BOINC credit across projects not consistent?

Message boards : Questions and problems : BOINC credit across projects not consistent?
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile marmot
Avatar

Send message
Joined: 16 Sep 13
Posts: 82
United States
Message 58843 - Posted: 21 Dec 2014, 20:46:34 UTC

Asteroids (resource 8500) is moving servers this week so my resource settings pushed work down to SETI (resource 1000), ClimatePrediction (resource 500), PrimeGrid (resource 200 GPU only) and Collatz(resource 200 GPU only.

I was hitting credit scores between 20k and 40k last month and now it's between 2k and 9k (even after adding a used i5 laptop early this month).

40% of my credit yesterday was from the 2 older nVidia GPUS running PrimeGrid while the 12 Intel cores did poorly. Checked the event logs and see no problems on the machines and they are not in a crashed state.

I found out that the credit reporting script at ClimatePrediction is malfunctioning (still show credit from Climate yesterday tho) but with SETI getting 8 of 12 cores and the 2 GPU's I should be easily breaking 15-20k a day.

So I'm left with the hypothesis of cross-project credit inflation.
ID: 58843 · Report as offensive
Profile Gary Charpentier
Avatar

Send message
Joined: 23 Feb 08
Posts: 2462
United States
Message 58848 - Posted: 22 Dec 2014, 1:26:23 UTC - in response to Message 58843.  

So I'm left with the hypothesis of cross-project credit inflation.

Just now coming to that realization? Has been discussed for several years. No one has a fix and there isn't the will for one anyway.
ID: 58848 · Report as offensive
noderaser
Avatar

Send message
Joined: 2 Jan 14
Posts: 276
United States
Message 58855 - Posted: 22 Dec 2014, 3:36:14 UTC

It's really coming to a head with Bitcoin Utopia and the high throughput of ASIC devices, though I've yet to hear a system proposed that would be fair to all parties concerned. Someone is always going to get the short end of the stick. I wouldn't get too caught up in it, be more concerned about the work that YOU are doing to support the projects you're interested in.
My Detailed BOINC Stats
ID: 58855 · Report as offensive
Aurora Borealis
Avatar

Send message
Joined: 8 Jan 06
Posts: 448
Canada
Message 58856 - Posted: 22 Dec 2014, 5:18:10 UTC - in response to Message 58855.  

It's really coming to a head with Bitcoin Utopia and the high throughput of ASIC devices, though I've yet to hear a system proposed that would be fair to all parties concerned. Someone is always going to get the short end of the stick. I wouldn't get too caught up in it, be more concerned about the work that YOU are doing to support the projects you're interested in.

I read on the mailing list that DA is working with Bitcoin to find a solution to the ridiculous Credit payout.
ID: 58856 · Report as offensive
Profile marmot
Avatar

Send message
Joined: 16 Sep 13
Posts: 82
United States
Message 58873 - Posted: 22 Dec 2014, 20:13:37 UTC - in response to Message 58848.  
Last modified: 22 Dec 2014, 20:41:12 UTC

So I'm left with the hypothesis of cross-project credit inflation.

Just now coming to that realization? Has been discussed for several years. No one has a fix and there isn't the will for one anyway.


I may have started in 1998/2005 but didn't really pay any attention or join the forums (except in 1998 when SETI@Home was the only game and Moo! was just starting). So your assumption of equal levels of involvement to your own, and making insults upon that assumption, seems to be a method of biting the newcomers and keeping new blood out of the BOINC tribe.

It's not healthy community building and I would hope you consider this when dealing with new or returning members to the forum.
ID: 58873 · Report as offensive
Profile marmot
Avatar

Send message
Joined: 16 Sep 13
Posts: 82
United States
Message 58874 - Posted: 22 Dec 2014, 20:16:27 UTC - in response to Message 58855.  
Last modified: 22 Dec 2014, 20:39:39 UTC

It's really coming to a head with Bitcoin Utopia and the high throughput of ASIC devices, though I've yet to hear a system proposed that would be fair to all parties concerned. Someone is always going to get the short end of the stick. I wouldn't get too caught up in it, be more concerned about the work that YOU are doing to support the projects you're interested in.


Is there a statistician tracking credit across projects per work hour an a reference machine? If so, do you know where that info is displayed?

Credit is a badge of honor to people here or they wouldn't use banners of their credit as tag lines.
ID: 58874 · Report as offensive
Profile marmot
Avatar

Send message
Joined: 16 Sep 13
Posts: 82
United States
Message 58875 - Posted: 22 Dec 2014, 20:35:52 UTC - in response to Message 58856.  

It's really coming to a head with Bitcoin Utopia and the high throughput of ASIC devices, though I've yet to hear a system proposed that would be fair to all parties concerned. Someone is always going to get the short end of the stick. I wouldn't get too caught up in it, be more concerned about the work that YOU are doing to support the projects you're interested in.

I read on the mailing list that DA is working with Bitcoin to find a solution to the ridiculous Credit payout.


Since Bitcoin is funding the projects, I'd agree that Bitcoin Utopia should pay out the best credit per work hour as another incentive. It's does seem ridiculously high though.
ID: 58875 · Report as offensive
Profile Gary Charpentier
Avatar

Send message
Joined: 23 Feb 08
Posts: 2462
United States
Message 58876 - Posted: 22 Dec 2014, 20:42:15 UTC - in response to Message 58873.  

So I'm left with the hypothesis of cross-project credit inflation.

Just now coming to that realization? Has been discussed for several years. No one has a fix and there isn't the will for one anyway.


I may have started in 1998/2005 but didn't really pay any attention or join the forums (except in 1998 when SETI@Home was the only game and Moo! was just starting). So your assumption of equal levels of involvement to your own and making insults upon that assumption seems to be a method of biting the newcomers and keeping new blood out of the BOINC tribe.

It's not healthy community building and I would hope you consider this when dealing with new or returning members to the forum.

Generally, if you don't read the forums before you post and the question you have, has been asked and answered dozens of times, well, we don't exist to do your searches for you.

First, understand the amount of credit granted is under the control of the individual project. Many "less sexy" projects give higher credits as a way of getting more people to crunch them. The BOINC program does not know how many CPU cycles a given task takes. Only the individual projects do, if they bother (some grant a flat amount per unit). BOINC simply takes the number reported as gospel. You might join some of the independent cross project aggregation sites and see what is on their forums. You could also write your own cross project aggregation and "correct" as you see fit for the differences in credit granting. I'm a bit surprised no one has done this, although one of the sites does (or did) publish a comparison chart. You could also search for "credit new" or "credit screw" as some called it.
ID: 58876 · Report as offensive
Profile marmot
Avatar

Send message
Joined: 16 Sep 13
Posts: 82
United States
Message 58878 - Posted: 22 Dec 2014, 21:36:45 UTC - in response to Message 58876.  
Last modified: 22 Dec 2014, 22:35:41 UTC


Generally, if you don't read the forums before you post and the question you have, has been asked and answered dozens of times, well, we don't exist to do your searches for you.


I did a search for 'credit inflation' and it wasn't found. A search on 'credit' was too generic coming up with 'credit to you' 'credit cards' 'credit is still zero' and every release notes for each new BOINC client. It's also not part of the FAQ.

Again you made an incorrect assumption and use it as an insult. If you wish to treat a new member to the forums in that manner that's your prerogative.

First, understand the amount of credit granted is under the control of the individual project. Many "less sexy" projects give higher credits as a way of getting more people to crunch them.


That seems like normal human behavior and it did seem likely to me.

You might join some of the independent cross project aggregation sites and see what is on their forums. You could also write your own cross project aggregation and "correct" as you see fit for the differences in credit granting. I'm a bit surprised no one has done this, although one of the sites does (or did) publish a comparison chart. You could also search for "credit new" or "credit screw" as some called it.


"credit screw" "credit new" is a search term I didn't think of so I'll search for that now.
EDIT1: Searching these forums for 'credit screw' or 'credit new' turns up no results outside of this thread.
EDIT2: It appears that the generic search defaults to 30 days and an advanced search is needed to get unlimited and better results. Here's one thread from 2012: http://boinc.berkeley.edu/dev/forum_thread.php?id=7897#45862

"But you can't compare the BRP4s with the SL6Vs that Einstein also runs. Or the Astropulse tasks at Seti. Or any of the applications at WCG. It's something that the developers are trying to make a system for, that eventually all projects will have to go use, but in the mean time, while it's not there yet, there's no really good way to measure efficiency between projects.
____________
Jord"
Makes me wonder how someone did actually create the comparison chart you mention when Jord claims it's not feasible. I'll look for that chart; I really want to see it (even an old version) now.

EDIT3:
Credit discussion from 2013 http://boinc.berkeley.edu/dev/forum_thread.php?id=8406#49300 with link to CreditNew proposal http://boinc.berkeley.edu/trac/wiki/CreditNew
Credit Proposal discussion in 2008. http://boinc.berkeley.edu/dev/forum_thread.php?id=2403#14747
It appears this is the current credit granting policy http://boinc.berkeley.edu/trac/wiki/ClientSchedOctTen

Don't really have time to spend on cross project credit comparison work but am thankful for your suggestion to check into the cross project aggregation sites.

Thankyou for kicking me in the right direction.
ID: 58878 · Report as offensive
Profile marmot
Avatar

Send message
Joined: 16 Sep 13
Posts: 82
United States
Message 58880 - Posted: 22 Dec 2014, 23:43:33 UTC - in response to Message 58878.  
Last modified: 22 Dec 2014, 23:56:04 UTC

OK, I found the heated debate in the thread about "Utopia Bitcoin not science project" http://boinc.berkeley.edu/dev/forum_thread.php?id=9473#54892 about credit inflation from the ASIC's and FLOP's benchmarks not being suitable for credit calculations. I see the suggestion that some donors with hundreds of cores doing actual science, could look to jump BOINC (over to Folding?) because of the ASIC credit. I'm guessing this kind of discussion happened at various team and cross platform forums.

I see there is a new generalized credit proposal:
http://boinc.berkeley.edu/trac/wiki/CreditGeneralized

Maybe, I'm all caught up now.

This makes things instantly more clear to me.
July 2014 onwards inflation.



And this one:
ID: 58880 · Report as offensive
Profile Gary Charpentier
Avatar

Send message
Joined: 23 Feb 08
Posts: 2462
United States
Message 58881 - Posted: 22 Dec 2014, 23:57:29 UTC

ID: 58881 · Report as offensive
Richard Haselgrove
Volunteer tester
Help desk expert

Send message
Joined: 5 Oct 06
Posts: 5077
United Kingdom
Message 58883 - Posted: 23 Dec 2014, 0:10:04 UTC - in response to Message 58881.  

Try here http://boincstats.com/en/stats/-1/cpcs

BOINCstats is an independent third-party stats site. The official BOINC record - the site that generates the "projects you are participating in" table on the account page at every project - is http://boinc.netsoft-online.com/e107_plugins/boinc/get_cpcs.php

Note the weasel words in the rubric at the top: "If a value in Row A and Column B is greater than 1, that indicates that project A is awarding more credit per cpu second than project B."

"Per cpu second" is a meaningless measure for any application which runs on a GPU or other coprocessor, and at the point the whole pack of cards falls over.
ID: 58883 · Report as offensive
Ken Putnam

Send message
Joined: 3 Nov 05
Posts: 15
United States
Message 58978 - Posted: 25 Dec 2014, 10:03:09 UTC - in response to Message 58876.  
Last modified: 25 Dec 2014, 10:05:56 UTC

Bitcoin may be the MOST ridiculous, but only by degree

I've been crunching since the SETI command line days, and joined Rosetta and Einstein in the 2009/2010 time frame

I joined Bitcoin, Asteroids, POEM, and Milky Way in June of this year, when I got a new box with a decent GPU

Take a look at my signature. SETI/Rosetta/Einstein have ALWAYS shared the largest resource share, but look where they rank, based on "credit"
ID: 58978 · Report as offensive
Profile marmot
Avatar

Send message
Joined: 16 Sep 13
Posts: 82
United States
Message 59086 - Posted: 30 Dec 2014, 10:47:50 UTC - in response to Message 58978.  
Last modified: 30 Dec 2014, 11:47:18 UTC

Bitcoin may be the MOST ridiculous, but only by degree

I've been crunching since the SETI command line days, and joined Rosetta and Einstein in the 2009/2010 time frame

I joined Bitcoin, Asteroids, POEM, and Milky Way in June of this year, when I got a new box with a decent GPU

Take a look at my signature. SETI/Rosetta/Einstein have ALWAYS shared the largest resource share, but look where they rank, based on "credit"


So going by your sig picture, Milkyway, POEM, Asteroids, Albert and CPDN are inflated compared to SETI/Rosetta. My personal experience has Asteroids giving me much higher RAC's than Einstein, Climate and SETI but it's not in either of the tables linked so I can't confirm.
ID: 59086 · Report as offensive
Profile marmot
Avatar

Send message
Joined: 16 Sep 13
Posts: 82
United States
Message 59088 - Posted: 30 Dec 2014, 11:04:35 UTC - in response to Message 58883.  
Last modified: 30 Dec 2014, 11:46:33 UTC

Try here http://boincstats.com/en/stats/-1/cpcs

BOINCstats is an independent third-party stats site. The official BOINC record - the site that generates the "projects you are participating in" table on the account page at every project - is http://boinc.netsoft-online.com/e107_plugins/boinc/get_cpcs.php

Note the weasel words in the rubric at the top: "If a value in Row A and Column B is greater than 1, that indicates that project A is awarding more credit per cpu second than project B."

"Per cpu second" is a meaningless measure for any application which runs on a GPU or other coprocessor, and at the point the whole pack of cards falls over.



Thanks for the link.

If PrimeGrid is given a decent CUDA GPU of the same time period (like Dell m6300's Quadro FX 1600 with an Intel t7500 CPU) it is equal or better, in credits per second, than Einstein@Home on the dual cores. It's very attuned to GPU calculations and so, yes, the 4400x figure in the chart is unhelpful.

Still, it's a nice guide for CPU based WU's.
A GPU chart would be interesting.

The http://boincstats.com/en/stats/-1/cpcs shows GPUGrid giving out WU credits per second at the highest rate of participating projects.
I don't see a key for how the table is produced and it's not mentioned in the FAQ's.
ID: 59088 · Report as offensive
Richard Haselgrove
Volunteer tester
Help desk expert

Send message
Joined: 5 Oct 06
Posts: 5077
United Kingdom
Message 59091 - Posted: 30 Dec 2014, 12:28:22 UTC - in response to Message 59088.  

The http://boincstats.com/en/stats/-1/cpcs shows GPUGrid giving out WU credits per second at the highest rate of participating projects.
I don't see a key for how the table is produced and it's not mentioned in the FAQ's.

Sorry, didn't see your edit. Forget it - it's meaningless.

The value 'credit_per_cpu_second' is (was) calculated by the project servers, and exported in the stats tables I was writing about this morning.

It was removed from the server code in April 2010 (changeset 2536797068de70d37b1d5b5d64e160720b3f8430), and as Joe Segur wrote at SETI in January 2012 (message 1189741),

Unfortunately, BOINC is no longer updating the credit per CPU second value. The update was removed from BOINC source code in April 2010, (changeset as above). The change would have been implemented quickly [at SETI], so any cpcs values for this project reflect processing done before then. There may be a few projects where the servers are still using older code so the values are current, but can only be meaningful at a project which has no GPU apps.

The field remains in the HOST records in the database and is included in the stats dumps. If a host is upgraded with a new processor, etc. but doesn't get a new hostID the old cpcs value remains. Any use of the cpcs values now should recognize they are mostly ancient history.
                                                                  Joe
ID: 59091 · Report as offensive

Message boards : Questions and problems : BOINC credit across projects not consistent?

Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.