> I think the problem now is that BOINC doesn't just sum up the user credits to
> get the total team credits.

The non-boinc way was due to a fix of another problem (http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=2619), so I guess the non-boinc way (sum of members credits) will return if not currently used ;)


Click below<br><a href="http://www.esea.dk/esea/boinc.asp"><img src="http://www.esea.dk/sig/eseasig.php?memberid=1"></a>

thx for the information :-)

So Raven's team should get his credits after the next update.

&gt; thx for the information :-)
&gt;
&gt; So Raven's team should get his credits after the next update.

Hope so! In my case, restoring my team's score would be easy: I've been the only member so far, so, my score is also my team's. Other peopls might be more difficult, granted, but, I'm still hopeful it can be worked out.

<IMG SRC="http://boinc.mundayweb.com/one/stats.php?userID=376&amp;prj=4&amp;trans=off"><BR>
<A HREF="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/team_display.php?teamid=1886"><IMG SRC="http://www.fandom.ca/bucket/bcboincrib.png"></A>

as far as I know credits (and recent avg credit) are now being calculated the BOINC way, even though I'm not using the BOINC system (since we're based on trickles etc). BOINC has credits in the result table, and the result table also has the userid, hostid, and teamid, so I do calcs off of that. So the result is update with the teamid that userid has at the time he/she gets the result, so it should be appropriate.

here is an example of the query I use to copy to a temp table twice a day to rerun the credit calcs:

create table cpdnexpt.recalcresult (select t.resultid,
exp(-(abs(unix_timestamp()-max(t.trickledate))*0.69314718/604800.0)) weight,
max((t.phase-1)*m.timestep + t.timestep) * m.credit_per_timestep total_credit
from cpdnexpt.trickle t, cpdnexpt.model m, cpdnboinc.result r
where t.resultid=r.id and r.appid=m.modelid
group by t.resultid)

then I make a temp "stats" table to quickly copy over the userid, hostid, teamid for a result and join with the above (by now indexes) recalcresult table:

insert into cpdnexpt.stats
select r.userid,r.hostid,r.teamid,id,c.total_credit,c.weight
from cpdnboinc.result r, cpdnexpt.recalcresult c where r.id=c.resultid

then I update result, user, host, team tables with the above information (there's a lot of locking logic I use that I haven't shown, since mysql isn't as nice about running these huge queries as Oracle is). here's an example of the user table update, the host &amp; team tables are similar just replace teamid &amp; hostid with userid etc:

update cpdnboinc.user u set total_credit=
(select sum(total_credit)
from cpdnexpt.stats r where r.userid=u.id),
expavg_credit=(select sum(total_credit*weight)
from cpdnexpt.stats rs where rs.userid=u.id),
expavg_time=unix_timestamp()

What does this mean for credits for someone who did not join a team for some time then joined a team?

Is it the case that previously these credits were counted in latest team stats while now they are not allocated to any team? Hence team totals have fallen.

Some large team totals are barely more than half what they previously were. I would find it suprising if this accounted for all of the fall in team credits (but for the moment I cannot rule it out).
_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Main_Page">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

Given
1) the substantial reductions in team credits (which could be annoying to some people),
2) That people may have based decisions to change teams on the way CP was doing team credits, and
3) I think it would be less load on the server.

Is it reasonable to calculate credit the old way at some time (perhaps last XML production). Store the results and the result reference of the last result included. Calculate new credit from result table as described above, but only copy results after the result reference stored, to the temp table. Add the stored totals. Save these totals together with result reference of the last result included again. This way you don't have many records to copy and add up.

(I could of course have completely mis-understood what you are doing.)

Edit for this to work properly you would have to work off trickles not results.

_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Main_Page">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

Thank You Carl for the explanation. I do see my individual cobbles growing each day, however, MadPenguin's team stats don't grow at the same rate and are somewhat less than my individual numbers. Looking at the history graphs <a href="http://www.boincstats.com/stats/team_graph.php?pr=cp&amp;id=67">here</a> (and following the link to my individual numbers) you can see what I mean.

I guess I don't understand how they can be different.
<br><img src="http://boinc.mundayweb.com/cpdn/stats.php/userID:22/trans:off/.png">

Thank you Deprecated,

now I have a second (and third) theory.

First theory:
You have 241,681 credits team has 208,694

Is it possible you gained ~32,987 credits before joining MadPenguin?


The second theory is that your team credits shows a jump of 6805 multiples. This imples (from recently) team does not get credit until you complete a result. So is 32,987 the credits in models in progress?
(Edit this would explain why all teams in Sulphur beta have no credit (No-one completed a run yet).)

The third theory is that it is a combination of these two issues.
_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Main_Page">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

Sorry Carl &amp; everyone time for a third theory (or up to 7th if counting combinations).

It appears you are counting failed results as well as full runs. The question is whether there are lots of credits in results that were abandoned or otherwise crashed but did not report back. Do these credits get counted for teams?

Looking at web pages a lot of hosts have lots of runs marked as in progress but clearly they are abandoned/have not reported failure.
_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Main_Page">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

Hi,

the last trickles increase credits for me (Nasicus) but not for my team (Crunchers@Freiburg).
Total credits of the team is staying at the same number and is still wrong.
And today I found a trickle listed twice in my latest trickles (result ID 966289, result name 0zb4_100066241_0, phase 1, timestep 216040).

What wrong?

Regards
Mike

Nasicus,

If I have understood, team credit will only increase when a result is finished.

Not sure whether Carl is telling us it is "working as designed" and thats the way it is going to be. OTOH he might listen to people complaining that they have 100 computers all over half way through a long sulphur cycle experiment and they still haven't got any team credit. ;) (Now who wants to sign up to BOINCSynergy in the Beta test so that we can get that situation and force the change ;) )


Haven't seen a trickle listed twice before, maybe your model rewound a day just at the trickle boundary. (I have a feeling Carl has said this can happen without a rewind but I can't remember why.)




_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Main_Page">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

Crandles,

All my numbers crunched have been crunched for MadPenguin.org.

For several runs, we linux users could not complete runs due to changes in the client that linux didn't like.

And finally, with the other outstanding issues we already know about is the "I sent in all my trickles and it still doesn't show complete" error.

I do have a few models in each of these last two catagories. Did I miss where Carl said that teams only get credit for completed runs only?
<br><img src="http://boinc.mundayweb.com/cpdn/stats.php/userID:22/trans:off/.png">

Carl wrote (see below):
"BOINC has credits in the result table, and the result table also has the userid, hostid, and teamid, so I do calcs off of that"

I didn't realise the implications of that until I saw your team had no credits for several days and a 6805 credits on one day.

Reported results should be OK, abandoned is much less clear...

Despite my 3 (7) theories, I still can't reconcile it. 32 models *6805.26 = 217,768 and you only have 208,694. (Unless you joined MadPenguin after completing first 2 models but you seem to have denied that. There wasn't anything like a team name change was there?)
_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Main_Page">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

the problem seems to be that I am relying on the teamid in the result table, but somewhere along the line BOINC hasn't been putting teams in there. since that isn't reliable, I'm going to just go back to the "old system" where your team in the user table determines how many credits the team gets (which means that a person "takes" their credits with them if they join another team). also don't forget that credit calcs are run twice a day so it may be up to 12 hours until you switch/join a team that you'll see the stats change.

on the RAC front -- my idea today is to use the average time of a result (i.e. the average date of all trickles for a given result) as the "exponential decay" for a result * the credits total for that result, so that should give more sensible BOINC-like numbers (instead of the last_trickle date * credit which is making huge #'s for RAC).

Great thanks Carl. :D

For the RAC, why not reset them all to 0 (or 10% of current value) and let the system start working again. AFAIK The silly numbers are in response to large chunks of credit in a short time not because the system was wrong. (I also fear for today's idea because of people with several abandoned work units that show as in progress.)


_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Main_Page">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

well the BOINC RAC system never really worked/applied to CPDN, since we use trickles, so there's no real point in trying to use my old kludges for that (and my new kludges are more fun ;-).

Gee, *THANKS* Carl ... you just broke the Wiki ... :)

Ok, so, is the team and take it with you been implemented already?
<p>
<a href="http://boinc-doc.net/boinc-wiki/index.php"><b>BOINC-Wiki</b></a>
<img src="http://www.boincstats.com/stats/banner.php?cpid=a6477942e70ed39f669d1ff2ede05be8">

Don't know if team has been done. It may have been fixed but has yet to hit the stats.

The RAC seems to have hit the Your account but not the top participants/xml yet.
_______________________________
Visit <a href="http://boinc-doc.net/boinc-wiki/index.php?title=Main_Page">BOINC WIKI</a> for help

And join <a href="http://www.boincsynergy.com/">BOINC Synergy</a> for all the news in one place.

Well, I made a change anyway ... probably a couple more places need to be changed but I can't recall where they might be ...

Make it a "hunt" for the Wiki readers ... :)
<p>
<a href="http://boinc-doc.net/boinc-wiki/index.php"><b>BOINC-Wiki</b></a>
<img src="http://www.boincstats.com/stats/banner.php?cpid=a6477942e70ed39f669d1ff2ede05be8">

&gt; the problem seems to be that I am relying on the teamid in the result table,
&gt; but somewhere along the line BOINC hasn't been putting teams in there. since
&gt; that isn't reliable, I'm going to just go back to the "old system" where your
&gt; team in the user table determines how many credits the team gets (which means
&gt; that a person "takes" their credits with them if they join another team).
&gt; also don't forget that credit calcs are run twice a day so it may be up to 12
&gt; hours until you switch/join a team that you'll see the stats change.

So in other words it'll go back to being the sum of current members' credits. Well, it's better than what I'm experiencing at the moment which is no new credit for the team, but a decaying RAC (well, the RAC thing is to be expected as it was way too high for normal curcomstances ;))

Anyway, happy to know something is being done, and cudos for great and hard work


Click below<br><a href="http://www.esea.dk/esea/boinc.asp"><img src="http://www.esea.dk/sig/eseasig.php?memberid=1"></a>