Many pending credits
1 May 2007 8:55:33 UTC
Topic 192651
(moderation:
After having crunched for 10 days I have 20K granted credits and 20K pending credits, is it normal to have so many pending credits in this project based on my production rate, 4k/day apperently?
Many pending credits
)
To give a example what can happen 3340314 if something goes wrong. It's been back in the unsent bucket. It can take some time befor my pending credits for this WU are granted.
"souls ain't born, souls don't die"
RE: RE: After having
)
well, 85% of my pending results are stil unsent to another cruncher
i think, if result is errored, it wait 1-2 days for re-sent
GOD is only word
Thanks for the replies. I
)
Thanks for the replies.
I have also been worried by this "Unsent" status. It looks better now, feewer are in this state.
Those big WUs do get your
)
Those big WUs do get your pendings up. Mine have been climbing quite steadily since the beginning of the run and are now at more than 2000, with an RAC of about 200...
RE: Those big WUs do get
)
well, i would crunch for next 2 days
if nothing changes (~85% pending+unsent), i switch E@H to S@H
it is only small difference (for my Intel pD 3.4)
E@H 12.5-13cr/hr
S@H 13-13.5cr/hr
GOD is only word
Update: I'm at 3600 pending
)
Update: I'm at 3600 pending now, which would be a few days worth of crunching on my boxes. It doesn't seem to be getting any less... I'm curious how this is going to end, I'm reaching a new personal record every day ;-)
RE: Update: I'm at 3600
)
My pending credits seem to have stabilized at around 20k, but I keep my fingers crossed and hope it will not increase to 18(!) times my RAC :)
I have the same problem - I
)
I have the same problem - I have 7.5K pending credit covering most of my WU's uploaded in the last week.
Is this really normal?
RE: I have the same problem
)
Whenever I feel the pending credits are mounting, I check the server status page to see if the validator might be stopped. You can also check the daily credit statistics at boincstats.com to see how the validator was doing in the last few days, here's the chart for Einstein@Home which looks ok as I'm writing this:
The two "indentations" are the transition from S5RI to S5R2 and the validator crashes last weekend.
In contrast, SETI@Home has some troubles lately validating (and distributing), so a graph looking like this would indicate problems:
CU
BRM
RE: I have the same problem
)
shauge's machines are hidden, so I can't comment on them specifically, but this machine is a fast dual dual-core that appears to be spending most of its time crunching einstein. Large amounts of pending credit are quite normal for a fast system, especially a fast multi-core system.
I couldn't find a good explanation of how einstein works before my patience ran out, so here's what happens as I understand it. When you download work from einstein, you download data files containing the data for many workunits. When you complete a result, einstein issues you another result from the data files you already have, if there is one available.
If your machine is really fast compared to the other machines that have downloaded the same set of data files, you end up crunching through result after result before anybody else, and have lots of pending credit. If your machine is fast enough ("fast" being dependent on how much time your machine spends crunching einstein, how many CPUs {cores} it has, and the raw speed of each CPU), you can end up with many results that you have returned that have not yet been issued to another cruncher.
As an example, take a look at this machine of mine.
I consider what I'm seeing in its results to be perfectly normal. I'll get my credits once the other crunchers who have the same set of data files catch up with my dual-quad box.
Maybe something about this should be added to the FAQs?