Comment for Dr. Allen & EAH Team
18 Apr 2007 15:26:51 UTC
Topic 192601
(moderation:
I was reading through the Boinc Dev mail list this morning and saw your reply to John's remark and graphs pertaining to the credit rates given by the various projects.
I would like to say you I think you should reconsider "devaluating" the intrinsic worth of EAH WU's more with the upcoming app for a several of reasons:
1.) Somebody has to set the "gold" standard for what the intrinsic value of BOINC work is. IMO, due to the close and continuing collaboration you have had with Berkeley as well as being one of the longer running projects, EAH and SAH are in the best position to fulfill this role.
2.) EAH in particular has gone through more effort to work application optimization into the main stream app on a regular basis than any other project, including SAH where they have the advantage of having the app open source and can draw on the community at large. As a result, I see no reason why the intrinsic value of the work here should be reduced further, considering that a major correction was made at the start of S5R1. At this point, I believe it is up to the other projects to bring their rates up, either by improving the performance of their app, or studying the credit issue better and set the intrinsic value of their work more appropriately.
3.) My experience on the SAH Beta Project indicates that the SAH team was targeting the rate at EAH for the new MB & AP apps, since my hosts at least are showing virtual credit parity at this point.
4.) I've read some commentary lately which indicates to me not all projects have full grasp of the finer points of the BOINC credit system, and the intracacies of cross project credit parity.
Regards,
Alinator
Comment for Dr. Allen & EAH Team
)
Well, I usually get more credit per CPU time on DRTG then I do here... maybe Einstein does give a bit more credits than average but imo it can't be that extreme. Besides, what Alinator says sounds logical to me, so I tend to support his opinion.
Still, I'm not really here for the credits but for the science, the great information you can get about computers and, last but not least, the awesome community. So if you decide otherwise, it's no problem for me.
I'll come back later to
)
I'll come back later to comment about this. Just now I need to go and get some sleep. (There are a handful of us who are starting to fade badly in the push to get our new search up and running. It's a bit scary since I am spending hours logged in working as root on the main E@H server. Not very sensible...)
Bruce
Director, Einstein@Home
Yeah, I think I can imagine
)
Yeah, I think I can imagine what that feels like. Not a really good idea. Just take your time, the science is certainly more important than credit issues and even scientists and admins need their sleep ;-)
We should all pitch in and
)
We should all pitch in and get Bruce a futon for his office! That way he can crash whenever he wants to.
RE: We should all pitch in
)
Or buy him a ticket for "One Free Clone" ..... :-)
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: .... the intracacies of
)
Hmmm .... I think the intellectual process would be somewhat similiar to the finalising of the value of the 'Euro', prior to it's introduction, amongst the various European nations that participate in that currency.
As always, the problem would be reaching sufficiently definite agreement upon the level of 'worth' or 'value', the rest is math. You'd have to exclude 'softer' indicators of merit ( alas like 'value to society' ), so as to avoid needlesly anxious dispute that could never settle. That would leave 'harder' and more measurable properties, some sort of 'fiducial' benchmark machine. It would be fictional in that a version of it doesn't actually exist but, as per Alan Turing, it can be deemed to have metrics that can correlate with actual devices. ( Recall that Turing Complete (?) machines can be equivalent in function - input and output identical - but differ in resources to achieve that, basically speed and memory )
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
Hate to say it, but this was
)
Hate to say it, but this was the the original concept of "cobblestomes". It was based on the work done...
Not long after BOINC went live, the term "cobblestones" faded, and "credits" was then used. (a "credit" is far different than what was defined as a "cobblestone" back when Boinc started...
Throw into the mix various optomized versions, multiple OS's, and other variables (like OS's and processors that spring up every few weeks), and both Cobblestones and Credits seem to provide an "apples and orages" comparison with other projects/HW, etc...
The only real measure of what a cruncher really did is "how much time was contributed to the project". This takes into account the slowest box and the 8x servers...
The guy that spends 60 minutes crunching on his 500mhz Pent on WU's gets the same "credit" as 60 minutes on a Pent 4 at 2gig crunching WU's..
They both contributed 1h of time... While othere won't agree, this is a valid measure that can be used for any project on any HW/SW, and will level the playing field...
Now if the crunching code for all projects could be changed to count the number of integer/FP math operations done, this also be used, but the "vampire code" to do this (vampire in that it serves no purpose to DC other than to "count for stats") serves no other purpose....
I can see the debate that "faster machines should get more credit", but in reality, didn't the guy on the slow machine contribute the same time as the guy on the fast machine?
Just food for thought... The guys looking for credits won't see it this way, but why is an hour of CPU time on your machine worth more than an hour of CPU time on a 500mhz machine? You both paid for 1 hr of electricity, right?
If you're in it for the science, the fast guys should be happy that they did more science during that hour, but if you're in it simply for credits, you'd kick and scream.
Don't many people get paid by the hour? Seems like that's a valid way to be "paid" here....
Hey, think about it.. when you describe DC to others, wouldn't it be easier to say "I donated 2500 hours of computer time to find a cure for Cancer" than to say "I got 376.876 credits for the computer time I spent finding a cure for cancer"?
RE: RE: We should all
)
RE: RE: .... the
)
Agreed, my position on that is in the interest of compromise and fairness you have to go by the assumption that all the science has the same intrinsic value.
I feel SAH and EAH best fufill the standard role since you both support the widest range of hardware and software platforms. Not to mention seem to have the most consistently open environment for participant input.
The social value question is left to the community to decide by either participating or not.
The other side of that coin is what happens if a project gets desperate for support and starts inflating their work worth in an attempt to attract the CW's?
In any event, the question is not as simple or cut and dried as many seem to think.
Alinator
RE: Hate to say it, but
)
Agreed, I'm not suggesting we go back to BM based scoring. That just doesn't cut it with BOINC being Open Source.
I don't see anything wrong with just stating CPU time donated and/or WU's processed as recognition either in theory , however as you well know that has gotten a pretty frosty reception whenever it's been brought up. ;-)
Alinator