Another Question About Credits
25 Feb 2005 11:47:09 UTC
Topic 187986
(moderation:
Hi
I'm trying to get my head around this whole credit thing. Does Per WorkUnit Credit increase or decrease if I allocate more CPU cycles to the project?
If my friend and I are both allocating the same number of CPU cycles - but he has a faster processor - which of use will receive more credits? Which is better more credits or less credits?
Can I use credits to compare computers processing speed?
Or..more simply...can someone explain what a credit is and what it is used for. I won't mind if you explain it to me like I am 6 year old.
Another Question About Credits
)
Hi agentsix,
every (distinct) WU should get the same (distinct) credit regardless of the machine it's crunched on. Another WU, another credit, but one WU, same credit. That's the theory.
But as the benchmarks are not that accurate, and counting of every single flop not that effective, and different CPU/OS constellations crunch in different ways, and so on...
BTW:
A good manual for Boinc is Paul D. Bucks BOINC Powered Projects Documentation and his FAQ therein.
For information regarding Credits look here!
Grüße vom Sänger
> Can I use credits to
)
> Can I use credits to compare computers processing speed?
>
Not too accurately. As Saenger said, the credit is caculated using the benchmark results. These are not very consistant across CPU types / operating systems etc. For example, in this host, the benchmarks are lower than on this host/. this is despite the fact that the first machine as 8 CPUs (well... 4 real and 4 virtual) and the second has 2 (1 real 1 virtual).
The result of that are that the claimed credit is always lower for the first host even though 1 CPU -v- 1 CPU on the second host, it actually takes longer to complete a workunit.
Plus there are differences in the WUs themselves which my cause them to run longer / shorter than a previous unit so that will also affect credit.
I guess if you have 2 identical machines crunching the same WU then youcould use claimed credit to compare the performace but other than that....
Paul.
Wanna visit BOINC Synergy site? Click my stats!
Join BOINC Synergy Team
> ...can someone explain what
)
> ...can someone explain what a credit is and what it is used
> for. I won't mind if you explain it to me like I am 6 year old.
>
With those credits and a fist full of ca$h, you can go into any "Starbuck's" coffee in the world and get yourself a coffee or go to any "McDonald's" and get a Happy Meal.
Dominique
*I still know CRAP when I see it.
> > every (distinct) WU
)
>
> every (distinct) WU should get the same (distinct) credit regardless of the
> machine it's crunched on. Another WU, another credit, but one WU, same credit.
> That's the theory.
>
If every distinct WU gets the same credit, and WU are distributed randomly, over time won't the avg credit of all particpants be equal?
Why do I see variance in avg credit between participants?
> Why do I see variance in
)
> Why do I see variance in avg credit between participants?
well if everything else was equal then on a per work unit basis, each participant should have the same credit.
however things are not equal.
For example. I have 2 PCs crunching for Einstein. Some people have 1. some people have 194 this person for example.
That taken into account, you user RA credit will never match his! '(a single host RAC may match one of his host RAC... but thats about it!)
Paul.
Wanna visit BOINC Synergy site? Click my stats!
Join BOINC Synergy Team
So the Recent Average Credit
)
So the Recent Average Credit is actually the sum of the average credit of each individual workstation processing the project(for each specific user)?
It seems like Sigma would be a better label for this if this is true.
> > Why do I see variance in avg credit between participants?
>
>
> well if everything else was equal then on a per work unit basis, each
> participant should have the same credit.
>
> however things are not equal.
>
> For example. I have 2 PCs crunching for Einstein. Some people have 1. some
> people have 194 href="https://einsteinathome.org/%3Ca%20href%3D"http://einsteinathome.org/account/596/computers">http://einsteinathome.org/account/596/computers">this person[/url]
> for example.
>
>
> That taken into account, you user RA credit will never match his! '(a
> single host RAC may match one of his host RAC... but thats about it!)
>
> Paul.
>
>
> So the Recent Average
)
> So the Recent Average Credit is actually the sum of the average credit of each
> individual workstation processing the project(for each specific user)?
>
> It seems like Sigma would be a better label for this if this is true.
>
No, the user RAC is not the sum of the host RAC. The calculation of the RAC is extremely time dependent as it is an exponentially reducing average. Each component (hoost, user or team) is calculated separately when credit is granted for that object.
BOINC WIKI