Nereid,

You can take what azazul says to the bank.

The BOINC wiki has a complete description, including the formula used for RAC, here.

BOINC does try to make credit equal across projects, but there are differences, and endless discussion on the boards over whether one project or other is "fairer." Different platforms or even optimized apps give differing amounts of credit too, IIRC.

The biggest difference is with CPDN, which doesn't go by computer time at all, but strictly by model time. Each type of model gives a fixed amount of credit per timestep, and thus a fixed amount per completed model. Each trickle adds credit; there is no verification of results, etc.

Generally, I've found that I get a quite a bit more with CPDN than for comparable time on, say, Einstein@Home.

I'm sure BOINC has approved this practice, as a reward for taking on the long times needed to do one of these massive models.

I'm sure you and (maybe) me and the largest number of BOINC users choose our project priorities by the science we care about. But it would possible to choose a mix that optimizes credit instead, though this would involve a lot of research and time. For example, you can't just compare credit granted for one WU of each of 2 projects that have a quorum system -- you'd have to do long term averages for each machine.

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

I meant to include a link to that and forgot, thanks.

I did not know this. I knew they awarded credit without checking against any other data, but thought they also awarded credit according to computer time. It is probably more fair the way they do it, I have seen some slower computers claim large amounts of credit in other projects, but were brought down by the quorum. If these computers were used for CPDN, then they would likely overcredit themselves.

__________________
www.csphysmath.com
BAUT Team Stats

Perhaps others have seen this before, but with Einstein@home, I've recently had some 6 WU where the other two folks in the quorum have received credit, but I have not. Take this one for example. My result must be OK since it was used to validate the other two.

__________________
Old laser physicists never die, they just become incoherent.
These days, every Tom, Dick, and Harry thinks he knows what a photon is, but he is wrong. - Albert Einstein

Hi LJ,

That is curious indeed.

The detail from your computation, clicking on the "16189328" on the page you mention gives this page, which claims the result is invalid. The stderr out message is the same as for valid ones; and AFAIK, is pretty normal.
So it's possible that the core client is reporting "valid" to the Data Server and "invalid" to the Scheduling Server, or vice versa; but this is a wild guess. Does this happen on all WUs on the box in question? Are all (or only) "r1_1263.0" WUs affected?

My memory is that one or two of the earlier 5.2.x BOINC versions had problems, and the recommended version was quickly (and quietly) bumped up to 5.2.13. Maybe your 5.2.6 is one of those?

So you might try 5.2.13 and see if it helps. (I'm assuming Windows.)

But even if that works, if you don't mind tracking down the numbers of the other WUs and posting them back here, I'll ask on the E@H board, so you can get credit for them.

Three valid results for credit is a long standing thing in E@H and I'm sure they'll want to know about this and fix it.

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

Hi all,
I rejoined the BOINC team.

Alas i couldn't revice my old account c00lk1d,
This is my new account:
39 168148 Jorge Schrauwen

The PC seems to be doing better than the old one i tried (BIONC kept crashing).

Is ther a way to remove the old one?

__________________
GCS/S d(+) s+:+ a--- C++(+++)>$ W+++>$ L>+ M+>++ w++ P+>++ tv@ PS b+ DI+ G e-> h! r-- !z+

~Jorge Schrauwen

Jorge,

Welcome back!

I hope the new machine works well for you in every way.

I don't think there is any way to remove an account, or merge two different ones. If you want the old name to be dropped from the stats pages, if you can somehow log in to your old account, you could quit the BAUT team. To do that you would need your old account key, which, I'm guessing, may be why you couldn't access the old one in the first place?

If you still have the email address you used for the old account, it might be possible to have E@H send you the account key, but I'd have to ask on the E@H board if there is a form for this somewhere.

If the above doesn't work, azazul, I believe, has the power to delete inactive members from the BAUT team, so you might PM him about this option if he doesn't post here. This way the account would still exist, but only near the very bottom of the projects' individual stats.

Otherwise, you will have the pleasure of watching your new self pass your old self in the standings, a feeling few of us get in life...

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

As noted on the main E@H page:

For anyone who hasn't come onboard yet: if you're willing to donate some CPU cycles, it's never too late to join us. You know you want to.

Oh, I do indeed have that power. Jorge, if you are unable to login, then I would be happy to remove your old account from the team.

Edit: And welcome back.

__________________
www.csphysmath.com
BAUT Team Stats

Afther digging trough some backups i found the old mail (took me 8cd's before i got the right one)... account removed from BAUT

I allready surpast my old account,
Old had around 90 credits (boohoo, pass that allready in one night )

See above maybe i can add a 2nd machine soon... not sure yet.
But can link it to my BAUT account name as with the old one?

Also: has anyone noticed that it works better on windows than linux? (better as in... more work done in the same ammount of time)

__________________
GCS/S d(+) s+:+ a--- C++(+++)>$ W+++>$ L>+ M+>++ w++ P+>++ tv@ PS b+ DI+ G e-> h! r-- !z+

~Jorge Schrauwen

Jorge,

Glad this worked!

Yes you can add additional machines to the same account. When you install BOINC on the new machine and it asks which project you want to attach to, check the option to use your existing account.

You will need your new password to log on, which we all hope you will not have to search 8 CDs for!! (But your experience shows the value of having the backups in the first place!)

This thread says......between Windows and linux. With the new "Albert" E@H application, different batches of WUs take different times. So your linux machine may have gotten a longer-running master file than your Windows box did. (See discussion in page 2 of this thread.)

If that is the case, you will have to wait for the 20 or so WU in each master file are done to really tell which is faster. But everyone in your quorum will have the same file, so the credit you get on each machine should be roughly proportional to your actual time spent.

Hope that makes sense.

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

I've seen it happen now with 2 different boxes running two different BOINC versions (5.2.6 and 4.?). Most of the problems are on my Athlon 4000+ (5.2.6). It has trouble with r1_1263.0, but my home computer has problems with a different one (Example). However, not every r1_1263.0 that my 4000+ works on becomes invalid (Example).

I've just done so. Let's hope it does.
Sure, here are the WU ID #'s that have had problems:

4082650
4077347
4072750
4046112
4042265
4037389
4174333
4233933
4213542
4211823

I'll let you know if any more pop up.

__________________
Old laser physicists never die, they just become incoherent.
These days, every Tom, Dick, and Harry thinks he knows what a photon is, but he is wrong. - Albert Einstein

Hi again LJ,

I just posted about this here.

I did notice you upgraded to 5.2.13 & at least one WU had the same problem, so my proposed solution was crap.

I apologize to you and your computers for suggesting over there that your results might be invalid, but I thought it more diplomatic to leave open that possibility there.

My experience of problem machines (limited, I admit) is that they either return all bad WUs or none at all, unless there's a real unusual circumstance. In particular, all the WUs in question ran for the full 11-12K seconds, the same as the many valid ones. So I think the most likely explanation is some mess up on the server-side.

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

Laser Jock,

The replies to my post show that I did not understand the quorum process:
So the validator has indeed decided that your result is the odd one out and accepted the other two as valid; and doesn't feel a need to send the result to a fourth machine.

So now we have the much tougher problem: why are your machines producing a few invalid results among many correct ones? Since this is an example par excellence of an intermittent error, it will probably be hard to track down.

The basic questions people usually ask are: Is either machine overclocked? Can you check processor temp while the E@H app is running?

It is still strange that two of your machines have this problem...I'm at a loss there. Unless both are in an overhot enclosure?

I do remember an early obscure error where WUs finished successfully but ended up invalid which might have been caused by overheating in the FPUs. E@H is very FPU-intensive and heat can apparently make the calculations go off some, just enough to produce invalid results in E@H. This even though the mobo reports overall processor temps as OK.

Since AMDs have much improved FPUs compared to Intel, one would guess they might be extra-sensitive to heat?

Maybe someone with AMD experience can help, as I'm really just guessing now.

__________________
Ken

Visit the BAUT Team Page
BOINCview: Network control
and a superior UI!

It sounds reasonable to me. Laser Jock, I would make sure that it is not in an area that allows it to overheat. Make sure it has enough fans, that the CPU fan is operating properly, and that it is not kept in a small enclosure. Many of my computers do not even have the side covers on them, this reduces the temperature inside the computer, but of course allows more dust to get inside the computer.

__________________
www.csphysmath.com
BAUT Team Stats

Both machines are overclocked. However, I've run Prime95 for hours on both machines and I'm convinced that they are rock-solid stable. In addition, I have good cooling on the machines and they rarely get above 40 C.

Nevertheless, I think I've been able to link the bad WU's to an online game crash (this one so you know to avoid it). The game would frequently crash when I played it, but never thought it was a problem. I now believe that that game and BOINC don't get along.

__________________
Old laser physicists never die, they just become incoherent.
These days, every Tom, Dick, and Harry thinks he knows what a photon is, but he is wrong. - Albert Einstein

I hope that is indeed the problem. It would just be a minor inconvenience compared to other things it could have been.

__________________
www.csphysmath.com
BAUT Team Stats

OK, now I'm sure that was the problem. And it's not just that game either; it can happen on any Flash game. Some are just less stable than others.

__________________
Old laser physicists never die, they just become incoherent.
These days, every Tom, Dick, and Harry thinks he knows what a photon is, but he is wrong. - Albert Einstein