But because of the credit cheat of the versions until 4.19 it's silly to compare with it ... so: forget the credit. the only thing that aims are correctly finished wus, nothing else.
I agree. I said the same thing on Cafe' Seti. The stats could show WU's completed and the work credits. The WU's completed is a number everyone can look at and understand. The WU credits is a number that will balloon from thousands, to millions, to trillions, to zillions and mean nothing when viewed.
> I agree. I said the same thing on Cafe' Seti. The stats could show WU's
> completed and the work credits. The WU's completed is a number everyone can
> look at and understand. The WU credits is a number that will balloon from
> thousands, to millions, to trillions, to zillions and mean nothing when
> viewed.
I really don't get it.
What's so nice about comparing oranges and apples?
I agree, that the data here @Einstein are not that divergent as on the other projects, especially Seti, but I also do not know, whether the WU-size will vary with more results and better science applications.
But the real donation is crunch time, or, if you don't want to give the older machines a push, flops. That would be a huge number, so the project team decided to create some kind of WU-size norm, and called it credit. It's calculated this way: ([whetstone]+[dhrystone]) * wu_cpu_time_in_sec / 1728000
and is only granted, if the result is validated.
There is a 'problem' atm with the benchmarks, but that's already corrected, and only lingers because not everyone has updated to the recent client.
Another possibility for 'measurement of devotion to the scientific cause' would be some project specific real values, like 'parsecs searched', 'proteins folded', 'years simulated' (that's done at CPDN) or whatever, so it's up to the project teams to implement this. But I think it's hard for the stats sites to make anything out of this data, as it's by definition not interchangable like CPU-time.
But so what? AFAIK it'll be a challenge, not a problem for the enthusiastic community to make it work.
My personal conclusion:
- WU-count is quick'n'dirty, but imho far too dirty
- credits reflect the real thing, but benchmarks still have to be improved
- real physical data would be a nice-to-have feature for the project teams to develop
- if the amount of calculations per wu would be constant, there won't be any need of credit ... so because this varies, it's good to have this credit system
-> but as you said the users aren't up2date at all and that's the problem
--> now, we all have the problem that the different versions aren't comparable ... so the credit system is more or less useless, I think in case of a fast fixing of the problem it would be in a certain way a good thing when only wus crunched by a minimum version would be accepted, the problem with this is that the project would lose crunchers just because of the comparability of wus what would be stupid (the scientific thoughts)
Credits
)
Hello David,
a bunch of them and some $$ added will get you a coffee at starbucks ;)
It's only for your bragging rights (or, to put it more polite: It's a measurement for your dedication towards the advancement of science ;)
BTW:
The (very good) manual for Boinc is the BOINC Powered Projects Documentation by Paul D. Buck and the FAQ therein.
For information regarding Credits look here!
Grüße vom Sänger
But because of the credit
)
But because of the credit cheat of the versions until 4.19 it's silly to compare with it ... so: forget the credit. the only thing that aims are correctly finished wus, nothing else.
Greetings, Santas little helper
I agree. I said the same
)
I agree. I said the same thing on Cafe' Seti. The stats could show WU's completed and the work credits. The WU's completed is a number everyone can look at and understand. The WU credits is a number that will balloon from thousands, to millions, to trillions, to zillions and mean nothing when viewed.
> I agree. I said the same
)
> I agree. I said the same thing on Cafe' Seti. The stats could show WU's
> completed and the work credits. The WU's completed is a number everyone can
> look at and understand. The WU credits is a number that will balloon from
> thousands, to millions, to trillions, to zillions and mean nothing when
> viewed.
I really don't get it.
What's so nice about comparing oranges and apples?
I agree, that the data here @Einstein are not that divergent as on the other projects, especially Seti, but I also do not know, whether the WU-size will vary with more results and better science applications.
But the real donation is crunch time, or, if you don't want to give the older machines a push, flops. That would be a huge number, so the project team decided to create some kind of WU-size norm, and called it credit. It's calculated this way:
([whetstone]+[dhrystone]) * wu_cpu_time_in_sec / 1728000
and is only granted, if the result is validated.
There is a 'problem' atm with the benchmarks, but that's already corrected, and only lingers because not everyone has updated to the recent client.
Another possibility for 'measurement of devotion to the scientific cause' would be some project specific real values, like 'parsecs searched', 'proteins folded', 'years simulated' (that's done at CPDN) or whatever, so it's up to the project teams to implement this. But I think it's hard for the stats sites to make anything out of this data, as it's by definition not interchangable like CPU-time.
But so what? AFAIK it'll be a challenge, not a problem for the enthusiastic community to make it work.
My personal conclusion:
- WU-count is quick'n'dirty, but imho far too dirty
- credits reflect the real thing, but benchmarks still have to be improved
- real physical data would be a nice-to-have feature for the project teams to develop
Grüße vom Sänger
- if the amount of
)
- if the amount of calculations per wu would be constant, there won't be any need of credit ... so because this varies, it's good to have this credit system
-> but as you said the users aren't up2date at all and that's the problem
--> now, we all have the problem that the different versions aren't comparable ... so the credit system is more or less useless, I think in case of a fast fixing of the problem it would be in a certain way a good thing when only wus crunched by a minimum version would be accepted, the problem with this is that the project would lose crunchers just because of the comparability of wus what would be stupid (the scientific thoughts)
Greetings, Santas little helper