Team MacNN has just completed testing on our new Optimized Macintosh clients. They can be found at:
http://members.dslextreme.com/~readerforum/forum_team/boinc.html

If you encounter any problems with them, please post in this thread in our team forum.
http://forums.macnn.com/showthread.php?s=&threadid=243456

Happy crunching for all Mac users from Team MacNN.

---

As far as I understand, these "optimized boinc clients" are just the boinc-side of the code? This shouldn't have no effect whatsoever on the actual computational speed of the climateprediction code (i.e. the worker). I can find source code for the setiathome worker and re-compile it with the IBM XLC compiler (to yield a huge speed boost over the crappy GCC 3.3 version) BUT so far I haven't managed to find the source for the climateprediction.net worker :(

IF the folks have compiled climateprediction worker with GCC 3.3, it is sluggish on the G4 and almost completely not properly optimized for the G5. Unfortunately, the GCC is not up to par on code generation for any other platform than x86 and sparc...

> As far as I understand, these "optimized boinc clients" are just the
> boinc-side of the code? This shouldn't have no effect whatsoever on the actual
> computational speed of the climateprediction code (i.e. the worker). I can
> find source code for the setiathome worker and re-compile it with the IBM XLC
> compiler (to yield a huge speed boost over the crappy GCC 3.3 version) BUT so
> far I haven't managed to find the source for the climateprediction.net worker
> :(
>
The sourse code is not puclic, and it is in Fortran.

> > As far as I understand, these "optimized boinc clients" are just the
> > boinc-side of the code? This shouldn't have no effect whatsoever on the
> actual
> > computational speed of the climateprediction code (i.e. the worker). I
> can
> > find source code for the setiathome worker and re-compile it with the IBM
> XLC
> > compiler (to yield a huge speed boost over the crappy GCC 3.3 version)
> BUT so
> > far I haven't managed to find the source for the climateprediction.net
> worker
> > :(
> >
> The sourse code is not puclic, and it is in Fortran.

A. Where does one get this IBM XLC compiler? Is it already prepped for Mac OS 10.x? Costs?
B. Climateprediction is not only not public, but the original converter to BOINC (previous project manager)...Christiansen...I think his name was...said the code was very proprietary.
C. The really don't want to release it, as they aren't doing cross WU validation against each other.
D. They really really don't want to, as it is converted from an original written for Cray Supercomputers, and they don't really quite understand all of it ;)