It does wait until BOINC thinks it will be late, but the time it thinks it
needs is based on the resource fraction, and with 1/1000 of the time, BOINC
believes that it will be late as soon as it arrives. EDF is not a problem.
jm7
Raistmer
<[email protected]
> To
Sent by: "Lynn W. Taylor" <[email protected]>
<boinc_dev-bounce cc
[email protected] [email protected]
u> Subject
Re: [boinc_dev] Suggestions
01/26/2010 02:49
PM
> ... and if you left it alone, it would reach a nice stable equilibrium,
> with a proper DCF (so it can predict deadlines and request work
> accurately) but instead you say "oh, no, it's not working!" and make it
> stop.
No. it wouldn't, I waited long enough.
> Of course BOINC crunches the Einstein work to completion immediately.
With
> a 1000:1 resource share anything else would devote less than two minutes
> per day, and that isn't enough to MEET DEADLINES.
>
> You'd be really angry if deadlines were missed.
Well, what is wrong with this approach: rise low-share project priority
only
if number of hours needed to complete downloaded task let say 1/2 or even
1/3 of time remaining to deadline?
Hope you get the point - it should increase priority when it approaches to
deadline, not right after downloading.
1/2 or 1/3 or even 1/4 can be used as safeguard to no miss deadline. But
currently we have 6hours/14*24 hours. Too much to really think it could
miss
deadline!
> We also have this amazing ability as crunchers to forget that there are
> other stakeholders out there -- that the projects themselves are vitally
> interested in how BOINC works.
No prob with that, but BOTH sides should be respected. Do you think it
would
be best for Einstein if I just detach this project completely and will
leave
my host idle when SETI has no work? Do you really think it would be best
solution? If yes, next will be proposal to drop all P3 cause"they only take
place in project/BOINC database"... bad way IMO...
> It's bad enough that a 1000:1 ratio of shares is allowed, because you
take
> up room in the BOINC database at Einstein, but you aren't going to
> contribute much.
Yes, I'm aren't going to contribute MUCH, but I still going to contribute!
Feel the difference.
> We've seen other projects suffer severe loading problems when s...@home
is
> down. It makes their capacity planning difficult, and it aggravates
their
> loyal users when all of the work gets sucked up.
It says only about popularity of SETI and probably bad server capacity of
other projects :P
> If you're willing to add a project as a "backup" you should be willing to
> do some work for them in exchange for that privilege.
I'm surely WILLING to do work for it, if not I would just leave that
project. I just willing to do that work under particular circumtances.
If user decides not to do any BOINC work when computer not idle - it's OK.
Well, I'm decide to do Einstein when SETI idle, look at this from such
point
of view. Or we should deny screen-saver mode? ;)
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
