Its a MAC, in name only. It's not from Steve Jobs, it's not a slow performer, it doesn't look weird, and it's not overpriced! It's actually the exact opposite!
For those that remember, the Stanford crew are using a new GROMACS core for the new workunits. This new core, as compared to TINKER, will be MUCH faster, as it now has 3DNow! and SSE support.
This doesn't mean that youre going to finish WU's any faster though, as they're going to start sending bigger (and, in many cases, more scientifically significant) proteins to process.
I noticed, on their website, that it also mentions parallelism - maybe, just maybe, we'll be able to run real SMP-supported F@H? Would be neat if we did. It also seems to have a primary focus on linux - that could also mean that the Linux machines gain back the 15-30% lost due to inefficient FORTRAN compilers for linux.
Their website is here:
http://www.gromacs.org/
To see some benchmarks that they did, to see how your current folders will perform (relative to others), check out this link:
http://www.gromacs.org/benchmarks/single.php
They also performed some 'clustering' benchmarks, as shown here:
http://www.gromacs.org/benchmarks/scaling.php
So far, it seems that the Athlon is still King of the Hill. Even comparing a Slot A Athlon to a Coppermine PIII of the same clockspeed results in the Athlon winning. Running Dual Athlons results in a 178% increase in work done... and judging by the comments on the page, they used an original MP board, not an MPX.
It seems to be a very fair benchmark for those Macs too. A PPC G4 at 800MHz is about 95% of a P3 800MHz, and 92% of a Slot A Athlon 800MHz. You guys won't have any problems blowing past those Mac teams. Bring on the fanboys!
.:Ttocs