> download so many WU's that they can not possibly complete them in the
> necessary time.
>
> Dominique
I found BOINC requests far more work than it should. Maybe this is a problem with the core client and not the user?
Example: Preferences set to 2 days cache (connect every 2 days). Two projects, resource for Einstein/Seti = 200/100. Meaning Einstein gets 66.67%, seti gets 33.33%.
WU's estimated to take 27.5 hours, and have one in progress 30% done, another "ready to run".
Yet BOINC requested 317242 seconds work. That almost 4 days at 100%, or 6 days at 66.67%. And no way would it complete it all in a week, even if the project got 100% of the resources. While estimated at 27.5 hours, they take over 35 hours.
Luckily the scheduler didn't respond and I disabled network communications before it could retry.
> > download so many WU's that they can not possibly complete them in the
> > necessary time.
> >
> > Dominique
>
> I found BOINC requests far more work than it should. Maybe this is a problem
> with the core client and not the user?
>
> Example: Preferences set to 2 days cache (connect every 2 days). Two
> projects, resource for Einstein/Seti = 200/100. Meaning Einstein gets 66.67%,
> seti gets 33.33%.
>
> WU's estimated to take 27.5 hours, and have one in progress 30% done, another
> "ready to run".
>
> Yet BOINC requested 317242 seconds work. That almost 4 days at 100%, or 6
> days at 66.67%. And no way would it complete it all in a week, even if the
> project got 100% of the resources. While estimated at 27.5 hours, they take
> over 35 hours.
>
> Luckily the scheduler didn't respond and I disabled network communications
> before it could retry.
>
>
> Walt
>
So in that case you now know to lower the cache/connect to 1 day. My machine does an E@H WU in 6 hrs. but I still set my cache to 1 day to be sure I can complete everything in time. The BOINC client does seem to have a problem with it's calculation of estimated times but I'm sure it will be fixed in the near future. So anyway, having set my projects to 25% each I know I can do 1 E@H WU per day. In my case I upload 1 and it downloads 1 and everything just flows along.
I wish people would not take like 3 days to finish a work unit. I've had pending credits for work units on there for like 2/3 days now because some doddery old 486 DX is still at 20% and E@H won't validate and give out credits until all results have been recieved. I thought boinc distributed work units depending on system ability/workload, not just randomly plying machines with work units they just can't complete on time.
> I wish people would not take like 3 days to finish a work unit. I've had
> pending credits for work units on there for like 2/3 days now because some
> doddery old 486 DX is still at 20% and E@H won't validate and give out credits
> until all results have been recieved. I thought boinc distrubuted work units
> depending on system ability/workload, not just randomly plying machines with
> work units they just can't complete on time.
If you're on dial-up, and connect every weekend, it's absolut fine to join the project.
If that's a fast machine just crunching Einstein, it's possible that it will deliver 15 - 20 once, DL the same amount, and go back to solitary crunching again.
Perfect fine, go on with this, just don't try too much, so that no deadline is passed.
I think one solution will be that the Einsteins servers would make distintions between "fast computers" and "slow computers" this way people that process results would get the result at about the same time.
Quote:
download so many WU's that they can not possibly complete them in the necessary time.
I Wish people wouldn't....
)
> download so many WU's that they can not possibly complete them in the
> necessary time.
>
> Dominique
I found BOINC requests far more work than it should. Maybe this is a problem with the core client and not the user?
Example: Preferences set to 2 days cache (connect every 2 days). Two projects, resource for Einstein/Seti = 200/100. Meaning Einstein gets 66.67%, seti gets 33.33%.
WU's estimated to take 27.5 hours, and have one in progress 30% done, another "ready to run".
Yet BOINC requested 317242 seconds work. That almost 4 days at 100%, or 6 days at 66.67%. And no way would it complete it all in a week, even if the project got 100% of the resources. While estimated at 27.5 hours, they take over 35 hours.
Luckily the scheduler didn't respond and I disabled network communications before it could retry.
Walt
> > download so many WU's
)
> > download so many WU's that they can not possibly complete them in the
> > necessary time.
> >
> > Dominique
>
> I found BOINC requests far more work than it should. Maybe this is a problem
> with the core client and not the user?
>
> Example: Preferences set to 2 days cache (connect every 2 days). Two
> projects, resource for Einstein/Seti = 200/100. Meaning Einstein gets 66.67%,
> seti gets 33.33%.
>
> WU's estimated to take 27.5 hours, and have one in progress 30% done, another
> "ready to run".
>
> Yet BOINC requested 317242 seconds work. That almost 4 days at 100%, or 6
> days at 66.67%. And no way would it complete it all in a week, even if the
> project got 100% of the resources. While estimated at 27.5 hours, they take
> over 35 hours.
>
> Luckily the scheduler didn't respond and I disabled network communications
> before it could retry.
>
>
> Walt
>
So in that case you now know to lower the cache/connect to 1 day. My machine does an E@H WU in 6 hrs. but I still set my cache to 1 day to be sure I can complete everything in time. The BOINC client does seem to have a problem with it's calculation of estimated times but I'm sure it will be fixed in the near future. So anyway, having set my projects to 25% each I know I can do 1 E@H WU per day. In my case I upload 1 and it downloads 1 and everything just flows along.
Gee, I'm just a dumb airhead,
Dominique
*I still know CRAP when I see it.
I wish people would not take
)
I wish people would not take like 3 days to finish a work unit. I've had pending credits for work units on there for like 2/3 days now because some doddery old 486 DX is still at 20% and E@H won't validate and give out credits until all results have been recieved. I thought boinc distributed work units depending on system ability/workload, not just randomly plying machines with work units they just can't complete on time.
> I wish people would not
)
> I wish people would not take like 3 days to finish a work unit. I've had
> pending credits for work units on there for like 2/3 days now because some
> doddery old 486 DX is still at 20% and E@H won't validate and give out credits
> until all results have been recieved. I thought boinc distrubuted work units
> depending on system ability/workload, not just randomly plying machines with
> work units they just can't complete on time.
If you're on dial-up, and connect every weekend, it's absolut fine to join the project.
If that's a fast machine just crunching Einstein, it's possible that it will deliver 15 - 20 once, DL the same amount, and go back to solitary crunching again.
Perfect fine, go on with this, just don't try too much, so that no deadline is passed.
Grüße vom Sänger
> download so many WU's that
)
> download so many WU's that they can not possibly complete them in the
> necessary time.
Does BOINC get smarter in it's WU requesting?
Nevermind, I've moved my question to here
it's -> its
)
it's -> its
"My other computer is a virus farm."
> it's -> its > :-( If
)
> it's -> its
>
:-(
If Type = fast & Caffeine Enough & Excited = very then
call useBadEnglish
> If Type = fast & Caffeine
)
> If Type = fast & Caffeine Enough & Excited =
> very then
> call useBadEnglish
One of my professors at university always said:
Bad english is the language of science
And I agree!
Grüße vom Sänger
That's Greek.
)
That's Greek.
"My other computer is a virus farm."
I think one solution will be
)
I think one solution will be that the Einsteins servers would make distintions between "fast computers" and "slow computers" this way people that process results would get the result at about the same time.
HOW TO - Full installation Ubuntu 11.10