Rosetta@Home - 05/09: Stats

[BlackMountainCow]

:beer: Please welcome EdvardasR to [TA]R@H :beer:

[TA]R@H is now 138 members strong.




Need some stats? Take a look at Free-DC, BOINC-Stats and BOINC-Synergy

Got some questions? Try the Official R@H FAQ (work in progress).

Want to know something about the medical relevance of Rosetta? Take a look at this article.

Interested in alpha testing for Rosetta? Take a look at RALPH.

Got problems with a stuck WU? Check out these threads here, here and here.




Yesterday's highest overall climber: Thehive (1,069 spots)

Yesterday's highest TeAm climber: Pokey (3 spot)


Yesterday's production for TA: 95,910 (day before: 97,681 = -1,771)

TA's RAC 95,321

TA's total score 13,360,667

TA's world rank is: #4




Milestone Makers:

Kelemvor passed 200,000 creds :beer:




Yesterday's top 25 producers:

• TeAm Enterprise (18,299 creds)
  
• Rebel Alliance (12,250 creds)
  
• Sofa King (9,268 creds)
  
• Fardringle (7,915 creds)
  
• BadThad (6,055 creds)
  
  
• TA_TheReaper (5,637 creds)
  
• OhioDude (5,095 creds)
  
• Dalephi (4,979 creds)
  
• TA_GeoffS (2,930 creds)
  
• ken008 (2,734 creds)
  
  
• Dennis (2,502 creds)
  
• biodoc (1,564 creds)
  
• chessaroo (1,165 creds)
  
• TA_andy (1,034 creds)
  
• Kelemvor (965 creds)
  
  
• strider7 (802 creds)
  
• Thehive (782 creds)
  
• Fox (762 creds)
  
• Climb4 (743 creds)
  
• Strikermike (715 creds)
  
  
• Pokey (715 creds)
  
• idlorj (630 creds)
  
• MDE (627 creds)
  
• GimpyOne (530 creds)
  
• Surge_On (515 creds)

SitRep for the TeAm and Rosetta News:

• Please welcome EdvardasR to TA-R@H! :beer:
  
• We had 57 active members (41.61%).
  
• Today's congrats go to Kelemvor for 200,000 crunched creds :gift:
  
• The Rosetta application was upgraded to 5.12. New version includes the ability to send out workunit descriptions and rescaled graphics for big proteins. For more information see
  
• Directly from the Rosy site: "We are sending out a newsletter that was written by the volunteers to all participants whose preferences allow (so, no, we've not become a SPAM host). The letter outlines the upcoming CASP contest and our greater need for computing power during that time. Sending out newsletters will be a very rare event - primarily because it is a big pain."
  
  I guess you've hall received the newsletter as well; I did and I've relocated another core to Rosey Hope you do so, too!

:beer::beer:

BMC with thx to CyGoR

 

[RobertE]

Some interesting information from Dr Baker found Here

The lead article in the June issue of Scientific American has an article descibing some of our work on engineering new molecules. I've been discussing with the editors an article on rosetta@home. With CASP starting in a couple of days, we will have a parallel competition for the "top 5 teams" just as in CASP we get to submit the "top 5 models". We will be keeping track of the total credits earned by each team from the period May 10 to Aug 1 when CASP ends, and will describe the winning teams and their contributions to the CASP prediction efforts in the above article and in book chapters on distributed computing we will be writing at the end of the summer.

We are solidly in 4th right now. So looks like sometime during the CASP run the TeAm is going to get some more exposure. :beer:

As always thanks for the stats BMC!:beer:

 

[Insidious]

Welcome to Rosetta@Home EdvardasR :beer:
Thanks for joining our team!

and a great milestone! Congratulations on 200K Kelemvor :thumbsup:

I can't wait to see how Rosey does in CASP7 (not that I would even be thinking about comparing results with that other Folding team. :evil: )

Thanks for the stats BMC!

-Sid

 

[RobertE]

I'm not sure anyone is at home at the other folding site. The one that starts with P correct? The main page still stalks about CASP 6. How sad. They went from one of my favorites to totally detached. I wonder why anyone still crunches for them.

 

[Insidious]

Originally posted by: RobertE
I'm not sure anyone is at home at the other folding site. The one that starts with P correct? The main page still stalks about CASP 6. How sad. They went from one of my favorites to totally detached. I wonder why anyone still crunches for them.

yeah, that one really didn't get off the ground. (Pred.)

I was actually thinking about our F@H folks. I am expecting that that project will do very well. (I crunched for them for a few years before switching over to BOINC)

-Sid

 

[petrusbroder]

Quotation edited by petrusbroder:

Originally posted by: RobertE
I wonder why anyone still crunches for them.

Well one explanation may run like this:
1. I have lost 1 052 crunching hours due to "bad WUs" in Rosetta and none (none at all) in Predictor. Had only 4 bad WUs, which crashed within seconds ...
2. Predictor does validation of the results (by using a quorum).
3. Predictor'a Mac client is rock-stable and the gives no problems with the WUs

Predictor is not as good at Rosetta in keeping in touch, but as far as I know (and I may be worng) the science is as solid as Rosetta's. Yes, they had problems with getting WUs out, but thats what spare projects are good for.

For me it is important that the crunching runs smoothly, with no resets, restarts, crashes, hung computers. I have had more of that with Rosetta than with all the other 11 project I have tried together (and I have run all projects for at least one month and on at least 10 computers at a time). Since I am a stats-freak I keep tabs.

# of Bad WUs _____ Lost Hours ______ Project ______ Total credits earned
___ 9 ____________ less than 1 ____ Seti@Home _____ 267 000
___ 2 ____________ less than 1 ____ Einstein@H _____ 142 000
___ 4 ____________ less than 1 ____ Predictor@H _____ 83 500
___ 0 ____________ none _________ CPDN __________ 47 500
___ 2 ____________ less than 1 ____ LHC@Home _____ 31 000
_ 1319 ___________ 1 052 hours ___ Rosetta@H _____ 103 000
___ 0 ____________ none _________ SIMAP _________ 56 500
___ 4 ____________ less than 1 ____ BBC-CCE _______36 500
___ 0 ____________ none _________ MalariaControl ___12 000 (alpha and beta)
___ 0 ____________ none _________ QMC@Home _____47 500 (beta)
___ 0 ____________ none _________ PrimeGrid _______11 000 (alpha, just for 11 days)

Less than one: usually a few seconds: most of the WUs crashed within seconds. BBC-CCE is the second most "crashing" DC-project: there I lost 20 minutes of crunching time without getting credits.

I think this explains a lot, at least for me. I have not left Rosetta. I like the project a lot but I am also cautious. Why so many hours lost in Rosetta? At the time of most crashes I could not baby-sit all my comps more than once or twice a week. One time all my computers hung on the 1%-WUs for 36 - 47 hours (for a total of 560 hours lost) because I was away on a business trip. I have been checking on the comps running Rosetta each day since then and still lost up to 24 hours now and then ... that adds up. This also shows that I still believe in Rosetta ...

 

[petrusbroder]

Looking at the numbers above --- I must be crazy! Or a sucker for punishment. I am very glad that I have had no Rosetta crashes at all for the last 2 weeks! The last version which crashed on me was 5.06.

 

[Dalephi]

Thanks BMC.

Welcome to EdvardasR!!!

Congratulations to Kelemvor for the nice milestone!

Somebody should post a copy of the letter from Rosetta on the forums. (I would do it but I am to shy).

Peter: I noticed that in the letter "All work units will receive credit even if they return late or with an error result". I don't know if this is new or they have been doing this all along, but it makes those bad work units a little easier to stomach.

 

[Freewolf]

Originally posted by: Dalephi
Thanks BMC.

Welcome to EdvardasR!!!

Congratulations to Kelemvor for the nice milestone!

Somebody should post a copy of the letter from Rosetta on the forums. (I would do it but I am to shy).

Peter: I noticed that in the letter "All work units will receive credit even if they return late or with an error result". I don't know if this is new or they have been doing this all along, but it makes those bad work units a little easier to stomach.

Dear Rosetta@home Participants, A wonderful aspect of Rosetta@home for me has been the marvelous contributions of the many dedicated volunteers around the world. We are about to begin the 7th biennial CASP Structure Prediction Experiment and from May 10th until August 1st our need for computing power will be even greater. In a bid to solicit more of your 'crunching time', R@H volunteers have put together the following letter which we are sending to all participants. With all of your collective efforts it should be possible to do wonders! Thank you! David Baker

--------------------------------------------------------------------------------

Dear Rosetta@home Contributors,
We are writing you to thank you for donating your computer time to Rosetta@home. We'd also like to bring you an update on recent improvements as well as ask for your help during this year's Critical Assessment of Techniques for Protein Structure Prediction (CASP7).

Rosetta@home is working to develop software to accurately predict the folding structure of proteins, which is a key function in developing cures for diseases such as AIDS, cancer, and Alzheimer's. The Rosetta software is used by other scientists and distributed computing projects. Any improvement we can make in this project will benefit other scientists and projects as well.

Cutting-Edge Research with an all new Application package
The Rosetta@home science application has improved greatly in recent weeks. We now have "smarter" software and new techniques which help find the best structures faster. We have also released a number of new features:

Work unit runtimes based on preference setting can minimize your internet use.
More frequent saves (checkpointing), means more science output and more credits for you and your team.
The "Watchdog" keeps work flowing by ending work units that appear stuck or hung.
All work units will receive credit even if they return late or with an error result.
All these improvements guarantee a problem-free and most useful contribution of your valuable computer time.

CASP7 is here - Be part of the competition and the winning team!

There are many science teams working on protein studies, and every 2 years they conduct a contest of sorts, to define the current state-of-the-art. Now the 2006 event is underway, and you can help. CASP will present proteins no one has ever seen before and ask the research teams to give their best structure predictions for this computationally intense problem. We are very hopeful the improved Rosetta software, now with the added power of distributed computing, will again find the most accurate predictions.

CASP7 runs from May 10th to August 1st. The more computing power we apply, the better the predictions become. Please be sure to run Rosetta@home, at least during the three month period of the competition. Read more about CASP7, and check for published results on the Rosetta@home website through the end of the year.

Be a part of the project team with new information and user recognition
A number of new informational features have been added to help you participate with the research team:

"Dr. Baker's science journal" where the project's lead scientist posts regular updates and project news.
Work Unit information describes the proteins and tests being performed for each work unit type.
The Application Version Log explains new features and scientific improvements in the application.
The project staff is very responsive to questions and suggestions and is working to incorporate them into Rosetta application improvements.

What you can do to help
If you discontinued Rosetta for any reason, please attach again and you will find the application improvements have resolved the problems many were having.
Dedicate a greater resource share to Rosetta, at least during CASP7.
Attach additional computers to the project.
Contact your friends and teammates and help them learn more about Rosetta so they will want to crunch more too. Here is a simple explanation you might use.
If you are a member of a team, please ask for more Rosetta processing time on your team message boards.
Thank you for your past participation, and we hope you can provide additional support to Rosetta@home in the future.

Mark, Joachim, Gerry and the Rosetta@home volunteers

P.S.: Still not convinced? Take a look at "10 reasons why I support Rosetta@home and you should too."

P.P.S.: Your account preferences suggest that you wouldn't mind recieving our rare newsletters. If this is not true, you can 'opt-out' of future mailings by changing the "Should Rosetta@home send you email newsletters?" setting in your account preferences here: Rosetta@home preferences

 

[BlackMountainCow]

@ Peter:

ZERO crashed WUs with CPDN? Wow, then you're one of the most lucky guys I know in CPDN! I really haven't heard of that before. Of my 20 CPDN WUs only one made it through to the end, all other failed more or less quickly, meaning some after 300 hours work, others after 10 hours.

 

[petrusbroder]

Yeah, I know. It was during TAS' CPDN-month I got all those points - I did not have the patience to crunch CPDN any more after that.
OTOH you get the trickle-credits in CPDN, so at least you get the stats. When something in Rosetta crashed (not now but before) you got no credits at all.
That has changed for the better.

 

[petrusbroder]

Originally posted by: Dalephi
...
Peter: I noticed that in the letter "All work units will receive credit even if they return late or with an error result". I don't know if this is new or they have been doing this all along, but it makes those bad work units a little easier to stomach.

I agree - but I lost most of my time (and credits) in 2005 before this new policy started.

 

[Insidious]

nm

 

[Dalephi]

Originally posted by: petrusbroder

Originally posted by: Dalephi
...
Peter: I noticed that in the letter "All work units will receive credit even if they return late or with an error result". I don't know if this is new or they have been doing this all along, but it makes those bad work units a little easier to stomach.

I agree - but I lost most of my time (and credits) in 2005 before this new policy started.

So I take it that this is a new policy then? I lost several thousand points in the last few months due to bad WU.

 

[Insidious]

It's been around for a little while now.

(They run a script every Friday looking for, and scoring those WUs )

-Sid

 

[Dalephi]

Thanks Sid.