Decreasing deadlines
log in

Advanced search

Message boards : News : Decreasing deadlines

Author Message
Profile rebirther
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 2 Jan 13
Posts: 7228
Credit: 42,729,227
RAC: 31
Message 485 - Posted: 3 Jan 2015, 18:08:24 UTC
Last modified: 3 Jan 2015, 18:17:13 UTC

Many WUs are running into timeout and it will take very long to get them back if they will be sent out again and again. If your host cant finish a WU in time like the long runners pls disable the app in your preferences. The main reason is that I cant add new work until all files are returned. This wastes a lot of time time. I will decrease all deadlines for new WUs.

frankhagen
Send message
Joined: 7 Jun 14
Posts: 36
Credit: 241,030
RAC: 0
Message 486 - Posted: 3 Jan 2015, 18:18:16 UTC - in response to Message 485.

good call!

Profile Steve Hawker*
Avatar
Send message
Joined: 8 Dec 14
Posts: 26
Credit: 610,307
RAC: 0
Message 488 - Posted: 3 Jan 2015, 18:59:19 UTC

Reb,

I don't really understand the need for short deadlines. These prime numbers have been there for an awfully long time and a few days isn't going to make any difference in the long run.

Having said that, you might find fewer WUs timeout if the DCF is corrected. My last batch of long tasks came in with estimates of less than 3 hours yet have taken up to 15 hours to crunch. BOINC wouldn't have downloaded so many had the estimate been more accurate.

Then again, you may be struggling with the same problem as at PG - there isn't a sub-project DCF. Beats me how there wasn't one from the start of BOINC and it's even more incomprehensible that there still isn't.

Okey Dokey, lets find some primes!

S.

Profile rebirther
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 2 Jan 13
Posts: 7228
Credit: 42,729,227
RAC: 31
Message 491 - Posted: 3 Jan 2015, 19:04:44 UTC - in response to Message 488.

Reb,

I don't really understand the need for short deadlines. These prime numbers have been there for an awfully long time and a few days isn't going to make any difference in the long run.

Having said that, you might find fewer WUs timeout if the DCF is corrected. My last batch of long tasks came in with estimates of less than 3 hours yet have taken up to 15 hours to crunch. BOINC wouldn't have downloaded so many had the estimate been more accurate.

Then again, you may be struggling with the same problem as at PG - there isn't a sub-project DCF. Beats me how there wasn't one from the start of BOINC and it's even more incomprehensible that there still isn't.

Okey Dokey, lets find some primes!

S.


The project doesnt use DCF, its fixed with 1.

Neo
Send message
Joined: 28 Dec 14
Posts: 18
Credit: 299,879
RAC: 0
Message 495 - Posted: 3 Jan 2015, 19:22:23 UTC - in response to Message 491.

Unless you have MANUALLY started a S/R base yourself, you will never really understand why all the work has to be returned back to the server.

After the results are returned (ALL OF THEM), Rebirther has to remove all the K's from the sieve file for each K that was prime.
It's not just one K... for each K there is, for example, 20,000 w/u's for that K depending on the range of N. After that's done, the sieve file (which is the LLR candidate file) is then sieved to a deeper N in preparation for further LLR testing.

God only knows what Rebirther has to do in order to make all of this work with BOINC and to get us a steady stream of work. Which, BTW, has been increasingly more and more steady.


The process is fun, if you're doing it for the first time, and only working on one (1) base....

Rebirther: If I oversimplified the above, my apologies.

Neo
AtP

Profile Steve Hawker*
Avatar
Send message
Joined: 8 Dec 14
Posts: 26
Credit: 610,307
RAC: 0
Message 496 - Posted: 3 Jan 2015, 19:34:40 UTC

Well, if we'll never really understand then the rest of your post is redundant. Still, lets have a go, eh?

I didn't say don't return work to the server, I suggested more patience would be appreciated.

The concept is simple. Need to eliminate candidates before taking the next step. Sure, I think we all get that. But the process still doesn't demand a short deadline. What's the big hurry?

Its not as if there is some shortage of other bases or other forms to be crunched, especially now Reb has sorted out multiple bases.

The process just doesn't need short deadlines. This isn't even approaching a systemic need.

You haven't come close to explaining why a short deadline is needed.

frankhagen
Send message
Joined: 7 Jun 14
Posts: 36
Credit: 241,030
RAC: 0
Message 497 - Posted: 3 Jan 2015, 19:40:19 UTC - in response to Message 496.

You haven't come close to explaining why a short deadline is needed.


plain simple: only after the last wu of a batch has been returned, the results can and have to be evaluated manually.

after that the next batch can be prepared and started.

not that hard to understand - isn't it?

Neo
Send message
Joined: 28 Dec 14
Posts: 18
Credit: 299,879
RAC: 0
Message 498 - Posted: 3 Jan 2015, 19:43:49 UTC - in response to Message 497.

You haven't come close to explaining why a short deadline is needed.


plain simple: only after the last wu of a batch has been returned, the results can and have to be evaluated manually.

after that the next batch can be prepared and started.

not that hard to understand - isn't it?


Good summary. :)

Neo
AtP

frankhagen
Send message
Joined: 7 Jun 14
Posts: 36
Credit: 241,030
RAC: 0
Message 499 - Posted: 3 Jan 2015, 19:50:04 UTC - in response to Message 498.

i can do even better...


crunch fast or stay hungry. ;)

Buster Gunn
Avatar
Send message
Joined: 2 Dec 14
Posts: 10
Credit: 4,431,348
RAC: 0
Message 507 - Posted: 3 Jan 2015, 23:02:40 UTC

Does the server eliminate any remaining tests on the server, for primed k's like SR5 does on PG? If so then just load the ENTIRE sieve file and let the server handle the deletions.
____________

Profile Steve Hawker*
Avatar
Send message
Joined: 8 Dec 14
Posts: 26
Credit: 610,307
RAC: 0
Message 508 - Posted: 3 Jan 2015, 23:17:47 UTC - in response to Message 497.

You haven't come close to explaining why a short deadline is needed.


plain simple: only after the last wu of a batch has been returned, the results can and have to be evaluated manually.

after that the next batch can be prepared and started.

not that hard to understand - isn't it?


No, easy peasy.

Where's the bit that says it needs to be be done NOW and not tomorrow?

Profile vaughan
Send message
Joined: 9 Dec 14
Posts: 81
Credit: 2,134,081,047
RAC: 3,458,849
Message 509 - Posted: 4 Jan 2015, 3:29:41 UTC

I also don't understand why reducing the time helps us to get long running tasks back any earlier.
I have a bunch of R648 tasks that seem to be stuck at 99.999% with time remaining 00:00:00 with a running time so far of over 45 minutes.

Profile rebirther
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 2 Jan 13
Posts: 7228
Credit: 42,729,227
RAC: 31
Message 511 - Posted: 4 Jan 2015, 8:04:45 UTC - in response to Message 509.

I also don't understand why reducing the time helps us to get long running tasks back any earlier.
I have a bunch of R648 tasks that seem to be stuck at 99.999% with time remaining 00:00:00 with a running time so far of over 45 minutes.


I will not change the deadline for long runners because a prime check needs much more time.

Werinbert
Send message
Joined: 9 Dec 14
Posts: 14
Credit: 10,009,744
RAC: 0
Message 527 - Posted: 5 Jan 2015, 4:21:55 UTC

I am going to agree with Steve Hawker*. We are simply looking for stationary primes not a moving target. The primes will still be there waiting in the same place whether we find them today, tomorrow, or next week. It is not like we are doing code-breaking during war time. I thought that this project is to last for a number of years.


I will not change the deadline for long runners because a prime check needs much more time.
??? So if a prime number takes longer to process it will be discarded for taking too long?
____________
35376214*3^107944-1 is prime! (20 Dec 2014)

Lemon
Send message
Joined: 1 Dec 14
Posts: 5
Credit: 7,563,028
RAC: 0
Message 528 - Posted: 5 Jan 2015, 4:48:09 UTC - in response to Message 509.

I also don't understand why reducing the time helps us to get long running tasks back any earlier.
I have a bunch of R648 tasks that seem to be stuck at 99.999% with time remaining 00:00:00 with a running time so far of over 45 minutes.


You can look at stderr.txt in the appropriate slot folder to see what percentage of processing has been done. (And that can then be used to calculate an approximation of the remaining runtime.)

Robert H
Send message
Joined: 21 Dec 14
Posts: 4
Credit: 13,824
RAC: 0
Message 530 - Posted: 5 Jan 2015, 6:56:31 UTC
Last modified: 5 Jan 2015, 7:44:35 UTC

@rebirther:

U seem to be in a hurrry so..

Maybe u should write some information about (minimum requirements) for each app.

:-)

Profile rebirther
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 2 Jan 13
Posts: 7228
Credit: 42,729,227
RAC: 31
Message 536 - Posted: 5 Jan 2015, 18:07:58 UTC

The max success results is 50% working. The task is getting a success but getting no credits. I need to find out whats wrong.

frankhagen
Send message
Joined: 7 Jun 14
Posts: 36
Credit: 241,030
RAC: 0
Message 537 - Posted: 5 Jan 2015, 18:12:26 UTC - in response to Message 536.
Last modified: 5 Jan 2015, 18:12:37 UTC

The max success results is 50% working. The task is getting a success but getting no credits. I need to find out whats wrong.


about half of them as far as i can see.

rocket-science - you know. ;)

Profile rebirther
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 2 Jan 13
Posts: 7228
Credit: 42,729,227
RAC: 31
Message 538 - Posted: 5 Jan 2015, 18:56:33 UTC

I will increase the deadline again to 24 hours except R3 for 50-200k ranges


Post to thread

Message boards : News : Decreasing deadlines


Main page · Your account · Message boards


Copyright © 2014-2024 BOINC Confederation / rebirther