I was just looking at the results for the top teams:

Name Members Recent average credit Total Credit Country
1 XtremeSystems 246 311,043.65 25,052,975.81 International
2 Dutch Power Cows 1071 261,066.85 24,927,849.52 Netherlands
3 Free-DC 154 174,032.69 28,027,746.46 International

this is serious computing power; I think the recent credit numbers correspond to
3,100, 2,600 and 1,740 computers crunching more or less full time
for these top three teams which is fantastic!

It looks like the DPC have nearly caught up to XtremeSystems in total credit, but XtremeSystems is
moving ahead faster. Why does Free-DC have the most total credits but considerably less than the other two teams recently?

We will have to have prizes when CASP finishes at the end of July for the top overall team and the
top team during CASP. definitely a citation in the science paper on rosetta distributed contributing at the minimum!

Actually, you caught those numbers after they'd just changed.
Xtremesystems has finally caught up to and overtaken DPC for total credit ( 2nd place ), and is moving ahead and creating seperation.
Free-DC is indeed ahead, but, lately, has not been putting out all that it can in Rosetta. Expect Free-DC's numbers to go up ( indeed, they've already started to rise ) as they move to fend of the challenge by Xtremesystems.

Scott

---

I was just looking at the results for the top teams:

Name Members Recent average credit Total Credit Country
1 XtremeSystems 246 311,043.65 25,052,975.81 International
2 Dutch Power Cows 1071 261,066.85 24,927,849.52 Netherlands
3 Free-DC 154 174,032.69 28,027,746.46 International

this is serious computing power; I think the recent credit numbers correspond to
3,100, 2,600 and 1,740 computers crunching more or less full time
for these top three teams which is fantastic!

It looks like the DPC have nearly caught up to XtremeSystems in total credit, but XtremeSystems is
moving ahead faster. Why does Free-DC have the most total credits but considerably less than the other two teams recently?

We will have to have prizes when CASP finishes at the end of July for the top overall team and the
top team during CASP. definitely a citation in the science paper on rosetta distributed contributing at the minimum!

Actually, you caught those numbers after they'd just changed.
Xtremesystems has finally caught up to and overtaken DPC for total credit ( 2nd place ), and is moving ahead and creating separation.
Free-DC is indeed ahead, but, lately, has not been putting out all that it can in Rosetta. Expect Free-DC's numbers to go up ( indeed, they've already started to rise ) as they move to fend of the challenge by Xtremesystems.

Scott

And expect the Xtreme vs Dc fight for the top to yield mind boggling production.
All of this in an environment of full respect and admiration as the Xtremes and Dcs respect each other capacities, work ethic and dedication to a project.

A historical note : The Cows went into stampede mode ( while leading the Xtremes) on April 1st. After their stampede fizzled, the Xtreme BBQ juggernaut took over and by May 6th some of the cows were roasting .

Who said science cannot be fun? LOL

---

Daivd Baker,

A worrd of caution (well one of the big problems with the current credit granting setup of rosett@Home)

Due to
- The open source nature of BOINC, we can compile our own BOINC client, and hence we can effect the benchmark substantialy, often a simple recompile can create a 1.5x credits/hour increase for older AlthonXP, PentiumIII's. Even more for the newer Athlon64 and P4's as they have SSE2 & SSE3 to bump the score up further.
We can of course just claim whatever we want in effect.
- Couple that with the 'we get what we claim' of Rosetta@Home situation, due to the nature of no redundency settings.

So 'credit' and 'RAC' do not show the REAL crunching ability, so if you yourselves are going to be giving money away or awards, you need to sort this out really, Seti-Enhanced has currently been released in circulation using a new fpops counting technique, I don't know if it will work over here but it there and works well at seti now. It ignores any benchmerk made by 'BOINC'.

In other words,
If team A is getting a RAC of 3000, and team B is getting a RAC of 2000, then they may actually be producing the same overall, team B could be using the standard boinc client, team B could be using 'optimised (benchmark inflating) clients', e.g Truxoft, Crunch3r, Boinc(french one forget name). Also by default the Linux client claim LESS credit than the Windows client.

I used the 'optimised, benchmark infalting client' myself, but for the added capability of it, extra credit is a nice side effect ;-)


I do like the idea, but not when the current credit situation is a shables, I think it is not the time. Maybe think of something else ?

I can also say the Free-DC and I beleive DPC certainly use these benchmark 'inflating' clients. BUT they are still the strongest teams and have a substantial number of computer behind them. Free-DC have a lot of there power covering over project in BOINC and outside of the BOINC world.


Just a note to consider.

---

Team mauisun.org

Fluffy: I can guarantee you that among the three top teams : Xtremes, Free DC and the Cows, the competition is being taken in a well humoured manner. A lot of subtle (okies sometimes not so subtle) baiting and trash talk but, the respect that these teams have for each others production capacities and for their "work-ethic" and dedication to the projects is evident at least to those of us who belong to the teams in question.

I can tell you that while being numero 1 is a great motivator, the biggest source of pride that both the the Xtremes and the Cows got from the "Stampede/BBQ" match was the amount of Wu's that were completed for Rosetta. We were proud that while having fun we were able to help and help a lot

It is a known fact that the potential and real computational power of the Free DC Teams is awesome, nobody among the opposing teams doubts their integrity, dedication and work ethic: they are a well oiled and powerful machine. I can vouch for the Xtremes, they are a team in all aspects of the word team. And even though I had some scrapes with some of the cows: it was all in good fun, they are an excellent team.

So when the DCs decided to engage now to defend their number 1 position, there was joy in the Xtreme hearts. For as I said: The DC's reputation as an honest, hardworking, dedicated team is well known. So I can bet you that what you will be seeing is a real production upsurge, measurable using any standard. Of course there will be baiting and calls to man the decks and all that hoopla. But rest assured that the teams currently involved in the fight for numero 1 , will also measure their success on the total amount of real work they produced for Rosetta. The competition and all the fun part is important to us but believe me, the progress of science is our pimary goal.

All I hope is that all the teams join in. That they ramp up their wu production and allocation of resources, specially during CASP7 time. Every bit of computation time we can provide is needed and I know that the David lead team of scientists will use the information we provide for the best.

---

I do like the idea, but not when the current credit situation is a shables, I think it is not the time. Maybe think of something else ?

Fluffy, I agree 100%.

If the project is going to give any "award" to crunchers, until R@H implements a fpops SETI-Enhanced type of credits or a F@H-type giving standard credits per WU type, let's do it measuring number of MODELS (predicted protein structures) returned, NOT Boinc credit claims.

Although model times can vary significantly between WUs, from 10min to 4+hr per model, I think overall it should even out, especially for big teams.

My 2 cents.

---

Best UFO Resources
Wikipedia R@h
How-To: Join Distributed Computing projects that benefit humanity

I totally agree with TioSuper.
(The first time I hear such wise words from someone of XS ... little joke :p)

To be serious:
All 3 top-teams (FreeDC, XS, DPC) uses the optimized clients. Within the teams it isn't accepted that 1 member will use other ways to increase his credits.

The main goal is the science. After that it's fun to have a battle.
With our stampede, we succeeded in a very important thing:
As a reaction, XS followed our production raise, activated a bunch of new members, and as an effort even the team-spirit became better.

We hope we can continue to this battle for the next period! And this in a sportive way.

---

Member of Dutch Power Cows

I've just spent an hour watching your lecture Mr Baker...

All I can say is wow!

I'm certainly going to have to watch it at least once more (probably many times over) to try and understand everything (Yes, eyes glazed over at least once!).

TBH, I didn't notice your umms and ahhs. More busy trying to understand what you said.

Have you any others that we can watch? Even if not by you, it'd be good to understand more about the science behind the Rosetta project.

As for the sociology, why don't you invite some sociologists to start watching (and helping?!) with the project? (As long as everyone doesnt mind being a bit of a guinea pig?)

As for the credit things, why dont you just set a production based credit granting system?
Instead of benchmarking jobs etc, why not just set say 20 credit per model processed, and then its up to each individual how fast they can proccess each mode. The faster the machine, the faster the model finnished, the more credit an individual can get. Obviously, you'd set a higher credit for more complex (longer protiens, or more detailed runs). But you'd set a value when you set up the work run, for each job that gets sent out. And it doesnt matter how long each computer spends on an individual work unit, because you're measuring by models produced, not by work units produced.

---

As for the sociology...

While you kind of acknowledge that some hand-holding is necessary with a many-personed project such as BOINC, your audience is also aware as you said that further purchases by you of computer farms is no longer necessary, and the cost of TLCing BOINC is much less expensive and time consuming than the time, energy, machine maintenance, power cost, etc. that your BOINC volunteers are donating to you! ;)

---

Regards,
Bob P.

But lets make this clear: the main goal is to motivate production, the competition cannot be allowed to degenerate into name calling and all the worst things that competition brings out.

Another thing to be think about and to be careful of: If you mix in non-CASP7 work units which continue your research, there may be a temptation for folks to delete these and work only on CASP7 units if these are the only ones to be credited in the competetion. The solution of course is to include all units in the competition so your non-CASP7 projects do not get short changed.

Part of the BOINC sociology! ;)

---

Regards,
Bob P.

PS. I know this is unpopular.

So let me follow this with another possibly unpopular observation: What I as a small fry in the Rosetta effort as well as the big time people are contributing is my CPU Processing time. I cannot guarantee a Wu or many Wu's per time unit I contribute because first of all: not all Wu's are created equal, not all computers are as efficient as the others and worst it seems that certain type of errors that keep creeping ("the 107s") up and that are still there are preventing me and others from completing the workload that was sent to us.

So if the credit is going to be based in number of WUs produced, to be honest, I will feel that my effort, worst what I am actually contributing to the project which is my cpu idle time is not being appreciated and recognized. What motivation will a person like me will have to keep running a WU that as of now has been more than 2 hour's at close to 1.4% with more than 6 hours estimated to complete when at the end I will be given the same consideration, similar credit to the to Wu produced by a more powerful computer, or of a smaller amino acid chain or of a type of computation that requires less time? Again what it is being donated is time.

Ditto for a model based on models produced: I can get a chain that spends 6+hours and still produce a single decoy because that is what can be produced given the algorithm/computational model/procedure I was asked to run while another person may get the type of project that is rabbit like in producing decoys in less time than it took me. Again TIME contributed is the commodity being donated, the number of decoys produced do not have a direct relationship to the time donated.


I am new here, but I seem to get the idea that the whole credit thing is being questioned in such a way that assumes that everyone if given a chance will cheat and is forgetting the simple fact: TIME is what is being donated. Those who contribute more time should be given more credit.

When you can develop a way to take into account (and somehow make uniform and or comparable ) the difference in WUs nature and size and in the processes that we as volunteers are being ask to run then we are given data to run, then we can start talking about a WU based credit system.

I am going to sound even more heretical: Let's not compare the purpose/priorities of the Seti@Home project with the Rosetta@Home Project: they are as different in the nature of what they are searching for as night and day. Let's not try to change the focus of the developers and scientists at Rosetta and divert , time and other resources in solving a problem that exists in the mind of a relative few people. And worst let's not demean the contribution of teams who are contributing a large share of resources : A LARGE Chunk of Computational time to the efforts.

Worst, lets not continue raising diversion at a time where our whole focus should be the CASP7 trials and not if X or Team A got 2 more points per second than I did . For what can happen is that many people can get tired of this diversion and stop contributing their seconds of CPU time regardless of the number of credits. ( There is a point this whole he/they cheat thing gets boring).

I rather have the team spend more time/resources in solving the hassles that the 107 type of errors that are still popping up are causing in time used than trying to solve the "credit issue". For all I can bet any of you $5.00 that once the credit issue is "solved" a new group or individual is going to raise the issue that the new model is unfair. Alas that is human nature.

Okies: Start roasting me :).

---

Tio,

Ah but the the point is 'you get paid for how much work you do'
So, either I could sit on my arse and doodle for a while, get the job done in 2 weeks and get paid more (2 weeks worht of pay) than the person that is working their arse off to get it done in 2 days (so only get 2 days worth of pay).
Where I believe in a project where 'work done' is what is important (as appose to projects like XtremLab where time done should be the measurment since there is no 'work done' as its the same no matter how fast the computer is) then that is what you should use.

As a side, how do you also include the time it takes me, physically, to manually send all 3 computers (and soon to be four when that hits 100% complete in a CPDN model in 3 day) on a pay per minute dial-up connection since BOINC seems unable to handle the proxy I have to go through on uploads. Downloads seem fine and some projects seem fine for uploads, rosetta unfortunaly is not one of them, same for CPDN, not even the latest 5.4.7). So I have to check daily for a collection of finished jobs, to then turn over the network to a different dial-up, easy but time-consuming.
But I do it for the fun and the nice people here :-) Credit can not score that :-)

Anyways,
On to a more important aspect of 'credit' It gives a project an overview of it's through-put in 'speed' terms, although it know a general feel, it cannot really compare if what you are measuring is false or skewed differently from week to week. It aids in planning if larger and lpnger job sets can be done and what sort of time fram to expect. It also give the project information.


-----
As for Free-DC I know them quite well ;-) and DPC well I've had chats with some of them during the FaD stampede. So I've seen it before :-) It's good and a enjoyable to watch the cows on the BBQ every time, lol

---

Team mauisun.org

Fluffy:

Well, if I were to use your argument: I shouldn't even be here. See for reasons unknown to me my computer has slowed down to the point of molasses when it comes to the Work Unit I am processing... a work unit that I have no control whatsoever in choosing and that at the rate is going it is going to take more than 200% the estimated CPU time I set my BOINC Manager to provide. So I wait patiently for this Wu to finish and to told then, that because it took me more time to do, the job I was assigned, when I have no control over what I got and because I cannot afford the luxury of a fast computer at this time, the time I contributed is worth less than the time other people contributed. (Again all that we can tribute is computing time)

The results are a direct result of the validity of the models the scientist/developers give us. The length and complexity of the protein chain is the scientists choice too and that too has a direct bearing on the number of decoys my machine will produce. Shoukd I refuse wus I know are large and will slow my production rate? Pass the hot potato to another?

All we as volunteers do is volunteer our computer time and that is the only thing we can control so, in fairness that should be the sine qua non criterion for evaluating us. Please remember that some of us have to use our computers, most of us do not have multiple rigs ( BTW the computers listed can be the result of the lack of a merge function) If we could choose the type of protein, computation mode and real time estimate of the units we get, then in fairness you can judge us for the speed of the results we get and the number of units we can produce.

Speed doesnt guarantee quality of results and reliability of them, nor does quantity of the units produced. All I hope my tortoise of a computer can produce something that helps: some valid, replicable and reliable science. That is all I can provide the scientists: time for the chance that good science is produced.

So I ask you what is this slow poke to do? ( And this can apply to others who like me: own a single computer and have to use our computer in other things other than distributed computing projects) Leave the project because I am slow, because my work product is not considered by some artificial standards and criteria I cannot and do not control not as worthwhile as those of others who can do unit work fast, who can do more decoys per unit than me?

That is what many of us are reading into the discussion that is being raised about credits. That is why I am finding so hard to recruit people to join me here. It seems that unless we are fast we are not wanted and not valued.

My 2 cents worth.

***********
One a more serious note: the cows do look nice on a grill :).

---

Tio, each model done at whatever speed is valuable, the team www.boincsynergy.com understand that. It's important you feel comfortable at what you're doing and I feel comfortable being a member of www.boincsynergy.com. www.boincsynergy.com welcomes all crunchers from any nation, any speed rig, and any project. Heck, www.boincsynergy.com even has "flash" games to relax with. Bring yourself over to where you might feel comfortable. you know where that is "www.boincsynergy.com".

This is starting to sound like an add. Seriously, Tio, every puter is valuable, I don't know many who actually think they can ever be number one or even in the top 10. Keep crunching because you like the project.

Tio, each model done at whatever speed is valuable, the team www.boincsynergy.com understand that. It's important you feel comfortable at what you're doing and I feel comfortable being a member of www.boincsynergy.com. www.boincsynergy.com welcomes all crunchers from any nation, any speed rig, and any project. Heck, www.boincsynergy.com even has "flash" games to relax with. Bring yourself over to where you might feel comfortable. you know where that is "www.boincsynergy.com".

This is starting to sound like an add. Seriously, Tio, every puter is valuable, I don't know many who actually think they can ever be number one or even in the top 10. Keep crunching because you like the project.

You may not wan to believe it but, I found the same spirit of camaraderie at the XtremeSystems Team. I am happy there being the "least among equals"

I am sorry to say that every time I hear an argument for a revision of the credit system: I cringe. Those calls seem to argue for the worthiness of some unit performance parameters over others when the only performance parameter we as volunteer can control and do control is the amount of time we run the projects. To be told that I am going to be evaluated on other criteria I cannot control or with criteria that do not take that TIME CONTRIBUTED is the only criterion I can control is to be told that the time I contribute is not as good as the time others contribute because the others are faster and more efficient ( efiiciency measured in ammount not quality of work) . That I am a lesser type of contributor because I do not own a mighty powerful fast machine.

Sorry if I sound harsh but, this how I am starting to feel every time a new thread on credit tweaking and cheats is opened. I am starting to get a tad annoyed and frustrated.

---

The point is to encourage more production, by increasing the amount of time the computer is running, the percentage of time Boinc runs Rosetta (if it's setup for multiple projects), by encouraging us to run Rosetta on multiple computers, or to go out and encourage others to run Rosetta.

All of us have the opportunity to do one of those 4 actions, if not most of them.

Basing the crediting system on hours - would encourage the stats freaks into setting up large collections of dirt cheap under spec-ed machines that will never return a single model/decoy. Via 100Mhz cpus with too little ram, and Win98 or Linux for example. 40 such machines wouldn't equal the model/decoy production of an Athlon running at 2Ghz. (The underspec-ed systems would be running with swap space - and that's much slower than running in Ram.)

Encouraging us to actually finish models/decoys - which is the main purpose that we're serving for this project - makes much more sense to me. (And this may want to be moved to another thread..)

To Dr. Baker: Don't they release the native structure of the protein we're working on - sometime after we're no longer allowed to submit data for that protein to Casp? i.e. Can't we have an RMSD result before the year end comparison data is released? (Just to sate our curiosity?)

---

....You may not wan to believe it but, I found the same spirit of camaraderie at the XtremeSystems Team. I am happy there being the "least among equals"

I am sorry to say that every time I hear an argument for a revision of the credit system: I cringe. Those calls seem to argue for the worthiness of some unit performance parameters over others when the only performance parameter we as volunteer can control and do control is the amount of time we run the projects. To be told that I am going to be evaluated on other criteria I cannot control or with criteria that do not take that TIME CONTRIBUTED is the only criterion I can control is to be told that the time I contribute is not as good as the time others contribute because the others are faster and more efficient ( efiiciency measured in ammount not quality of work) . That I am a lesser type of contributor because I do not own a mighty powerful fast machine.

Sorry if I sound harsh but, this how I am starting to feel every time a new thread on credit tweaking and cheats is opened. I am starting to get a tad annoyed and frustrated.

"Tio"

I understand your point. But I will quickly add that your contribution is valued by this project on an equal footing with all others who contribute. Rosetta uses the credit system that is standard in the BOINC system. The credit system will not change unless BOINC changes. That change is on the horizon in the form of "Tflop" counting. What is interesting about that system is that it is really a complicated way of counting work units.

The current system of credit awards used by BOINC relies on Redundancy as a validity check to prevent cheating. Rosetta does not use "redundancy" in its credit awards. Redundancy (sending the same work to multiple systems) is not just impossible with the work Rosetta is doing, it is wasteful of the valuable contributions of users like yourself. So the fact is that a number of users will continue to revisit this issue at every opportunity.

No matter the system used someone will not like it or think it is unfair in some way.

Your points are well taken. But please remember you and the time you contribute are important to Rosetta. Arguments over the way credits are awarded will not change that. One single model that is processed on your system may hold the key to the riddle of protein folding. The speed that you process that model is not important. What is important is that you be here with us to create that model.

---

Moderator9
ROSETTA@home FAQ
Moderator Contact

"Tio"

Rosetta does not use "redundancy" in its credit awards. Redundancy (sending the same work to multiple systems) is not just impossible with the work Rosetta is doing, it is wasteful of the valuable contributions of users like yourself. So the fact is that a number of users will continue to revisit this issue at every opportunity.

No matter the system used someone will not like it or think it is unfair in some way.

So why go then into credit models that will undersestimte/undervalue the only variable we as voluntees can and do contol: the time we can offer and DO offer?

Mark my words: There are people that are getting tired of the credit issue and who are feeling like Tio, and like me. The vibe we get is that the length of time we spent processing is a function of cheating to accumulate points. That is blatantly unfair to us, but that is the feeling that we (the slow ones) are getting.

The level of frustration I seem to be getting among we the slower ones with those who are continuously harping about the tweaking of the credit system is growing to the point we are starting to question why we stay here.

I detached from the project because of the level of frustration with the 107 type errors (Which BTW seem to be popping in many computers without solution to them in sight) I decided to return and now I am faced with models and Wu's that take a lot of time (something I cannot control) and see the same people harping (and that is the softest word I can find in my vocabulary) ad nauseam about the time based credits that were harping before I left.

Hey maybe the solution to the credit issue is for all of us, the slow unproductive ones to leave this project in the hands of the efficient ones, the fast ones. To be honest with you there is a DC project that recognizes the time we spend and dedicate as that is the prominent data that is displayed in the "screen saver". That simplicity of credit , that absence of credit complainers is making that project more and more attractive to s slow poke like me.'

I am not a patient person by nature and to be honest my patience is running out on the credit issue.

---

This and no other is the root from which a Tyrant springs; when he first appears he is a protector.”
Plato

@mmciastro
I agree with you that the optimized clients can be interpreted as cheating.
For me, it isn't important which crediting system is used. The only thing that counts is:
If someone else is getting x points for a job, I want to get the equal credits for my work done.

If other teams uses the optimized clients, we use this also.
When the other ones decide tomorrow to disable the optimized clients, I'll be the first to disable mine.

---

Member of Dutch Power Cows

I am going to go really radical on all of you;

How about a Rosetta@Home without the so called "CHEATERS' aka the ones that use optimized clients?

Let's have all those people/teams that use optimized clients ("the cheaters" as you accuse them to be) STOP producing work for Rosetta. Do not accept any WU that comes from an optimized client.

Can the rest of you take over the slack and keep the current production levels as they have been for the last Month or so?

If you cannot then please do allow those who are carrying a heavy work load keep working without having to hear from you that they are cheaters.

---

This and no other is the root from which a Tyrant springs; when he first appears he is a protector.”
Plato