Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummers-lowrise
|
Message boards :
Seventeen or Bust :
Time extension
| Author |
Message |
|---|
|
|
|
I'm about 55 hours from finishing a SoB task, but it's due in 47 hours. Is it possible to get a deadline extension so that my ~380 hours of crunching won't go to waste? If so, how?
If it's pertinent, the WU is llr_sob_44330130.
____________
| |
|
 Michael Goetz Volunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14037
ID: 53948
Credit: 477,382,143
RAC: 302,741
                               
|
|
See my response here.
____________
My lucky number is 75898524288+1 | |
|
|
|
|
I have 4 or 5 WU's on this Host 128184
that the deadline is approaching tomorrow, but still need 80+ hours to finish..
The deadline is too short... and they will not finish in time
When these finish, will I get credit for them? | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14037
ID: 53948
Credit: 477,382,143
RAC: 302,741
|
I have 4 or 5 WU's on this Host 128184
that the deadline is approaching tomorrow, but still need 80+ hours to finish..
The deadline is too short... and they will not finish in time
When these finish, will I get credit for them?
I assume you read the response in my message directly above yours, which (when you follow the link) gives you the answer "probably yes" along with an explanation.
To be a bit more specific about your WUs, you have two that are due on the 9th. In one WU, the WU has one completed result and no others in progress other than yours. If yours times out, another result will be sent out to another computer. As long as you either A) return the results with 5 days of your result timing out or B) return the result before the other computer does (IF it does), you will get credit. So, most likely, unless the WU gets sent to a very fast computer that processes the WU immediately, you'll have about 10 days of leeway.
In the other WU there's also one completed result, but there's two results in progress, yours and another result that timed out on March 16th. It's possible that the computer that timed out in March is still chugging away on this WU, and could return a result at any time. So, in this WU, you still have the 5 extra days to get credit -- or until another computer returns a valid response. That could, theoretically, happen at any time if the other overdue result comes back, but the most likely scenario is also that you have around 10 extra days.
____________
My lucky number is 75898524288+1 | |
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
                 
|
A) return the results within 5 days of your result timing out...
Several weeks ago the purge delay was extended. Please see this thread for further discussion. Unfortunately, at this time, the ideal solution is not available.
Will I get credit for this
FYI...newer WU's currently have a 28 day deadline. Once those older WU's finally get cleared, these expiration cases should be minimized.
____________
| |
|
|
|
|
It seems that some SOB WUs need much more time to finish than others. One of mine, 47464643_0 will need (based on the actual progress, not the predicted time) about 345 hours, while 44331317_5 will need only about 225 hours, what seems quite normal, compared to my other SOB WUs.
While I expect both to be completed in time, I guess we might see some further problems with the deadline if needed CPU-time will increase with growing numbers and not everyone crunching 24*7 at 100 % CPU. | |
|
|
|
|
can you check the k-values
perhaps it has triggered an fft-size change.. | |
|
|
|
can you check the k-values
perhaps it has triggered an fft-size change..
Sure, will do so, if you please explain me what and how to do it. I'm just one of those guys who lets BOINC use unneeded PC-resources. | |
|
|
|
can you check the k-values
perhaps it has triggered an fft-size change..
Sure, will do so, if you please explain me what and how to do it. I'm just one of those guys who lets BOINC use unneeded PC-resources.
Go to your prime grid folder (on my machine its c:\programs\BOINC\Data\projects\www.primegrid.com) and open the file llr_sob_44331317 with a text editor. At the moment I have no WUs on my machine to give an example, but it will be two lines. The first line is header, the second line gives the candidate which is tested.
____________
There are only 10 kinds of people - those who understand binary and those who don't
 | |
|
|
|
|
Thanks!
The one for llr_sob_44331317 reads
10223 17146205
the one for the longer running llr_sob_47464643_0 reads
55459 18053398
Which value means what? So I can learn a bit more about it. Thanks! | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14037
ID: 53948
Credit: 477,382,143
RAC: 302,741
|
Thanks!
The one for llr_sob_44331317 reads
10223 17146205
the one for the longer running llr_sob_47464643_0 reads
55459 18053398
Which value means what? So I can learn a bit more about it. Thanks!
Sierpinski numbers are of the form k*2^n+1. The first number (55459) is k, the second (18,053,398) is n.
It seems we've started doing the 18m range as well as the 17m range we started with, and as n goes up, so does the run time. As I understand it, in addition to the runtime going up kind of linearly as 'n' increases, at certain points the runtime abruptly increases as well.
On the positive side, should you be fortunate enough to find a prime here, a number with an 'n' in the 18 millions will be around 300,000 digits longer than one with an 'n' that is lower by one million.
Hopefully (hint, hint!) the admins are sending out the n=18m SoB WUs with even longer deadlines! The 21/28 day deadlines for the n=17m WUs really only work for CPUs of the Core2 class or better; most earlier Intel chips (anything that says Pentium or Celeron), and their AMD equivalents, were too slow to finish those in time, even running 24/7. (My single-core crunchers would take around 32 to 40 days to crunch one of the n=17m WUs.)
With this jump in runtime, Core2 chips that aren't running 24/7 will start being under time pressure again.
Just for giggles, I wonder how long it would take the Atom N450 in my wife's netbook to do an SoB?
____________
My lucky number is 75898524288+1 | |
|
|
|
As I understand it, in addition to the runtime going up kind of linearly as 'n' increases, at certain points the runtime abruptly increases as well.
Somewhere between 17.8M and 18.5M, the FFT size increases from 1536K to 2048K, in Prime95 at least, but it sounds like LLR is similar in this regard.
| |
|
|
|
|
Thanks for the answer!
They might not only have to adjust the deadline, but maybe the credits too. | |
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14037
ID: 53948
Credit: 477,382,143
RAC: 302,741
|
Thanks for the answer!
They might not only have to adjust the deadline, but maybe the credits too.
Indeed, yes.
____________
My lucky number is 75898524288+1 | |
|
Message boards :
Seventeen or Bust :
Time extension |
