I am asking for a very special favor.

[Jose]

Please check the Top Computers section at Rosetta http://boinc.bakerlab.org/rosetta/top_hosts.php

Please look at the nine computers that are listed above MAOJC's . I want your unbiased opinion about them. :umm: Let's say I do not have a good feeling about most of them.

So what say you all? Should I present those 9 to the Rosetta Developers for severe corrective action?

[tnerual]

you can try with the one placed in position 7

217724 7 Anonymous 4,898.18 80,376.53 GenuineIntel
x86 Family 6 Model 8 Stepping 1 498MHz Microsoft Windows XP
Professional Edition, (05.01.2532.00)


498 mhz and in 7th place ... strange

[LAURENU2]

Jose
Don't you think it would make more sense to ask the Rosetta Developers to make a standardized Boinc+Rosetta client?
As it is now I have heard from the Rosetta forum that people just have to change the value in 1 file to get any score they want.
:umm: Calling up a witch hunt might do more damage then good!:rock:
Pushing for a standardized Boinc+Rosetta client with better encryption is the way I would go:thumbs:

[Jose]

Jose
Don't you think it would make more sense to ask the Rosetta Developers to make a standardized Boinc+Rosetta client?
As it is now I have heard from the Rosetta forum that people just have to change the value in 1 file to get any score they want.
:umm: Calling up a witch hunt might do more damage then good!:rock:
Pushing for a standardized Boinc+Rosetta client with better encryption is the way I would go:thumbs:


Laurenu: this is not a witch hunt. It is a very careful review of the scores. BTW close to 1000 top scores have been reviewed. For very obvious rasons care (a lot of care) is being excercised. The sad part is that whatever is finally decided for the new credit formula is not going to make everyone happy.

That said : man there is no way those scores are true. Some swift action in those cases may result in sending the message to some of the egregious cheaters.


Would you care to develop thetheme of that little file that can be changed? Please PM or email me at joseantonio(dot)choicecable(dot)net

[Paladin*]

Rant On ... I Grind my Teeth when ever I think of the Rosetta Project ... :bang: ...The Cheating started out at 60 Credits Per hour & then it jumped to 600 Per Hour & now it's up to 6000 Per Hour ... :swear: ... Where will it all end, I didn't even run the Project for a couple of Months I was so disgusted with it ...

As far as I concerned the Rosetta Project has let the People make a Mockery of the Credit System & it has Forever Tainted the overall Credit Standings for everyone involved with running BOINC. It has let people put up Hundreds of Thousands & maybe even Millions of Credits they would have never got if the Project would have done something right from the start.

They don't need a Quorum if they don't want one, but a simple Cap on the Credit allowed per Hour of Processing like a lot of the other Projects have that don't have a Quorum would have at least kept the Cheating somewhat under Control.

The way it is right now anybody can join the Project & get whatever Credit they want & thats exactly what a lot of people are doing ... Rant Off ... :smoking:

[cwhyl]

I read somewhere Rosetta will switch to fpops counting soon?
That benchmarking system is a joke.

[Jose]

Lets make this clear contrary to other projects, Rosetta @ Home has a way to verify what is produced in the work units other than agreement by quorum: it has had the confirmed structures of what it sends in the Wu's and in the case of the CASP 7 Trials , it gains access to the structure models once the Protein deadline ends. What we are testing in Rosetta@home is the software they use: the Rosetta Tool. We have been offer an opportunity to help in the development of a tool. So a quorum higher than one is actually it is a waste of resources: the structures we have been "predicting"are known or will be known as soon as they are released at the respective deadlines.

Second and this is obvious: regardless what they choose (and for obvious reasons do not expect any change until after CASP 7 ends) as their credit granting procedure there is going to be gnashing of teeth and howls. There are too many computer types and too many computing "personalities". Several months ago it used to be Opty's , now we are talking about air cooled, water cooled, phase cooled Conroes and soon it is going to be Kentfields so the power to compute hard and compute fast is going to be accessible to many. This at the same time where a large number of computers of the 60 thousand plus involved in Rosetta are your run of the mill (whatever that means ) PC run by people that range from those who do not care about credits to others that are very competitive. So it is going to be a given that when the change comes there are going to be howls. One thing the people at Rosetta do not want is the brouhaha at SETI. That said there are some credits claims that are out of the credibility range and have to be dealt with sooner than latter . ( That is why I need your opinion with the first nine I asked about.)

Does any of you really want what is happening at one of the projects where if your machine gets over the max decided for the project the work unit is not accepted as valid ? So the computing time volunteered goes to waste.

Or are you are arguing that what happens in the newest Vault project , where the first computer of the initial quorum that reports sets the unit credit?

BTW there are ways to "cheat" ( read render a quorum meaningless) with a quorum, speciall if a large number of teamates join a project.

I have been asking you and in another teams MB because I value your opinion. If sometimes I sound like a Devil's advocate it is because I need to get a clearer view of what the dc community stands regarding credits and benchmarks and because I need to present real options to those who asked me to help them.

Again, your opinion is valued. Keep it coming.

[LAURENU2]

Would you care to develop thetheme of that little file that can be changed
Jose
I know nothing of files OR how to change them
All I know is what I have read over at the Rosetta forum
I am not in the IT field and at the most I can only install
programs that install them self:looney:
But it is clear that Rosetta / Boinc security and scoring And Memory requirements are crap:swear:

[Helix_Von_Smelix]

was thinking of switching over to Rosetta, which of the anti cancer BOINC does the best and most relevant to science??

[Exci]

Helix - I'd say the answer to that question will be coming out this fall when the results of Casp 7 are released. Rosetta was the 'winner' of casp6 when they only had access to their in-house servers (~1000 cpus). . I'm excited to see what they're able to do with 2 years of improvement to the code and 50x the number of processors. If they do in fact come out on top, it's likely a large number of researchers (cancer, alzheimers, you name it) will start using Rosetta in their projects (for free) in addition to Baker's lab.

[Jose]

I hope that people have realized that some of the most egregious cheaters (including a full team) were removed.