Here's what a "cobblestone" is according to the Boinc website. P.S the Cobblestone is named after Seti scientist Jeff Cobb

A BOINC project gives you credit for the computations your computers perform for it. BOINC's unit of credit, the Cobblestone 1, is 1/100 day of CPU time on a reference computer that does both

1,000 double-precision MIPS based on the Whetstone benchmark, and
1,000 VAX MIPS based on the Dhrystone benchmark.
These benchmarks are imperfect predictors of application performance, but they're good enough.
Eventually, credit may reflect network transfer and disk storage as well as computation.

Here's what a "cobblestone" is according to the Boinc website. P.S the Cobblestone is named after Seti scientist Jeff Cobb

A BOINC project gives you credit for the computations your computers perform for it. BOINC's unit of credit, the Cobblestone 1, is 1/100 day of CPU time on a reference computer that does both

1,000 double-precision MIPS based on the Whetstone benchmark, and
1,000 VAX MIPS based on the Dhrystone benchmark.
These benchmarks are imperfect predictors of application performance, but they're good enough.
Eventually, credit may reflect network transfer and disk storage as well as computation.

lol :-)

---

Team mauisun.org

Movieman:
You took exception to a quote by Biggles who quoted yet another person, who made reference to the reason a 3rd person stated for using a particular optimized client. Biggles did not claim in that post that he was talking about a team, he did not claim to be talking about your reasons for using the client. Re-read the quote.

The Rosetta team mentioned that it was okay to use optimized clients. During CASP and during his current trip, Dr. Baker has been very quiet on the boards. I missed his daily updates and comments. Real life intruded and ran off with two of the project's moderators. The comments about the optimized client use were not directed at any one team, but at the users of the optimized client. As the clients had already been said to be okay to use on Rosetta by the Rosetta team, the comments should not have been taken to heart by those using the optimized client. Those that were talking about the editing of a certain file that allowed them to increase their benchmark by 1,000 times were accused of insulting your team when your teammembers were represented in the group that was identifying those editing their benchmarks.

When it comes to presenting, what those of a different mindset would call mis-information, over and over, there's more than one group guilty of this. We need to stop the practice of shooting first, and asking questions later. And if neither side can see the position of the other side, let us agree to disagree.

If you want a comment from Dr. Baker, wait until he's actually taking part on the board on a regular basis again.

Regardless, the topic is closed for this thread. We have a new credit system. This thread is for discussing the new credit system. If you have suggestions on how to improve the work based solution we have now, then make them. Members of XS asked for a work base credit system. Members of other teams asked for a work based credit system. >WE< asked for this. Now it's >OUR< responsibility to understand how the current system works, how the scores average out, point out any perceived flaws, and make constructive suggestions. Perhaps we can get the programming staff to step in and describe the system in better detail. But give the Rosetta team a bit more time.

---

Rosetta Moderator: Mod.Tymbrimi
ROSETTA@home FAQ
Moderator Contact

.......Now it's >OUR< responsibility to understand how the current system works.......

Erm, if I remember correctly, we did that kind of basic arithmetic/maths very early on here at primary school in Scotland BTW.

---

I like the new system, and it is here to stay.

Get over it.

So.... about the discussion of the new credit system....

---

Add this signature to your EMail:
Running Microsoft's "System Idle Process" will never help cure cancer, AIDS nor Alzheimer's. But running Rosetta@home just might!
https://boinc.bakerlab.org/rosetta/

This new credit system seems fair to all participants in that it places everyone on the same playing field. Some would like a different field (I.E more credit/hour). With the implementation of this new field the Top Paricipants by "Total credit" will be skewed for months/years to come by the use of third party boinc clients. I was thinking that the Top Participants by RAC would somewhat quickly transform into something more representative of the new system, BUT this morning I realized that until the project implements the RAC decay feature, that this wouldn't be true. Any user who just left would not be contacting the server with work so that it could make adjustments to the users rac. Would it be possible for the project to turn on this feature?

tony

This new credit system seems fair to all participants in that it places everyone on the same playing field. Some would like a different field (I.E more credit/hour). With the implementation of this new field the Top Paricipants by "Total credit" will be skewed for months/years to come by the use of third party boinc clients. I was thinking that the Top Participants by RAC would somewhat quickly transform into something more representative of the new system, BUT this morning I realized that until the project implements the RAC decay feature, that this wouldn't be true. Any user who just left would not be contacting the server with work so that it could make adjustments to the users rac. Would it be possible for the project to turn on this feature?

tony

Or just run it once after 2 to 4 weeks. That should drop the RAC to realistic levels for participants that no longer crunch.

---

BOINC WIKI

BOINCing since 2002/12/8

Another attempt to get backdating and/or adjustments after the fact has started. Now it is called Rac Decay.

Perhaps they do have it on, or it's just the system doing it's thing, but here's a comparison of the top forty by RAC. This compares the postitions of users on the day the new system was implemented vs today.

This new credit system seems fair to all participants in that it places everyone on the same playing field. Some would like a different field (I.E more credit/hour). With the implementation of this new field the Top Paricipants by "Total credit" will be skewed for months/years to come by the use of third party boinc clients. I was thinking that the Top Participants by RAC would somewhat quickly transform into something more representative of the new system, BUT this morning I realized that until the project implements the RAC decay feature, that this wouldn't be true. Any user who just left would not be contacting the server with work so that it could make adjustments to the users rac. Would it be possible for the project to turn on this feature?

tony

So I take it after reading your last table that this post was made without factual knowledge of what was happening.

This new credit system seems fair to all participants in that it places everyone on the same playing field.

It does not. See David Kims comments/dialogue about Power Mac users.

This new credit system seems fair to all participants in that it places everyone on the same playing field.

It does not. See David Kims comments/dialogue about Power Mac users.

It may not be fair, that nobody compiles a Rosetta build for Power Macs, that does use all of its power. But the credits given (same credits for same work done) are fair.

Norbert

---

David Kim mentioned that he turned the RAC decay feature on. It will be running once per week:

https://boinc.bakerlab.org/rosetta/forum_thread.php?id=2219#25556

However I'm not sure whether it will run automatically or whether David Kim has to run it each time manually. It seems to me after the one run 8 days ago it did not run another time.

@Jose

RAC decay is activated on all BOINC projects I know. It updates the RAC not only when a result was reported but for all users/hosts even if they don't report. That is not backdating, that is just showing the actual RAC. If not activated a RAC would stay forever if a host/user departed, which is certainly not the intention of RAC (recent average credit).

David Kim mentioned that he turned the RAC decay feature on. It will be running once per week:

https://boinc.bakerlab.org/rosetta/forum_thread.php?id=2219#25556

However I'm not sure whether it will run automatically or whether David Kim has to run it each time manually. It seems to me after the one run 8 days ago it did not run another time.

@Jose

RAC decay is activated on all BOINC projects I know. It updates the RAC not only when a result was reported but for all users/hosts even if they don't report. That is not backdating, that is just showing the actual RAC. If not activated a RAC would stay forever if a host/user departed, which is certainly not the intention of RAC (recent average credit).

Translation of what I said:

In the context the other person used it, the issue of Rac decay could as well been the restart/re-visitation of the backdating issue.

See my bias against that person reads his post about the RAC as a questioning of why the number 1 RAC is still being held by a member of XtremeSystems. Maybe had he seen the daily stats , that person would have realized that the member of XtremeSystems is still crunching (albeit a minimal amount of work ) Rosetta work units , so his RAC will not drop down as fast as those who have abandoned completely the project. Nevertheless, it has fallen significativelly.

Sad, that person could post the lyrics of Happy Birthday on September 9th in a post dedicated to me and I would check the wording for ulterior motives. Yes, that is the level of mistrust I have for him and his "facts" .

Yes that is BIAS ( I like to see it as well earned dislike.). I am not proud of it but, I will admit to it. Maybe one day... maybe when ET calls home. :)

PS: I know the meaning of RAC and I also known that it being an average it is not a very reliable indicator and I also know that RAC measurements are quirky and in occasions they reflect the fact that a person just started crunching.

This new credit system seems fair to all participants in that it places everyone on the same playing field.

It does not. See David Kim's comments/dialogue about Power Mac users.

It may not be fair, that nobody compiles a Rosetta build for Power Macs, that does use all of its power. But the credits given (same credits for same work done) are fair.

Norbert

The basic unfairness is that Power Macs and other powerful CPUS are not being used efficiently thus their work is not credited fully as it should be. The inefficiency in the use of the CPU is reflected in lower credits. The basic unfairness is that people with power macs are discouraged from joining the projects and thus the project science is denied the results of the new technology. If it is fair to have a systems that keeps using the low end of technology so those people that cannot afford or are not willing or capable of upgrading their hardware participate, it is equally fair to have a systems that encourages those who can afford the technology, are willing to upgrade or are capable of upgrading to participate.

A lot have been said about attracting new members to boinc projects. Why not starting in developing a BOINC client that allows for the efficient use of all the CPUS available, specially the new powerful ones ( this will attract the increasingly large number of people that is investing in new processors and hardware) as well as developing one doesn't punish the users of non-windows OS ( Linux user still suffer under crediting )?

IMHO , want to talk about a fair system, that is a start.

@T.Castro:
FYI:
Vietnam Soldiers has currently one X2 3800 on Rosetta versus the 20 machines that were reporting daily so you should take that into your calculations.
That one will be off of Rosetta as soon as I can contact the owner of the rig.
I think that premise will also hold true across a lot of the top contributers on your chart from XS.
By my calculation there are app 100-130 machines currently on Rosetta versus the 2000-2500 that were on 3 weeks ago.
This will greatly effect the RAC and was expected.
Jose my friend, lets leave these gentlemen to their discussions and get on with our lives.
Thank you.
Movieman

---

@T.Castro:
FYI:
Vietnam Soldiers has currently one X2 3800 on Rosetta versus the 20 machines that were reporting daily so you should take that into your calculations.
That one will be off of Rosetta as soon as I can contact the owner of the rig.
I think that premise will also hold true across a lot of the top contributors on your chart from XS.
By my calculation there are app 100-130 machines currently on Rosetta versus the 2000-2500 that were on 3 weeks ago.
This will greatly effect the RAC and was expected.
Jose my friend, lets leave these gentlemen to their discussions and get on with our lives.
Thank you.
Movieman

My dear friend: I was pointing the obvious to some people. My fault!!! LOL LOL LOL

As to lives: I almost died yesterday: major electric storm poor sloth came close to being fried ( TY TY APC) Some of the wiring in my house have to be repaired and rechecked. Diablo Legion cannot be hooked until that check is done.

Why not starting in developing a BOINC client that allows for the efficient use of all the CPUS available, specially the new powerful ones ( this will attract the increasingly large number of people that is investing in new processors and hardware) as well as developing one doesn't punish the users of non-windows OS ( Linux user still suffer under crediting )?

In Rosetta, Linux guests would suffer because the BOINC benchmark runs a lot slower on Linux than the same benchmark in BOINC on Windows. With the new credit system, it's not so. So this IS a BOINC problem - but not one of using the machine efficiently as such, but rather a problem of compilers doing more or less good work out of some particular set of code - many compilers are SPECIFICALLY tweaked to do well on Dhrystone/Whetstone, which is what BOINC uses for the benchmark. Unfortunately, it's not the BOINC source-code that is causing this problem, but rather the fact that the BOINC project uses different compilers/compiler settings to compile the Windows vs. Linux vs. other OS's BOINC code. Using the SAME compiler with the SAME settings would make it more fair...

Just as a comparison: A 2.6GHz Opteron running Linux gets 2200/1360 Integer/Float OPS, vs. a 2.2GHz Athlon64 running Windows gets 4150/2250 - that's the same processor architecture, just using different builds of the code. And since the benchmarks are (for most BOINC projects) used to calculate how much work per hour the machine can do, it will give a better result on Windows.

Since the actual project code is often more complex, the actual result of the calculation project may be less affected by compiler differences [Whetstone and Dhrystone are code that can benefit greatly from small improvements in the compiler, whilst big complex functions like those in for exampel Rosetta are much less likely to gain much from a better compiler architecture].

--
Mats

---

Why not starting in developing a BOINC client that allows for the efficient use of all the CPUS available, specially the new powerful ones ( this will attract the increasingly large number of people that is investing in new processors and hardware) as well as developing one doesn't punish the users of non-windows OS ( Linux user still suffer under crediting )?

In Rosetta, Linux guests would suffer because the BOINC benchmark runs a lot slower on Linux than the same benchmark in BOINC on Windows. With the new credit system, it's not so. So this IS a BOINC problem - but not one of using the machine efficiently as such, but rather a problem of compilers doing more or less good work out of some particular set of code - many compilers are SPECIFICALLY tweaked to do well on Dhrystone/Whetstone, which is what BOINC uses for the benchmark. Unfortunately, it's not the BOINC source-code that is causing this problem, but rather the fact that the BOINC project uses different compilers/compiler settings to compile the Windows vs. Linux vs. other OS's BOINC code. Using the SAME compiler with the SAME settings would make it more fair...

Just as a comparison: A 2.6GHz Opteron running Linux gets 2200/1360 Integer/Float OPS, vs. a 2.2GHz Athlon64 running Windows gets 4150/2250 - that's the same processor architecture, just using different builds of the code. And since the benchmarks are (for most BOINC projects) used to calculate how much work per hour the machine can do, it will give a better result on Windows.

Since the actual project code is often more complex, the actual result of the calculation project may be less affected by compiler differences [Whetstone and Dhrystone are code that can benefit greatly from small improvements in the compiler, whilst big complex functions like those in for exampel Rosetta are much less likely to gain much from a better compiler architecture].

--
Mats

So are you arguing for optimizers?