BOINC scoring..
12 Dec 2005 20:47:22 UTC
Topic 190359
(moderation:
It takes me about 3 times as long to crunch an EAH data unit as it does a SETI or LHC unit, yet I only seem to be getting around half more credits. Doesn't seem fair and I fear it will discourage people from participating.
I understand there's no real value to your score, but the human world runs on what we call 'bragging rights' here.
BOINC scoring..
)
KWSN,
If you were claiming 65, you were using an optimized client, to compensate for running an optimized Seti app that really cut your crunchtime, right? Einstein, first of all, does real science, not some fairy-tale search for a cute little alien with a healing finger. For that reason, the calculations are meaningful, and are done to a precision several orders higher than most other projects, especially Seti, so Einstein is very intolerant of the slight deviations in calculation that enter the picture with third-party apps. They simply would_not_give_valid_results.
Long story short, that leaves us comparing a standard Einstein app against an optimized Seti app, or, apples and oranges.
I hope that participating in hard science research will somewhat compensate you for whatever less total credit it earns you, and your bragging rights within Einstein will be worth just as much as they are within any other project. :-)
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: It takes me about 3
)
Hi KWSN-GMC,
Welcome to EAH. We hope you enjoy your stay here.
Your last group of validated claims over at Seti seemed to range between about 17 and 27 or thereabouts. I added up the last 10 and it averaged 22.848.
You have only three validated results here so far and the average is actually over 80. So if it's only taking 3 times as long for an EAH result and you are getting nearly 4 times the return, I'd say you were doing quite nicely :).
Of course it's still very early days and the numbers do vary quite a bit but you know what they say about statistics - you can make them show whatever you want. I'd be interested in how you worked out you were only getting 50% more (half more you said) credit?? That's a pretty neat way of interpreting the numbers :).
Anyway, have fun, that's all the credits are good for. It's the science that counts and we appreciate your contribution towards that.
Cheers,
Gary.
How silly of me. I hadn't
)
How silly of me. I hadn't actually gone to check the seti credits.. I thought I had an accurate picture in my head of getting low 40's and more for those.
As to the optimized SETI client..no I have not been running one. I have loaded one, but only since this weekend and the current SETI jamup prompted me to take a look at the boards.
I had not been running an optimized BOINC client till then either.
I've been doing this program so long I more or less ignore it now. 5.5 years is a long time. :D
RE: KWSN, If you were
)
Thanks for the info on the increased precision of these calculations, i was unaware. I am WELL aware of the near nill chance of actually detecting other civilizations signals with a ground based, under the atmosphere antenna. Lets not forget, however, it was the SETI@HOME that validated and moved forward the distributed computing program so more valuable research could benefit.
RE: ... it was the
)
KWSN,
Yes, I have said here before that I consider the development of Distributed Computing to be Seti's major, if not only, worthwhile contribution, and certainly not to be trivialized. I did SetiClassic for 17 months, from Dec 4, 2002 to May 2004, to the tune of 3148 results, then reckoned that a move onward was long overdue.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
I believe there is
)
I believe there is 'secondary' data being extracted from our crunching that has benefited other radio astronomy programs as well.
Be that as it may, you will note I have been very happily doing my small part to help design the LHC and now einstein.
I only regret that after running the program so long, I was ignoring it to the point I was so late switching to BOINC.
RE: I believe there is
)
I really hope that is the case, rather than continue thinking that I totally wasted 10000 hours of computer time. :-)
And I sincerely thank you for your contributions.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
As a slight bit of a tangent
)
As a slight bit of a tangent from the original post, but still on the same topic of low credits...does anybody know why Einstein credits seem to be consistently behind Climate Prediction?
I have been watching project stats on boincstats.com for about a month now. Einstein has about 25% more users, 21% more hosts registered, more teams, more countries etc. and yet on average Climate Prediction still gets more credits per day. I have tried looking at percent of active users and other such criteria, but have yet to find a single reason why Einstein credits/day are lower. Furthermore, compared to Climate Prediction, Einstein@Home continues to get more new users and new hosts per day but we don't seem to be catching up very much.
I am not going to jump to any conclusions and say that something is definitely wrong here, but I certainly can't see any reasons for these trends. Does anybody have a clue what I could be missing, or does Einstein really give less credits on average?
RE: As a slight bit of a
)
smith,
Good question, and I have no direct answer. There are other factors, though, which are less apparent, they don't show up on statsheets. Share percentages are one of the more obvious possibiliities, i.e. if someone has Einstein set at 25%, Rosetta at 25%, and Climate at 50%, they show up as 1 host on each project, but more work is being done on Climate. Another possibility, nore likely, is the large number of Einstein host that are primariliy used for their screensavers, with any work being done an ancillary side-benefit. Screensaver/Graphics use somewhat slows the processing. Another factor is the Graphics Bug that afflicts many hosts, and the people who suffer or just plain allow workunits to error-out without seeking help for the problem, for example, this fellow. These hosts return little or nothing to the project.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
RE: As a slight bit of a
)
You have to remember how the BOINC scoring system works. Each computer is benchmarked with a benchmarking program independent of the project (hard coded into the BOINC client, separate from the different flavors of science code). This benchmarking program outputs the equivalent of a credit rate. When a result is completed, regardless of the project, the amount of credit claimed is simply the output from the benchmarking program times the amount of time. Thus, the amount of credit a computer will earn is necessarily independent of the project run. The previous sentence is true, except when it comes to Climate Prediction. Instead of granting credit the way all other projects (to my knowledge) do, they instead grant 6,805.26 credits for each WU, independent of run-time. Thus, the Climate Prediction is different from all other projects. As you have observed, this rate turns out to be significantly higher given the average computer. Disregarding the use of optimized clients and the like, all other projects will grant credit at the exact same rate.