Are you serious? Credits...

Message boards : Number crunching : Are you serious? Credits...
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
[SG]sl1d3r

Send message
Joined: 30 Aug 11
Posts: 12
Credit: 616,565
RAC: 0
Message 11064 - Posted: 23 Oct 2011, 20:28:36 UTC

http://burp.renderfarming.net/workunit.php?wuid=1662288

More than 450.000 CPU seconds worked and only 203.63 credit points???

That's not funny... :(

sl1d3r
ID: 11064 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4568
Credit: 2,100,409
RAC: 55
Message 11065 - Posted: 24 Oct 2011, 6:37:55 UTC - in response to Message 11064.  

It is weird that the number 203.63 should pop up for all the workunits for this session so far. I think we may have hit a credit limit somewhere inside BOINC.
If that is indeed the case then the credit will be adjusted up to the real value later.

I'll look into this.
ID: 11065 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[SG]sl1d3r

Send message
Joined: 30 Aug 11
Posts: 12
Credit: 616,565
RAC: 0
Message 11066 - Posted: 24 Oct 2011, 19:30:45 UTC - in response to Message 11065.  

That would be nice. Thank you!
ID: 11066 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile nick
Avatar

Send message
Joined: 27 Jul 09
Posts: 28
Credit: 1,546,451
RAC: 0
Message 11074 - Posted: 3 Nov 2011, 10:10:41 UTC

My credits have been flat since 24 Oct.
ID: 11074 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
LordOfWar

Send message
Joined: 3 Jun 08
Posts: 9
Credit: 435,513
RAC: 0
Message 11081 - Posted: 5 Nov 2011, 1:49:00 UTC - in response to Message 11074.  

nick,

You task list shows you have been working on session 1279 since the 24th of october and that is what this session [SG]sl1d3r is referring to in his first post.

As Janus stated we suspect that this session might have reached a credit limit in burp and it will be fixed in the future. You will be awarded the correct amount of credits when the problem is fixed, and the actual credits you should have received is calculated.

Thanks and happy rendering!.
ID: 11081 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[SG]sl1d3r

Send message
Joined: 30 Aug 11
Posts: 12
Credit: 616,565
RAC: 0
Message 11083 - Posted: 5 Nov 2011, 22:53:47 UTC - in response to Message 11081.  

Some news/update info about that problem?

Greets
sl1d3r
ID: 11083 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4568
Credit: 2,100,409
RAC: 55
Message 11084 - Posted: 7 Nov 2011, 19:25:55 UTC - in response to Message 11083.  

Yup, credit will be granted once the session has completed.
ID: 11084 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Speedy

Send message
Joined: 25 May 06
Posts: 208
Credit: 676,104
RAC: 0
Message 11118 - Posted: 21 Nov 2011, 23:38:33 UTC

Looks like the credit issue has been sorted on session 1279. I got 5,172.00 for result id 6954534 it works out to be 309.33 credits an hour. Nice work.
Have a crunching good day
ID: 11118 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4568
Credit: 2,100,409
RAC: 55
Message 11120 - Posted: 22 Nov 2011, 18:10:25 UTC - in response to Message 11118.  

Looks like the credit issue has been sorted on session 1279. I got 5,172.00 for result id 6954534 it works out to be 309.33 credits an hour. Nice work.

Unfortunately the issue hasn't entirely been solved. What is even worse is that it seems to be rooted quite deeply in the way that BOINC handles workload estimation right now.
ID: 11120 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile DoctorNow
Project donor
Avatar

Send message
Joined: 11 Apr 05
Posts: 403
Credit: 2,189,214
RAC: 116
Message 11248 - Posted: 16 Mar 2012, 15:43:20 UTC
Last modified: 16 Mar 2012, 15:56:16 UTC

The thread might be old but the title fits... ;-)

Heck, CreditNew is going on my nerves, the values are going down and down on some sessions:
7031880
7031912
Looks like an older 2-core-system with 64-bit doesn't earn much here anymore...
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 11248 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Loup

Send message
Joined: 18 Feb 12
Posts: 18
Credit: 3,051,672
RAC: 0
Message 11249 - Posted: 16 Mar 2012, 18:08:05 UTC - in response to Message 11248.  

Yeah, the credit system is pretty much broken, just checking the points per WU for the session 1403 shows that (anything from 1000 - 5000 per WU for the ones I looked at).

I am wondering if the fixed GFlops per WU setting is screwing something up, it certainly causes the task estimates to be pretty much pointless:

If a PC churned through a few long WUs the system thinks that that PC is actually slow as hell since it suddenly took ages for a WU of the same complexity aka GFlops. Afterwards it gives you insane task durations for small WUs. Do a lot of small WUs and the opposite happens - if it judges the credits using the same mechanism then I am not surprised that those are pretty random.
ID: 11249 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4568
Credit: 2,100,409
RAC: 55
Message 11251 - Posted: 16 Mar 2012, 19:33:07 UTC

This concern was raised back in 2009 before CreditNew was implemented:

David Anderson wrote:
> The design of a new credit system is here:
> http://boinc.berkeley.edu/trac/wiki/CreditNew
>
> This design is joint work of me and Kevin Reed from WCG.
> It replaces an earlier proposal, now deprecated.
> I'm in the process of prototyping it to make sure it works as planned.
> Comments welcome.
>
> -- David

Janus:
It is a little unclear to me how the normalization and averaging is done in the case of a project with workunits of unknown lengths (i.e. different for each workunit from a given app A, no estimates known in advance). Doesn't that mess up the convergence of VNPFC*(A) and VNPFC*(H, A)?

David:
Such variance makes these converge more slowly,
but they still (according to the Law of Large Numbers) converge.

I'm personally still worried that convergence will never happen due to time-local interference (as correctly pointed out by Loup). This issue has been raised numerous times. Recently (in 2011) the reply was:

David:
The goals of CreditNew involve long-term averages.
It makes no promises about individual jobs or about credit/hour.

If a project has highly variable jobs,
this translates into highly variable credit for individual jobs.
But the long-term average stuff should still hold.

If anyone has a specific suggestion for how to make credit
less variable on the short term, while still preserving the
long-term goals, let me know.

I've really tried to think about this a lot but haven't been able to come up with a system that fits within the framework of CreditNew and yet will work reliably with "nondeterministic" projects like BURP (where WU run-length is not known in advance).
I'm open to a discussion about what could be done about this, but please stay civil and to the point - credit flamewars are not accepted here, please take those elsewhere. Also please understand that even if some kind of super nice solution is found in this thread it may still take quite a while to implement it, test it and convince the powers that be about it.
ID: 11251 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Tomasz R. Gwiazda
Avatar

Send message
Joined: 20 Jul 09
Posts: 6
Credit: 82,543
RAC: 0
Message 11423 - Posted: 18 Jul 2012, 7:41:00 UTC - in response to Message 11251.  

how about that:

Completed and validated 98,542.63 550,783.32 3,839.39
Completed and validated 98,279.40 550,674.36 203.63


why there is a difference in points?? 3839 to 203

Join us at www.boincatpoland.org
ID: 11423 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Tomasz R. Gwiazda
Avatar

Send message
Joined: 20 Jul 09
Posts: 6
Credit: 82,543
RAC: 0
Message 11456 - Posted: 23 Aug 2012, 14:52:25 UTC - in response to Message 11423.  

and again

94,446.46 595,741.10 203.63
94,772.84 598,240.25 203.63


is anyone here alive from the project staff?


Join us at www.boincatpoland.org
ID: 11456 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile DoctorNow
Project donor
Avatar

Send message
Joined: 11 Apr 05
Posts: 403
Credit: 2,189,214
RAC: 116
Message 11457 - Posted: 23 Aug 2012, 17:46:00 UTC

Since your computers are hidden I assume this is from session 1490? I have lots of of them too. ;-)
It was already mentioned in other threads also, but some sessions cause this credit miscalculation.
Don't worry, after SunFlower will be finished Janus will take care of all the wrong given credits...
Life is Science, and Science rules. To the universe and beyond
Proud member of BOINC@Heidelberg
My BOINC-Stats
ID: 11457 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Tomasz R. Gwiazda
Avatar

Send message
Joined: 20 Jul 09
Posts: 6
Credit: 82,543
RAC: 0
Message 11458 - Posted: 24 Aug 2012, 15:50:23 UTC - in response to Message 11457.  

That means exactly what and when ? :)
when this set will end? and what will be done?

Join us at www.boincatpoland.org
ID: 11458 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John P. Myers

Send message
Joined: 10 Aug 11
Posts: 2
Credit: 53,826
RAC: 0
Message 11505 - Posted: 11 Nov 2012, 19:00:15 UTC
Last modified: 11 Nov 2012, 19:00:47 UTC

I'm assuming the credit problem applies to session 1491 and will also be corrected?

http://burp.renderfarming.net/result.php?resultid=7258457
ID: 11505 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Janus
Volunteer moderator
Project administrator
Avatar

Send message
Joined: 16 Jun 04
Posts: 4568
Credit: 2,100,409
RAC: 55
Message 11507 - Posted: 13 Nov 2012, 18:44:25 UTC - in response to Message 11505.  

Yes, pretty much everything since the introduction of "CreditNew" is going to be corrected.
ID: 11507 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
TeeVeeEss

Send message
Joined: 29 Nov 11
Posts: 1
Credit: 137,365
RAC: 0
Message 11524 - Posted: 6 Dec 2012, 5:54:18 UTC

One WU from session 1339 is giving the 'right' credits, the next one the famous 203.63:

ses0000001339frm0000000103prt00001_4 | 1776726 | 54690 | 5 Dec 2012 | 4:56:03 UTC 6 Dec 2012 | 2:15:18 UTC | Voltooid en gecontroleerd | 76,614.19 | 558,776.71 | 203.63 | SunflowerBlender v4.81 (mt)
ses0000001339frm0000000077prt00001_3 | 1776692 | 54690 | 4 Dec 2012 | 7:04:42 UTC 5 Dec 2012 | 4:56:03 UTC | Voltooid en gecontroleerd | 76,273.03 | 557,417.95 | 4,696.26 | SunflowerBlender v4.81 (mt)
ID: 11524 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Bruce Kennedy

Send message
Joined: 5 Sep 09
Posts: 1
Credit: 286,244
RAC: 462
Message 11531 - Posted: 15 Dec 2012, 10:10:57 UTC - in response to Message 11524.  

I have SunflowerBlender v4.81 from 3 weeks ago that needs to be corrected.

7260216 1776625 56589 Completed and validated 49,480.27 359,982.99 5,399.10

7260215 1776629 56589 Completed and validated 49,240.97 359,068.73 203.63


ID: 11531 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : Are you serious? Credits...