|
KWSN Orbiting Fortress KWSN Distributed Computing Teams forum
|
View previous topic :: View next topic |
Author |
Message |
Atomic Booty Prince
Joined: 10 Sep 2005 Posts: 589 Location: New Orleans
|
Posted: Wed Aug 02, 2006 9:31 pm Post subject: Credit Amounts |
|
|
I've mostly been shrubbing for CPDN, but I have been trying out different projects recently, and noticed that the average credit granted for your time can vary quite significantly from project to project. For instance, Einstein seems to give me about 1 credit every 170 seconds (or 0.00587 cr/sec). But for QMC & Rosetta, I get 1 cr/265 sec (0.00377 cr/sec). Is this correct? And if so, why are different reward scales used within the BOINC system? It would seem to me that could lead to point bribery or something. I mean sure, some people are going to still do the projects they like, but others could be swayed unfairly to shrub for a project based on a more generous credit incentive. What gives? And Ni! |
|
Back to top |
|
|
A Shrubbery Prince
Joined: 24 Jun 2004 Posts: 1861 Location: Pacific NW
|
Posted: Thu Aug 03, 2006 12:05 am Post subject: |
|
|
The climate projects assign fixed credits based on the amount of work done. QMC (and a couple others) have a quorum of 1 which means your credit will be entirely dependent upon the most recent CPU benchmark. Seti (possibly Einstein) use fpop counts to set credits.
Each system has it's benefits and drawbacks. I beleive BOINC in general is looking at a way to even out the disparities in credits between projects but don't expect to see major changes anytime soon. For the most part the system is policed by the users to highlight problems and prevent abuse. _________________ A Shrubbery for Prez. Let's put more bushes in the whitehouse.
KWSN - A Shrubbery
Shrubbing since 2000 for the KaNI!ghts. |
|
Back to top |
|
|
ToeBee KWSN Castellan
Joined: 30 Jun 2004 Posts: 601 Location: Manhattan, KS
|
Posted: Fri Aug 04, 2006 7:15 pm Post subject: |
|
|
Yes, this issue has been identified as a problem (credit inflation they call it) and they are looking at several ways of dealing with it. I know CPDN does an analysis every once in a while and adjusts their credit/time factor to bring it into sync with other projects. There was a big row at seti when the new client version was released and people thought it was giving too little credit. We shall see what happens |
|
Back to top |
|
|
Atomic Booty Prince
Joined: 10 Sep 2005 Posts: 589 Location: New Orleans
|
Posted: Sat Aug 05, 2006 8:43 pm Post subject: |
|
|
Thanks for the explanations. From my reading, it seems that your OS (Windows/Linux) and CPU brand (Pentium/AMD) can also affect the outcome as well, yes? With quorums of 3 or more I suppose it doesn't matter so much, but there's still got to be times when it does. It just makes me wonder, if the flops/benchmarks etc. are known to be unreliable, why use them at all? Why not simply give fixed amounts for work like CPDN?
I imagine it's not as simple as that though, since the kinds of work done in other projects can change, perhaps affecting the time it takes to complete them. For instance, with CPDN, the shrubs are huge, and the operations required to do them are probably pretty similar throughout, making it easy to predict how long it should take and assign it a concrete value. On the other hand, with a project like QMC, I have had shrubs as short as 1 hour, and as long as 26+ hours, and the average credit/sec varies. My theory is that different phases of the project require different types of operations to be performed, and some of them are more CPU intensive than others. Still, it seems that they should be able to assess that in advance, and take an educated guess at a fair fixed credit amount, They could always adjust it later, right? Or maybe it's time for me to go grab another ice cold Looney Lager from the fridge.
Anyway, I do hope they can settle on a standardized cross-project credit system. I will shrub for whichever projects I like, regardless, but after reading about all the drama over at seti (found a link on one of the threads around here), I can see why that Crunch3r guy packed up and went to Einstein. If you're going to move your armada to a new port, it might as well be the one that pays the best, even if that payment is in worthless Mardi Gras doubloons.
|
|
Back to top |
|
|
Sir Papa Smurph Cries like a little girl
Joined: 18 Jul 2006 Posts: 4430 Location: Michigan
|
Posted: Sun Aug 20, 2006 5:44 pm Post subject: |
|
|
Credit differences only matter for Boinc combined. the credit within a project is the same for all persons in that project. There are differences based on processor and OS as well as aparently swap file size and memory. But the thing is if you are only competing against those doing the same projects this does not matter much. For those who are looking at Boinc stats for all projects then that is a different matter. All these projects are not just like comparing Apples and Oranges, there are bananas and lemons, limes, strawberries as well. There are at least 30 different projects listed on the stats page. I personally do not believe that they will ever be able to devise a fair credit system that will make everyone happy. So just concentrate on the project you like. _________________ a.k.a. Licentious of Borg.........Resistance Really is Futile.......
and a Really Hoopy Frood who always knows where his Towel is...
|
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|