Pending credit rising fast
10 Jul 2006 9:33:38 UTC
Topic 191551
(moderation:
Hi there,
I'm not sure if this topic is already on this boards (if it is, i'm very sorry), but i have a question about my pending credit rising fast.
Since the new s5 units my pending credit is rising fast and hardly any unit is being validated. I currently have over 3200 pending credit. Is there something wrong at E@H? Or is it my computer? I have a AMD 64bit 4800+ dual core at 2,4Ghz and 1gig ram.
Could someone provide me some info about this?
Thanx!
Pending credit rising fast
)
thats all ok.
The big WUs lead to longer return times. If you have a computer with a small WU cache, you have to wait longer for your quorum partner.
After some time the difference between finished WUs by your computers and validated WUs by your quorum partners will balance to a (nearly) constant value.
But the credits will come...
Udo
Ah ok .. Thanx for the
)
Ah ok ..
Thanx for the info ;)
My sincere appologies to
)
My sincere appologies to anyone who has any of my machines as a quorum partner at the moment. I was fiddling around with my Connect-to interval to try to grab a few extra LHC units from the last batch and somehow ended up caching 10 days worth of everything except LHC. :-(
After running in EDF mode for a couple of days to clear out all the Predictor@Home work, I'm finally back to crunching the backlog of Einstein. Please be patient, folks. Your credits are on their way! :-)
Regards,
-- Tony
RE: My sincere appologies
)
Tony, to actually get those WUs, do the following:
1)Set all projects(except LHC) to NNT/NNW
2)Set all projects(except LHC) to Suspend
3) increase your "connect to" setting
4) Update LHC
If you don't set them to NNT/NNW first then the remaining projects will request work before LHC does.
Note: follow the reverse order to turn the projects back on without getting boatloads as well.
tony
RE: RE: My sincere
)
Another possibility, which includes avoiding fiddling with connect every setting, is to suspend the rest of projects and allow work fetch of your belowed one (LHC). After you get the work from LHC, resume all projects. Possibly set BOINC CC to suspend network while you're suspending and resuming projects to avoid some project fetching surplus work.
If a project is suspended, its time share is split among the rest. If only one project is not suspended, it obviously gets all of the time and as BOINC CC doesn't know when you're about to resume some other project, it'll topp up the WU cache.
Depending on your resource share you can get a lot of work. Example: if your setting of connect every is 2 days and if you're attached to LHC, Einstein and Predictor and you have resource share split evenly (eg. every project gets 1/3 of CPU time), then you'll end up with (ideally) 2 days worth of LHC work. When you resume other projects this translates to 6 days of LHC work.
The lower the resource share, the biger 'over-subscription'.
Metod ...
@ Tony and Metod, Thanks
)
@ Tony and Metod,
Thanks for the tips, guys. I realised I had forgotten to suspend/NNW the other projects about 5 seconds after they had all requested 10 days of work (or as much as they could get and still make deadlines).
I don't normally try to hoard work units for any project - my default cache is 0.1 days. But since LHC work is quite scarce at the moment and I like credits almost as much as I like science , I thought I'd join all the max-cache crunchers and try to grab a bigger chunk of the available work. If I decide to go that route again the next time LHC work is available, I'll make sure that only LHC is able to download work before I increase the cache setting.
Regards,
-- Tony
[Edit: spelling]
RE: @ Tony and
)
I got some LHC work last nite, might wanna try this now. LOL
[edit]dang, Now they're out of work again. It's a pain doing this when you're attached to 13 projects. LOL
I also have some pending
)
I also have some pending credits that have been sitting there for a long time. Has anything changed that could influence this?
RE: I also have some
)
Unfortunately Grinz you had some work units assigned to a quorum partner who didn't crunch them. Einstein then had to wait until the deadline (2 weeks later) before it realised that a result was not coming in. It then reassigned the units and you should see them validated in the next few days. Many of these problems were caused by the increase in the length of the work unit size at the end of June but they are very nearly worked out of the system. Your credits are on their way. Keep crunching, dAVE
Thank you Dave - much
)
Thank you Dave - much appreciated. At least I know it is not a major problem of some sort.
It will be nice when I suddenly get all 880 credits due to me :)
Lee AKA Grinz