I read in one of the faqs that when results are reported, credits are granted based on the computer that claims the smaller amount of credit.

Looking at all this data that's suddenly available i notice that of the three reporting computers, the credits seem to be awarding to the middle claim. Which does seem fair. I'm often granted more credit than I claimed it seems.

Just an observation.

North

I saw the same thing. I also saw that most of my computations are "invalid". Should say 90% of them. That can come from an error "too many rasults...", or no error at all, but the wu is still "invalid". What is amazing is that these WUs are taken into account in the crediting scheme: they are one of the 3 needed results. But the 2 other computers get credits, and mine 0.
Clearly, the credit computation is far from its specifications.
Another thing is the most WUs lead to invalid results. From the forum, it seems that it is mostly a problem of G4/G5 running optimized clients. Is it a poor testing of the optimisation ?
I was planning to move to official non optimized Boinc and seti, but one of my computer is unable, since 2 days, to upload any result (the famous error "no filename...". Thats a terrible beta. Better to move to cpdn.

---

Lately my results have been valid. (I only went to BOINC a few weeks ago)

I was running 10.1.5 on my old beige G3 (with a Sonnet upgrade to G3). ALL of my results from that were invalid. Then, with help from the Mac forum people, I managed to upgrade it to 10.3.5 (Panther). Now all my results are valid.

Am I running the optomized Boinc? I'm not sure, but I don't think so. It's just the regular client downloaded from the Boinc site. With SETI control on top.

I've also noticed some work units where I got zero credit but the other two hosts got 30 or so units. I would think this is just a little bug. It doesn't happen very often on my units.

On the whole, once I got up to speed, and since they've started to get through that validator queue, things seem to be moving along pretty well.


North

Wow, i just looked at your results. You really do have a big problem with all those zero credits!

I'd get rid of that optomized client and see what happens. What's wrong with the plain old non-optomized client anyway? Maybe you're crunching faster, but you're getting no credit for it somehow.

North

I have this problem with those zero credits too. I hope that Berkeley can fix this prob soon.

---

M.f.G., HaQuiX

http://www.unitedmacs.com

CREDITS by PAUL

BOINC VALIDATOR and Credits

---

Semper Eadem
So long Paul, it has been a hell of a ride.

Park your ego's, fire up the computers, Science YES, Credits No.

> Wow, i just looked at your results. You really do have a big problem with all
> those zero credits!
>
> I'd get rid of that optomized client and see what happens. What's wrong with
> the plain old non-optomized client anyway? Maybe you're crunching faster, but
> you're getting no credit for it somehow.
>
> North
>
>
>
>
>
I just loaded the basic "non-optimized" clients (boinc 4.13 and seti4.02). Got an error "O status but no finished file" after 3 minutes...Will leave it running a few days.
I just remeber an error we got recently with the scientific data processing pacakge IDL: when moving large blocks in memory, datas were sometimes randomly modified. That disapeared with 10.3. But clearly, C compilers can easily produce wrong codes which give random errors.20 years, computers gave stable results (either good or bad), but that is no more the case with the terrific tandem Unix+C...

---

> > Wow, i just looked at your results. You really do have a big problem
> with all
> > those zero credits!
> >
> > I'd get rid of that optomized client and see what happens. What's wrong
> with
> > the plain old non-optomized client anyway? Maybe you're crunching
> faster, but
> > you're getting no credit for it somehow.
> >
> > North
> >
> >
> >
> >
> >
> I just loaded the basic "non-optimized" clients (boinc 4.13 and seti4.02). Got
> an error "O status but no finished file" after 3 minutes...Will leave it
> running a few days.
> I just remeber an error we got recently with the scientific data processing
> pacakge IDL: when moving large blocks in memory, datas were sometimes randomly
> modified. That disapeared with 10.3. But clearly, C compilers can easily
> produce wrong codes which give random errors.20 years, computers gave stable
> results (either good or bad), but that is no more the case with the terrific
> tandem Unix+C...
>
>
I was getting client errors with the optimized client. So, I just loaded the basic "non-optimized" clients (boinc 4.13 and seti4.02). I am now getting .11 to .14 claimed credits for all of the new results on the computer that I loaded it on. I have been granted credits in the 40.0 range though even though it has taken over 10000 secs to process a wu. Any suggestions.

Jon