Hmmm...

As every day, I connected my (otherwise Offline) Network to the Internet to allow flushing Results this morning, then left it offline again (Network Access Disabled via BOINCview).

This evening I connected everything again, and noted an abnormally high amount Credits being added to my Account.

Timeframe :
Morning Flush : 0520 Local time
Credits (acc. BOINCstats.com) : 28128

Evening Flush : 1720 Local time
Credits (live observed) : 32835

Time elapsed : 12h
Credits added : 4707
(Note : all CPDN Systems share 50/50 with SETI)
Thus, effective CPU time for CPDN : 6h per Host

Active Systems : 13
Actual Trickles delivered by them between 0520 and 1720 : 10

I manually hand-counted those, per each Host and verified by Timestep .

Specials/Abnormals I noted :
- One of my Systems (Host 5592) delivered its first finished CPDN WorkUnit in the Morning Flush.

However, due to reported past abnormal Credit behaviour with CPDN experienced by other CPDN Users delivering a finished WorkUnit, I closely monitored my Credits but found only the one last Trickle had been accounted for... so nothing unusual noted.

- After Evening Flush, that Host (5592) had finished its first Trickle for the new Model.
Credits Granted : 94.52

Wasn't that supposed to be a fixed number of, like... ~70 Credits in the past?
---------------
Anyway, even if I assume 95 Credits were now to be granted, and even if I doubled the amount of Trickles to 20 (assuming Stats were fooling me, and accounting for some of the last day's Trickles), I only come up with about ~1900 Credits to be expected.
(and that using a very optimistic scenario)

My Systems simply cannot deliver ~4700 Credits worth of work in 12 hours, while each one is sharing 50% CPU time with SETI (with worked, the normal number of SETI WU's were finished in the evening).
Even my fastest System took 6h53m for its latest Trickle, thus at 50:50 Share with SETI, effectively takes almost 14 hours realtime for one Trickle.
--------

Is there anyone that could bring some light into this ?
The observed case also does not quite fit the known "Credit Bug when delivering a finished CPDN Model", in which case I would have expected the 5444.21 Credits surplus the completed WorkUnit #20532 was worth.

Very odd, and I don't like being Credited more than I deserve.
Either the Stats were fooling me big time (I'll have to monitor the daily stats for a 'peak' with 3rd Party Stats Services), or there's indeed something wrong with the Crediting System.

==============
Since that is a 'live' Case of possible Credit System error, I'd be glad if an Admin could isolate the Data while it's fresh & available, and check what possibly went wrong.
___________________________________________
<p>Scientific Network : <a href="http://www.falconfly.de/network.htm">36200 MHz «» 8460 MB «» 854.3 GB</a> </p>

Hmmm...me too.

I turned in one trickle this afternoon and my credits jumped by 400 or so. I think I have the answer though. It looks like, in addition to getting a new higher credit for each trickle (Carl had mentioned this possible adjustment in another thread soemtime ago), that all previous trickles have been adjusted upward to match the new credit. That is, all those previously completed trickles that got ~76 credits have been adjusted to the new credit level of ~94 credits. At least mine seem to add up that way.

&gt; Hmmm...me too.
&gt;
&gt; I turned in one trickle this afternoon and my credits jumped by 400 or so. I
&gt; think I have the answer though. It looks like, in addition to getting a new
&gt; higher credit for each trickle (Carl had mentioned this possible adjustment in
&gt; another thread soemtime ago), that all previous trickles have been adjusted
&gt; upward to match the new credit. That is, all those previously completed
&gt; trickles that got ~76 credits have been adjusted to the new credit level of
&gt; ~94 credits. At least mine seem to add up that way.

Hm, I never read that, but at least it explains the new, higher Value per Trickle.

But if I do the math, recalculating my old value of 28128 Credits from the Morning Flush, I would have already been at ~34790 Credits before flushing in the evening which doesn't quite add up.

~5444 Credits for the finished WorkUnit this Morning also seems like a normal value for a finished Model, using the old Credits per Trickle figures. (?)
___________________________________________
<p>Scientific Network : <a href="http://www.falconfly.de/network.htm">36200 MHz «» 8460 MB «» 854.3 GB</a> </p>

There's a <a href="http://www.climateprediction.net/board/viewtopic.php?t=2596">thread</a> on the phpBB about the extra credits lottery that might interest you. Carl is hoping to track down what's going on before he leaves in a couple of weeks, but there doesn't appear to be a common factor so far.
<br><a href="http://www.teampicard.net"><img src="http://www.teampicard.net/templates/fisubice/images/phpbb2_logo.jpg"></a><a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/team_display.php?teamid=3">Join us here</a>

I think I should give a bit more detail on what I observed with my inflated credit...

First, the box is a P4HT 2.8 WinXP SP1, but running only one CPDN model (part time mostly since I particpate in 5 projects with that machine). I was watching pretty closely to see the 19th trickle come in. Before it came in, credit was as expected at about 76 per trickle. With the 19th trickle, credit increased by a large jump. FalconFly's note made me think to check the calculations, and his estimate of about 94 credit per trickle is dead on my total for 19 trickles on that box now. (I also noticed that the only other person in my team had a credit jump at about the same time...I think more than coincidentally).

Hopefully I will get another trickle from this box this afternoon or tomorrow (it is only on during business hours when I am in the office). I'm curious to see what credit the next one gets???

I have my personal CPDN Credit Bug quite visible now :

<a href="http://www.boincstats.com/stats/user_graph.php?pr=cp&amp;id=3434">http://www.boincstats.com/stats/user_graph.php?pr=cp&amp;id=3434</a>

<img src="http://www.boincstats.com/stats/user_graph2.php?pr=cp&amp;id=3434">
(11 Nov 2004)

I did not see even the most active Members of my Team (having alot more Credits than me) reflect a possible, general "uprating" of previous Credit calculations themself, their numbers look normal, no Peaks or Jumps observed.

<a href="http://www.boincstats.com/stats/user_stats2.php?pr=cp&amp;st=0&amp;to=100&amp;or=&amp;ti=18">http://www.boincstats.com/stats/user_stats2.php?pr=cp&amp;st=0&amp;to=100&amp;or=&amp;ti=18</a>

------------
Looking at some other of my Systems that recently begun Trickling, I see an average Credit of ~94.5 per Trickle as well, so that definitely changed.

I'll take some averages and calculate the Credit per Trickle for some of my longer running ones. IF those turn out as ~94.x as well per Trickle, that would indicate an "uprating" of past/old Trickles indeed.
___________________________________________
<p>Scientific Network : <a href="http://www.falconfly.de/network.htm">36200 MHz «» 8460 MB «» 854.3 GB</a> </p>

Hmm...

I did check.
And indeed!

Average even for Machines having delivered alot of Trickles since September (where it was definitely in the 70's Credits per Trickle Range ) has been 'uprated' to ~94.5 ...

Looks like that was responsible for my Jump.
Odd though, that multiplying my existing (old) Credits by the factor of 1.26 (94.5 vs. 75) would yield 35441 Credits at my Morning Flush.

Since I haven't reached that Level yet, even after my evening Flush, this is either an ongoing process (progressively re-processing the entire Results base, and with that, a progressively uprated Credit), or it did not effect all of my Trickles (?)

-- edit --
Found one not affected (yet) :
My recently finished Model.

That one still figures as 74.578 Credits per Trickle (5444.21Cr from 73 Trickles)
Looks like if that had been reprocessed for 94.5Cr/Trickle as well, my figures would all add up of having been entirely uprated from ~75 to 94.5.

___________________________________________
<p>Scientific Network : <a href="http://www.falconfly.de/network.htm">36200 MHz «» 8460 MB «» 854.3 GB</a> </p>

Something is definitivly afoot.

Credit is taking weird jumps. A host which is only capable of one
trickle a day has been granted 415.88 in the last 24 hours. Another
which averages 1.5 per day has been granted 945.18 in the same time
frame.

Yesterday, I received a 5K+ bonus when a third host finished a WU!
Thought that little problem was fixed.

-P

Concerning the Credits per Trickle, the answer is in the CPDN Classic Forum I've been pointed to by a Team mate :

<a href="http://www.climateprediction.net/board/viewtopic.php?t=2619">http://www.climateprediction.net/board/viewtopic.php?t=2619</a>
___________________________________________
<p>Scientific Network : <a href="http://www.falconfly.de/network.htm">36200 MHz «» 8460 MB «» 854.3 GB</a> </p>

I agree that they increased from the seventy something to the ninety something figure per trickles, but the particularity that I have noticed now is that each time that a change of step begins, then there is a total higher bonus in credits. It happened in my unit when finishing fase 1/3 and then again at the end of fase 2/3, will it happen again when it finishes 3/3??

PFFFFT!!!!
Looks like I got robbed for 100,000 points... I'd really like to know what the heck is "up"

&gt; PFFFFT!!!!
&gt; Looks like I got robbed for 100,000 points... I'd really like to know what the
&gt; heck is "up"
&gt;
From the home page " Also note, credits have been recalculated to be more fair for users, since a bug was causing "unauthorized bonus credits" to appear on random runs. Team/Host/User credits are now recalculated every few hours (similar to how other BOINC projects "validate" their credits)."

They took back 11,000 credits from me as well. Oh well we now get 95 credits per trickle now so it will only take me 115.78947 trickles to get my 11,000 credits back.
<img src="http://boinc.mundayweb.com/one/stats.php?userID=70">