The problem is with the Windows benchmark. It has been compiled in such a way that part of the benchmark has been optimised away, and it provides an inflated score. This has been an issue for some years, and I am at a loss to explain why it hasn't been addressed properly before now.

It has been reported to the BOINC devs on at least three occassions. They finally seem to be moving on it.

Being as we have to accept that Windows is the average system OS we take that benchmark as the norm and bring other OS`s inline with it.

I have also emailed Mr. Anderson

carl.h, the benchmark doesn't use an arbitrary scale. It attempts to define a processor's capabilities in the precise number of operations it can perform per second. BOINC uses the well-known and fairly well respected Dhrystone and Whetstone benchmarks.

Mr Anderson (I feel like I'm in the Matrix when I say that) has used your email as the basis for a request to the BOINC alpha testers to gather statistics on the issue.

Lawrence,

Have started to run Rosetta a few months ago for 5%, just to see how it works against the theory. Done 50 or so and it works very well. Each structure/decoy has a given value.....the more structures one works thru the more credit. U can set a WU to run 2, 4, 6, 8, 24 hours or more. After the chosen time it will stop and send the result with how ever many decoys completed. Some WU's have been run for Tens of thousands of times from different seeds by different crunchers. They are thus extremely well known, thus the predictability of how much credit should be awarded can be accurate.

Now on FAAH or HDC, i care to think that model may not work, not technically informed enough if it is comparable. That said, FAAH has now been crunched 29 million times. There must be statistical data derivable from that to come to an award system, but per the table i put somewhere, we have single structure, 10, even 26 and above structured WU's, but all seem to go in 7.5 to 8 hours on my machine...... thus why not give it just the K.I.S.S. and make them e.g. 60 BOINC credits and whilst at it 25 for HDC, Period.....kills all discussion on points keeping all camps happy, ignoring the truth police.

Oh, but only for 'valid' as i've now seen too many with excessive hi claim that without fail, can be predicted to be invalid.

ciao

Hello Lawrence,
Contrary to what you may have heard, XS left Rosetta in force( some stayed) just before the change in the credit structure.This was an issue between us and David Baker. So did others from Teddies,Free DC and other teams as well as individuals not affiliated with any teams.
From what I saw after we left, the first weekend Rosetta had issues with bad WU but sorted that out and to the best of my understanding has a pretty reliable system in place that also is pretty fair to Linux users but isn't to any mac user.
Let's face it, any system that uses an artificial benchmark is never going to be "fair" especially when it fails to make use of the capabilities of 70%(a guess) of todays computers. I'd love to see a system that is work based that's fair to all no matter the paltform or OS.
I know, very easy to say, much more difficult to do.
That said, and this is my personal opinion only as I do not speak for XS, I'm here to stay no matter if you change or not.
It's become a matter of principle to me now.
So as much as I'd love to see some changes, it is not at the top of my list and you won't hear me passionately speaking on the topic.
You have my best wishes no matter which direction you go.
In 2-3 weeks we'll be bringing in a few "surprises" that you will enjoy.
Some of which will rank up there with the fastest air cooled Intel based systems on the planet.
When they come online I'll give you a yell with the specs.
Good Luck!