I now officially have the single-machine that does the most work for Rosetta@home.

Screenshot at the bottom of this post:

http://michaelgr.com/2008/08/29/big-cruncher-redux/

Guess it really helped to overclock the Mac Pro. Faster FSB/memory makes a fairly big difference too, apparently.

I'm not really a stats/credit fiend, but I do like knowing that my little comp is contributing to the science that much.

Happy crunching over the long weekend, everybody!

---

congrats! 7 of the top 8 are running Darwin!

---

congrats....

---

Very nice!

After looking at the top performing machines, I can't stop wondering why there's such a huge spread sometimes between the exact same processors. Is it that some machines spend more time crunching for Rosetta than others?

On a similar note, it surprises me that the Xeons are so much better, often 75-100%, than the latest Core 2 Quad's? It can't all be FSB?

RAC does not take in to account the % of time your computer is running BOINC, nor the % of time your computer is running Rosetta. So if a machine also runs another project (even Ralph), it will reduce the Rosetta RAC figure you see.

Also, if the computer actually does some other useful work, this will take CPU seconds away from BOINC. Even when it is running, and is running Rosetta.

---

Rosetta Moderator: Mod.Sense

@Mod.Sense

I had considered taking a representative sample and charting out the top performers of each CPU type, the assumptions being the top performers might be dedicated, but as you point out, the obvious problem is not knowing how much Rosetta work they actually do.

What would be neat is if there were a standardized sample work unit that could be downloaded and run manually by the end user to test their theoretical speed and create a baseline with.

My point with this is to come up with a real world comparison of CPU speeds and power consumptions (as far as that's possible) as they relate to the project.

Thanks for your input!

very nice ...

Very nice!

After looking at the top performing machines, I can't stop wondering why there's such a huge spread sometimes between the exact same processors. Is it that some machines spend more time crunching for Rosetta than others?

On a similar note, it surprises me that the Xeons are so much better, often 75-100%, than the latest Core 2 Quad's? It can't all be FSB?

The rosetta pages don't show overclocking. For example, my Mac Pro has 2 Xeon Quads at 2.8ghz, but I've overclocked them to 3.129ghz (and the FSB gets overclocked too).

The Xeons on that list probably do better than Core 2 Quads because the Xeons are in 2 socket machines while the Core 2s are in 1 socket machines.

---

The Xeons on that list probably do better than Core 2 Quads because the Xeons are in 2 socket machines while the Core 2s are in 1 socket machines.

not only they are in 2socket machines, they usually run 247 on servers...and quad-cores are often shutdown for the night, for example, play video, games, etc..