I have 12 Work Unit's Pending....
18 Jan 2006 22:13:14 UTC
Topic 190629
(moderation:
Hello Everyone... I have 12 work units pending credit... a few of them have been pending for a bit of time (a week or so). Is this typical of einstein? I have been working Seti for several years... and am used to having them pend only for a day or so....
I have 12 Work Unit's Pending....
)
Jeff,
Welcome to Einstein, and join the club. My machine, single-core, simple Athlon XP, has 45 WUs pending, nearly all of them from the z1_0137.0 datafile, and no end in sight. I seem to be paired with crunchers who like to hold a 6-day cache. I just pity those who are running quad Xeons and such. This 3-crunchers-to-a-WU gig just isn't working out too well for us.
Regards,
Michael
edit - I suppose it will get a little better when we've drained all the doggone tiny alberts, but this is getting old real quick! I'm about to scrap the z1_0137.0 file in my einstein folder, and keep on scrapping it until it doesn't return again.
microcraft
"The arc of history is long, but it bends toward justice" - MLK
Actually, it won’t be any
)
Actually, it won’t be any different then the smaller Albert’s are done, it might just seam that way. As it is now any host has exactly the same chance to be assigned to a particular data set, as any other host. You will be assigned work on whose datasets you already have downloaded as long as there is available work from them. So in theory, any 3 hosts could be pared up to work on a WU as long as they belong to different participants.
If you want to make the number of pending WU’s smaller, you would have to pare up hosts with similar turnaround times. So for an example, the project could assign host with an average turnaround time of less then 4 days to some data sets, and host with turnaround times of more then 4 days to other data sets. This would reduce the number of pending WU’s and the number of WU’s in the database. Then we were beta testing the project, Bruce was recommended not to use the “average turnaround time� to divide hosts. After thinking about this for over a year, I can only se two problems with it.
If you divide hosts in to too many groups, one of the groups could contain so many hosts with problems that WU’s will error out because of that.
Credit. Older/slower hosts tend to claim more credit/result. Dividing hosts would affect the amount of granted credit/WU within that particular group. Changing to flops counting for claiming credit like SETI will do, would eliminate this problem.
Then you're really interested in a subject, there is no way to avoid it. You have to read the Manual.
I've just added 4 additional
)
I've just added 4 additional stations to Einstein from Grid. All I'm getting now is pending WU from all 7 of my stations.
When they get verified I should have a load.
Question: is Albert causing all these pending? My original stations would have a pending for a day at most, now they process, submit and keep processing. All the time building pending files for each station.
RE: I've just added 4
)
I've had 2 WUs pending for a while now. The person we're waiting to report downloaded 59 WUs & hasn't submitted any results at all. These 59 WU become overdue on January 26th. Then they will be reassigned to others who, hopefully, will get results.
RE: Question: is Albert
)
The answer to this question is no.
The correct answer is when they changed to Albert, they now only send 3 results out, and all 3 have to be returned before they verify. Previous 4 were sent, and only 3 were needed.
I personally have seen my PENDING double since this change. They are trying to do more work, with not having to send the 4th result, but I think it's starting to make some small issues, and may grow in the future.
RE: RE: I've just added 4
)
Yeah, I've had this too. Lots of the reissued WU's go to similar users, so now I have to wait 28 days before they fail again. Lord knows whom they will be issued to then!
But apart from that, this is one of the best run BOINC programs.
Join the #1 Aussie Alliance on Einstein
When I look at my account, I
)
When I look at my account, I see I have units pending back from when I attached to the project.
The answers I have seen are only partially addressing my specific question and only in a general way.
RE: When I look at my
)
What is your specific question?
RE: When I look at my
)
Credit is given only when a quorum of 3 resultss have been returned successfully. All 3 do not need to be valid, only 2 of them do, but all 3 must at least have been crunched and returned in good order. Since they only send out 3 results, and some people do not return them in the 2 week period, then it needs to be sent to a 4th. If that doesn't come back a 5th, and so on. Each at 2 week increments. Since it can go up to 20 times, it could take several months before a WU is validated.
RE: When I look at my
)
Please see the FAQ, linked to from the E@H front page.
Director, Einstein@Home