Massive drop of credits per CPU hour

Message boards : Number crunching : Massive drop of credits per CPU hour
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 · Next

AuthorMessage
reklov

Send message
Joined: 21 Aug 11
Posts: 2
Credit: 1,011,939
RAC: 0
Message 97 - Posted: 2 Sep 2011, 6:22:06 UTC

In the last days, I observed a massive drop of credits, e.g.
48817 22978 196 1 Sep 2011 17:07:37 UTC 2 Sep 2011 5:25:24 UTC Fertig und Bestätigt 44,220.27 41,472.14 63.13 Get Decics with Bounded Discriminant v1.04
42843 21368 196 28 Aug 2011 11:59:03 UTC 30 Aug 2011 4:31:50 UTC Fertig und Bestätigt 66,358.80 63,553.96 301.98 Get Decics with Bounded Discriminant v1.04

It's the same host and same application. Credit droped from 17.1cr/h to 5.5cr/h CPU time. (Other projects grant in general about 20-25 cr/h, with some spikes in both directions.)
Is any explanation of the drop of credits from server side available?
(I didn't observe such a drop in my other projects in the last days, thus it must be something specific to NumberFields@home.)
ID: 97 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Dagorath
Avatar

Send message
Joined: 2 Sep 11
Posts: 57
Credit: 1,274,345
RAC: 0
Message 99 - Posted: 2 Sep 2011, 7:24:06 UTC - in response to Message 97.  

If this project uses the CreditNew scheme for calculating credits then you can expect to see a drop in credits after completing a few tasks. After that the credits should level out and not change much.

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux
ID: 99 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1318
Credit: 403,736,438
RAC: 288,164
Message 106 - Posted: 2 Sep 2011, 23:54:11 UTC - in response to Message 99.  

Not sure if we use the "CreditNew" scheme. We are using whatever is in the current server_stable boinc code. Our WUs do have a big swing in run times, from a few minutes to 20 or 30 hours; I'm not sure if that affects the way boinc computes the credit.
ID: 106 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Dagorath
Avatar

Send message
Joined: 2 Sep 11
Posts: 57
Credit: 1,274,345
RAC: 0
Message 109 - Posted: 3 Sep 2011, 8:19:00 UTC - in response to Message 106.  

If you're on the current server_stable code and haven't customized the parts of the code that calculate/award credits then I am pretty sure you're using CreditNew. I don't really follow the server versions as much as I do the client versions so I could be wrong. I crunch for the science and don't give 2 hoots for the credits but you'll find a lot of crunchers hate CreditNew with a passion because it doesn't yield the credits per hour they expect to receive.

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux
ID: 109 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
zombie67 [MM]
Avatar

Send message
Joined: 19 Aug 11
Posts: 31
Credit: 63,078,601
RAC: 27,106
Message 111 - Posted: 4 Sep 2011, 3:41:56 UTC - in response to Message 109.  

I crunch for the science and don't give 2 hoots for the credits but you'll find a lot of crunchers hate CreditNew with a passion because it doesn't yield the credits per hour they expect to receive.


Heh. It's a LOT more complicated than that. Yes, sometimes credits are too low, and sometimes they are too high. The reality is that there is no rhyme or reason. It's completely random. On top of that, it penalizes application optimization, which should be encouraged. It penalizes building faster machines, which should be encouraged. There are a number of other problems with it, when you start to include 32 vs 64 bit, GPUs, etc. But enough said. I am not here to argue it really. I will leave it with this: Yes, a lot of crunchers hate CreditNew, and for valid reasons.
Reno, NV
Team: SETI.USA
ID: 111 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Sabroe_SMC

Send message
Joined: 4 Sep 11
Posts: 1
Credit: 5,832,306
RAC: 9,574
Message 116 - Posted: 4 Sep 2011, 17:50:32 UTC
Last modified: 4 Sep 2011, 17:51:35 UTC

Credits granted are a bad joke: http://stat.la.asu.edu/NumberFields/result.php?resultid=31646
51 cr / 3,6 hours on a 3.5MHz i7 LOL

Set to NNW

Edit: It was the first WU, so no creditdrop....^^
ID: 116 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1318
Credit: 403,736,438
RAC: 288,164
Message 117 - Posted: 4 Sep 2011, 21:39:44 UTC - in response to Message 116.  

I agree, the credit granting scheme does seem screwed up. I took a quick look at some results and noticed that credits per cpu hour had a huge variance. I also noticed that it seems to be biased towards linux. Credit granted to linux hosts were between 2 and 3 times more than windows for the same cpu time. I promise to look into this when I find some time.
--Eric
ID: 117 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Conan
Avatar

Send message
Joined: 3 Sep 11
Posts: 30
Credit: 7,701,817
RAC: 3,703
Message 120 - Posted: 5 Sep 2011, 4:59:16 UTC

Don't know if the bias people have seen towards Linux has been compared like for like (64 bit Linux against 64 bit Windows), as I don't have any 64 bit computers (yet).
However my 32 bit Windows are only giving 12 to 15 cr/h with little variance between work units.
My faster computer is getting less than my slightly slower computer which seems a bit backward to me.
A recently completed 17 hour WU only returned 12.6 cr/h, that is not very good.

Have to go away this week so will leave this project running and will make a decision about the poor credit when I return.

If this project is like Optima@home then I will have to upgrade to 64 bit just to get decent credit (Optima only gives 14 cr/h on 32 bit but 25 plus on 64 bit).

Conan
ID: 120 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
vaughan

Send message
Joined: 20 Aug 11
Posts: 3
Credit: 27,052,804
RAC: 2,907
Message 140 - Posted: 6 Sep 2011, 1:23:07 UTC

Reached the 10k milestone.

Credit on this project is too low.

Set to No New Tasks.

Bye.
ID: 140 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile tocx
Avatar

Send message
Joined: 29 Aug 11
Posts: 4
Credit: 2,904,753
RAC: 0
Message 156 - Posted: 6 Sep 2011, 12:26:08 UTC - in response to Message 117.  

..Credit granted to linux hosts were between 2 and 3 times more than windows for the same cpu time. ...
--Eric

Means this that the 64Bit LinuxApp is faster as 32Bit WindowsApp,or is this a problem with the creditgranting?
ID: 156 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile rebirther
Avatar

Send message
Joined: 19 Aug 11
Posts: 4
Credit: 1,673,636
RAC: 0
Message 157 - Posted: 6 Sep 2011, 14:50:59 UTC

With long tasks on windows I got around 7cr/h ^^
ID: 157 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile tocx
Avatar

Send message
Joined: 29 Aug 11
Posts: 4
Credit: 2,904,753
RAC: 0
Message 160 - Posted: 6 Sep 2011, 16:57:31 UTC - in response to Message 157.  

until yesterday i got up to 17 cr/h (AMD Athlon X4 630, host with most cr/h), with the new LinuxApp (1.06) cr/h increased
ID: 160 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ChertseyAl

Send message
Joined: 19 Aug 11
Posts: 45
Credit: 1,014,069
RAC: 0
Message 164 - Posted: 6 Sep 2011, 18:00:41 UTC

Hmmm, there was a brief spike when the credit became reasonable, but now it's dropping again.

It's so low that I've only got 3 machines left getting enough credit to 'justify' (hehe!) the electricity costs, and those are still dropping.

Don't know what to suggest. Fixed credit is obviously not viable, and whatever scheme is in place now seems even more random/arbitrary than the delightful (sic) CreditNew.

I guess you could take an average runtime for all WUs so far and grant a fixed credit and hope that it averages out, but some WUs are going to be way over or way under :/

Al.
ID: 164 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1318
Credit: 403,736,438
RAC: 288,164
Message 165 - Posted: 6 Sep 2011, 20:05:07 UTC - in response to Message 156.  

..Credit granted to linux hosts were between 2 and 3 times more than windows for the same cpu time. ...
--Eric

Means this that the 64Bit LinuxApp is faster as 32Bit WindowsApp,or is this a problem with the creditgranting?


Interesting. I have noticed that the same wu runs twice as fast on my linux box as on my windows. The linux version is built with gcc with O3 optimization; the windows version is built with visual studio with optimization turned on (optimize for speed setting). So maybe part of the problem is the inefficiency of the microsoft compiler. Greg is working on a windows build using mingw, it will be interesting to see if it is any faster.
ID: 165 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1318
Credit: 403,736,438
RAC: 288,164
Message 166 - Posted: 6 Sep 2011, 20:25:51 UTC - in response to Message 164.  

Hmmm, there was a brief spike when the credit became reasonable, but now it's dropping again.

It's so low that I've only got 3 machines left getting enough credit to 'justify' (hehe!) the electricity costs, and those are still dropping.

Don't know what to suggest. Fixed credit is obviously not viable, and whatever scheme is in place now seems even more random/arbitrary than the delightful (sic) CreditNew.

I guess you could take an average runtime for all WUs so far and grant a fixed credit and hope that it averages out, but some WUs are going to be way over or way under :/

Al.


It looks like maybe I should monkey with the boinc code to make the credit more fair. Otherwise we may start losing volunteers. Does anyone know if I even have control over this? The results being returned from the clients already have the credit set and that gets committed to the database. Maybe it's possible to intercept the result, modify the credit, and then put it into the database. Not sure how "ethical" this is though. I can foresee projects having a "credit war"; whoever grants the most credits attracts the most volunteers. Just doesn't seem right...
ID: 166 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Dagorath
Avatar

Send message
Joined: 2 Sep 11
Posts: 57
Credit: 1,274,345
RAC: 0
Message 168 - Posted: 6 Sep 2011, 23:25:55 UTC - in response to Message 166.  

I am not familiar with the server code so all I can offer is what I've heard. The admin at AQUA@home recently said that the CreditNew scheme is kind of intertwined into the server code in a way that makes it difficult to monkey with or opt out of. (With earlier server versions you could apparently easily substitute your own credit granting code for the built-in credit code). He also said that monkeying with CreditNew would make it difficult to upgrade to subsequent server code. Maybe he was right, maybe he was just lazy, maybe he just wasn't imaginative enough, we'll never know for sure. They've shutdown that project, btw, nothing to do with credits.

Don't be discouraged by what I have said about AQUA admin's "experience". Monkey away with it as you will. You might want to consult with David Anderson first, not to obtain permission because you don't need it, but to get an idea of the pitfalls. Don't be afraid of a credit war either, we're all used to it and most just don't care anymore. The problem with credits has been with us for so long many crunchers just ignore it.

If you don't do something to boost your credit payout you will definitely have some crunchers leave. Exactly how many remains to be seen. My advice is let the credit whores go. You'll retain a solid base of crunchers dedicated to the work you do. It's up to you.

BOINC FAQ Service
Official BOINC wiki
Installing BOINC on Linux
ID: 168 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ChertseyAl

Send message
Joined: 19 Aug 11
Posts: 45
Credit: 1,014,069
RAC: 0
Message 172 - Posted: 7 Sep 2011, 8:33:41 UTC - in response to Message 166.  

Credit wars belong in the same category as Father Christmas, the tooth fairy and unicorns. They don't exist, so don't worry about it :)

OK, unicorns do exist, but the argument is still valid ;)

Al.
ID: 172 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
frankhagen

Send message
Joined: 19 Aug 11
Posts: 76
Credit: 2,002,860
RAC: 0
Message 173 - Posted: 7 Sep 2011, 15:23:11 UTC - in response to Message 172.  

crazycredits at its best:

53436 35396 7 Sep 2011 11:39:55 UTC 7 Sep 2011 13:08:25 UTC Completed and validated 5,242.71 5,127.42 120.96 Get Decics with Bounded Discriminant v1.06
53429 35389 7 Sep 2011 11:41:14 UTC 7 Sep 2011 13:00:45 UTC Completed and validated 4,425.94 4,301.67 135.36 Get Decics with Bounded Discriminant v1.06
53420 35380 7 Sep 2011 11:39:55 UTC 7 Sep 2011 12:32:37 UTC Completed and validated 3,089.19 3,038.72 124.95 Get Decics with Bounded Discriminant v1.06
53398 35358 7 Sep 2011 11:46:46 UTC 7 Sep 2011 13:37:27 UTC Completed and validated 2,869.65 2,722.25 52.74 Get Decics with Bounded Discriminant v1.06
ID: 173 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Conan
Avatar

Send message
Joined: 3 Sep 11
Posts: 30
Credit: 7,701,817
RAC: 3,703
Message 200 - Posted: 14 Sep 2011, 23:11:00 UTC
Last modified: 14 Sep 2011, 23:52:47 UTC

Well I can't work out why my faster computer (by only 200 MHz, an AMD Phenom II 1100T @ 3.3 GHz, my other is AMD Phenom II 955 @ 3.2 GHz), is getting consistently much lower results than my slower machine.

My slower machine gets from 12 up to high 20's with an average around 15 cr/h.
The 1100T is getting from 6 to 24 with an average around 12 cr/h.

These are some results from the faster computer (Host 415) over the last few days:--

68466 49717 14 Sep 2011 17:30:56 UTC 14 Sep 2011 18:36:59 UTC Completed and validated 443.02 441.05 1.05 Get Decics with Bounded Discriminant v1.05
60841 42263 12 Sep 2011 18:51:12 UTC 14 Sep 2011 12:42:25 UTC Completed and validated 11,966.36 11,804.94 26.74 Get Decics with Bounded Discriminant v1.05
58599 40262 11 Sep 2011 15:32:00 UTC 14 Sep 2011 0:00:55 UTC Completed and validated 10,258.75 10,181.86 19.70 Get Decics with Bounded Discriminant v1.05
58112 39775 11 Sep 2011 3:54:50 UTC 11 Sep 2011 8:58:45 UTC Completed and validated 18,161.44 18,036.39 62.99 Get Decics with Bounded Discriminant v1.05
58001 39664 11 Sep 2011 3:54:50 UTC 12 Sep 2011 18:51:12 UTC Completed and validated 48,209.78 47,959.63 132.41 Get Decics with Bounded Discriminant v1.05
58000 39663 11 Sep 2011 3:54:50 UTC 12 Sep 2011 2:18:41 UTC Completed and validated 50,830.19 50,413.69 174.31 Get Decics with Bounded Discriminant v1.05
57881 39544 10 Sep 2011 23:38:17 UTC 11 Sep 2011 8:09:40 UTC Completed and validated 13,210.59 13,113.48 45.41 Get Decics with Bounded Discriminant v1.05
56307 37970 9 Sep 2011 14:35:22 UTC 11 Sep 2011 3:54:50 UTC Completed and validated 1,677.22 1,665.88 5.12 Get Decics with Bounded Discriminant v1.05
55823 37783 9 Sep 2011 11:51:09 UTC 11 Sep 2011 6:24:55 UTC Completed and validated 10,136.31 9,949.55 33.66 Get Decics with Bounded Discriminant v1.05
55690 37650 9 Sep 2011 9:05:02 UTC 9 Sep 2011 17:00:03 UTC Completed and validated 3,678.39 3,661.20 20.80 Get Decics with Bounded Discriminant v1.05
55502 37462 9 Sep 2011 4:52:53 UTC 10 Sep 2011 20:36:11 UTC Completed and validated 24,761.98 24,593.31 78.03 Get Decics with Bounded Discriminant v1.05
54866 36826 8 Sep 2011 21:53:31 UTC 10 Sep 2011 20:36:11 UTC Completed and validated 55,231.34 54,818.78 208.09 Get Decics with Bounded Discriminant v1.05
54518 36478 8 Sep 2011 9:22:37 UTC 10 Sep 2011 9:22:53 UTC Completed and validated 49,931.89 49,742.33 232.18 Get Decics with Bounded Discriminant v1.05
54389 36349 8 Sep 2011 7:32:10 UTC 9 Sep 2011 17:00:03 UTC Completed and validated 39,597.83 39,464.34 223.95 Get Decics with Bounded Discriminant v1.05
53978 35938 7 Sep 2011 22:33:18 UTC 9 Sep 2011 11:51:09 UTC Completed and validated 50,653.42 50,402.27 296.11 Get Decics with Bounded Discriminant v1.05
53193 35153 7 Sep 2011 8:07:57 UTC 8 Sep 2011 21:05:25 UTC Completed and validated 22,549.56 22,437.20 150.71 Get Decics with Bounded Discriminant v1.05

And these are some results from the 955 (Host 416)

66305 47556 13 Sep 2011 17:37:32 UTC 14 Sep 2011 12:32:07 UTC Completed and validated 1,644.83 1,338.45 5.32 Get Decics with Bounded Discriminant v1.05
60098 41520 12 Sep 2011 10:15:41 UTC 14 Sep 2011 4:21:22 UTC Completed and validated 40,287.70 36,171.00 140.07 Get Decics with Bounded Discriminant v1.05
59999 41421 12 Sep 2011 10:15:41 UTC 12 Sep 2011 18:07:36 UTC Completed and validated 4,853.20 3,253.34 17.51 Get Decics with Bounded Discriminant v1.05
59983 41405 12 Sep 2011 10:15:41 UTC 12 Sep 2011 18:07:36 UTC Completed and validated 3,313.23 2,612.08 11.96 Get Decics with Bounded Discriminant v1.05
58403 40066 11 Sep 2011 10:25:36 UTC 13 Sep 2011 20:02:03 UTC Completed and validated 59,509.64 54,062.41 295.46 Get Decics with Bounded Discriminant v1.05
58051 39714 11 Sep 2011 3:17:57 UTC 13 Sep 2011 3:18:17 UTC Completed and validated 45,160.77 37,294.91 100.52 Get Decics with Bounded Discriminant v1.05
55773 37733 9 Sep 2011 9:07:58 UTC 11 Sep 2011 8:11:43 UTC Completed and validated 9,686.56 9,238.53 45.21 Get Decics with Bounded Discriminant v1.05
55701 37661 9 Sep 2011 9:07:27 UTC 11 Sep 2011 3:17:57 UTC Completed and validated 5,703.13 5,236.75 26.73 Get Decics with Bounded Discriminant v1.05
54751 36711 8 Sep 2011 12:52:40 UTC 9 Sep 2011 2:38:55 UTC Completed and validated 1,610.58 1,050.55 9.79 Get Decics with Bounded Discriminant v1.05
54702 36662 8 Sep 2011 11:44:08 UTC 8 Sep 2011 12:53:06 UTC Completed and validated 12.55 7.44 0.09 Get Decics with Bounded Discriminant v1.05
54591 36551 8 Sep 2011 11:34:27 UTC 10 Sep 2011 19:05:22 UTC Completed and validated 22,650.25 18,939.38 110.83 Get Decics with Bounded Discriminant v1.05
53589 35549 7 Sep 2011 15:40:02 UTC 10 Sep 2011 1:50:09 UTC Completed and validated 39,293.05 34,230.11 207.20 Get Decics with Bounded Discriminant v1.05
52235 34249 6 Sep 2011 22:06:07 UTC 9 Sep 2011 4:13:24 UTC Completed and validated 26,405.25 18,113.33 159.71 Get Decics with Bounded Discriminant v1.05

You can see the differences, amounts awarded for the same or similar run times are wildly different with no consistency.

I am going to struggle to 5,000 points then call it a day as 6 credits an hour is not worth the effort, with 12 and 15 extremely poor.
It has taken me 2 weeks to get 4,500 points.

Conan
ID: 200 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1318
Credit: 403,736,438
RAC: 288,164
Message 203 - Posted: 15 Sep 2011, 2:42:50 UTC - in response to Message 200.  

Well, sorry to see you go. Check back in a few weeks to see if the credit granting is fixed. I'll post a news item if I manage to fix it.
ID: 203 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · 3 · 4 · Next

Message boards : Number crunching : Massive drop of credits per CPU hour


Main page · Your account · Message boards


Copyright © 2024 Arizona State University