently has the largest capability for actual GPU installation and use. NViDIA makes CUDA drivers for all sorts of Mac platforms and GPU installations, while ATi seems to hide their advanced OpenCL drivers.
Bear in mind that (I think) the APP and OpenCL drivers one would need for BOINC are already installed with your OS.
That said...at the moment only PrimeGrid has a functioning ATi (ati13ati) app for their PPS sieve, and I have not tested it myself yet.
Collatz's admin has long told me and others that if he ever gets a Mac he'll write a better CUDA app, but that nowadays seems moot since if he gets a Mac now it'll have ATi in. That's great, actually, since many projects like that are better suited to ATi cards and the types of math they do better.
You're welcome to try running a few of the BOINC projects (like PG) with ATi apps for Mac...or try your hand at getting Parallels to recognize your GPU, which I've never had success in doing.