Slower and slower, then sticks at 100%

Message boards : Number crunching : Slower and slower, then sticks at 100%
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Peter Hucker

Send message
Joined: 6 Sep 11
Posts: 4
Credit: 7,533
RAC: 0
United Kingdom
Message 1063 - Posted: 1 Oct 2017, 11:12:50 UTC

I keep getting 4C tasks that say 21 minutes needed. They go well till about 99%, then go slower and slower, eventually reaching 100% and sitting like that overnight (but with task manager showing all 4 cores in full use). Often this makes the task go over the deadline.

Are these tasks broken or will they finish eventually?

If they go over the deadline, is the deadline extended if they're already running?
ID: 1063 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 734
Credit: 17,574,526
RAC: 510
Germany
Message 1065 - Posted: 1 Oct 2017, 13:25:57 UTC - in response to Message 1063.  

Let them run.
The progress indicator and remaining time is some kind of artificial intelligence of the BOINC client. The yafu application itself deosn't has any progress indication.
You will get credits up to 10 days after the deadline.
ID: 1065 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Peter Hucker

Send message
Joined: 6 Sep 11
Posts: 4
Credit: 7,533
RAC: 0
United Kingdom
Message 1066 - Posted: 1 Oct 2017, 15:04:32 UTC - in response to Message 1065.  

Thanks.
ID: 1066 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile ChristianVirtual

Send message
Joined: 5 Aug 17
Posts: 3
Credit: 10,351,404
RAC: 0
Japan
Message 1067 - Posted: 15 Oct 2017, 7:12:55 UTC

have a similar one ... but while waiting the system is not putting other task ? Have that with 8C tasks under Ubuntu with Ryzen
ID: 1067 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mikey

Send message
Joined: 13 Apr 17
Posts: 15
Credit: 14,932,922
RAC: 370
United States
Message 1068 - Posted: 15 Oct 2017, 23:18:06 UTC - in response to Message 1067.  

have a similar one ... but while waiting the system is not putting other task ? Have that with 8C tasks under Ubuntu with Ryzen


The admin said they are still running when they are done you will start new tasks, the 'finish time' is just a guess by the Boinc software and only after a few hundred or even a few thousand workunits will it be even somewhat close to reality.
ID: 1068 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile bcavnaugh
Avatar

Send message
Joined: 6 Jan 17
Posts: 4
Credit: 2,473,394
RAC: 0
United States
Message 1089 - Posted: 9 Dec 2017, 15:50:47 UTC - in response to Message 1065.  

Let them run.
The progress indicator and remaining time is some kind of artificial intelligence of the BOINC client. The yafu application itself deosn't has any progress indication.
You will get credits up to 10 days after the deadline.


So even at 100% for 20+ Hours allow them to Complete?
ID: 1089 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[AF>Le_Pommier] Jerome_C2005

Send message
Joined: 22 Oct 13
Posts: 21
Credit: 7,716,074
RAC: 4,382
Mexico
Message 1093 - Posted: 17 Dec 2017, 17:48:24 UTC

I have this WU running for less than 10mn and with less than a minute of reported CPU time, and then this one running for almost 3 days that I finally canceled : I had to suspend it / stop the computer, when I restarted it had lost all the running time and went back to 16 hours, and then I let it run for almost 2 days before I canceled it and now it says 1d18 hours running and less than 16 hours of CPU credited... this is all nonsense and very poorly managed : the cruncher needs a precise information on run time estimate and deadline, it is not boinc that "invents" runtime but the application that provides information to boinc about it's internals, we don't want some WU "working the time they want" and "deadlines that mean nothing don't worry".

Crunchers deserve a little more respect than this, I'm giving my CPU power and I pay the energy to run it, I don't like to be treated as an ignorant kid and "you don't need to understand that" kind of behavior.
ID: 1093 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 734
Credit: 17,574,526
RAC: 510
Germany
Message 1094 - Posted: 17 Dec 2017, 18:14:53 UTC - in response to Message 1093.  

Workunit runtime is not predictable. Each workunit is predicted with it's worst case. But very often it runs much faster because the different trial factoring method which are tried at the beginning. Out of this the BOINC client calculates some artificial remaining runtime.
ID: 1094 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile bcavnaugh
Avatar

Send message
Joined: 6 Jan 17
Posts: 4
Credit: 2,473,394
RAC: 0
United States
Message 1102 - Posted: 3 Feb 2018, 18:28:32 UTC
Last modified: 3 Feb 2018, 18:43:40 UTC

So I should let the below tasks run or abort them, going on a week running now?
Seems a lot of waste of power for such a Long time for this.
My Cost is now what I am looking at because the two computer are really doing nothing.

2 Each YAFU-16t v134.05 (16t) windows_x86_64 going on 6 Days now
99.99% 1 second left resets to 99.98% back to 1 Second left on one task for 4 Days now
99,94% 23 Seconds left resets the same and been doing this for over 3 Days now
http://yafu.myfirewall.org/yafu/results.php?hostid=23988&offset=0&show_names=0&state=1&appid=

1 Each YAFU-8t v134.05 (8t) windows_x86_64 going on 4 Days 11 hours now
99.99% 3 seconds left reset to 99.98% back to 3 seconds left and this has been going on for 3 days now.
http://yafu.myfirewall.org/yafu/results.php?hostid=30493&offset=0&show_names=0&state=1&appid=

Thank you

Crunching@EVGA The Number One Team in the BOINC Community. Folding@EVGA The Number One Team in the Folding@Home Community.
ID: 1102 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 734
Credit: 17,574,526
RAC: 510
Germany
Message 1104 - Posted: 3 Feb 2018, 21:39:16 UTC - in response to Message 1102.  

If the files in the slot are still updating and still consuming CPU, than the jobs are still running.
ID: 1104 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile bcavnaugh
Avatar

Send message
Joined: 6 Jan 17
Posts: 4
Credit: 2,473,394
RAC: 0
United States
Message 1105 - Posted: 4 Feb 2018, 21:38:38 UTC - in response to Message 1104.  

Thank you,
I will give them a few more days.

Crunching@EVGA The Number One Team in the BOINC Community. Folding@EVGA The Number One Team in the Folding@Home Community.
ID: 1105 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
[AF>Le_Pommier] Jerome_C2005

Send message
Joined: 22 Oct 13
Posts: 21
Credit: 7,716,074
RAC: 4,382
Mexico
Message 1109 - Posted: 10 Feb 2018, 10:26:25 UTC
Last modified: 10 Feb 2018, 10:27:57 UTC

It is very difficult to participate when you use a windows machine where boinc cannot access internet most of the time (I use a USB key once a week to move WUs to a connected win VM in my Mac).

I've had the machine exceptionally at home for a few days (thanks to the snow in Paris :) ) so I could let it run connected to the net (out of corporate network) so I decided to give another try with yafu, I got a very short WU first (3 secs of run only !) and then I have one of those long ones that already reached 100% and is running, I can see the logs are updating in the slot, apps running in memory so I know "it's OK" but what worries me is that if it doesn't finish by Monday morning (and the deadline is tomorrow afternoon), that machine will go back to the office with no connection for boinc.

Supposing the WU then finishes during the week it won't be sent back before the next week-end so who knows if it will still be considered valid and accepted by yafu...
ID: 1109 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile marsinph

Send message
Joined: 1 Apr 18
Posts: 22
Credit: 715,524
RAC: 0
Belgium
Message 1235 - Posted: 14 Aug 2018, 18:24:25 UTC - in response to Message 1109.  

Ben non Jerome !!!
Entretemps, vous l avez surement remarqué.
Le delai actuel pour renvoyer le resultat est de moins de 48 heures sur ce projet !!
De plus beaucoup de contraintes
Si vous cherchez des credits : Collatz et PrimeGrid sur GPU
Sur CPU : CitizenScienGrid (long, mais beaucoup de credit .
PrimeGrid aussi en CPU

Tout les chiffres cidessous sont basés sur ma plateforme de reference : I7-2600K-16RAM-GTX950
ID: 1235 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Chooka

Send message
Joined: 4 Mar 19
Posts: 11
Credit: 28,616,045
RAC: 0
Australia
Message 1373 - Posted: 18 Mar 2019, 7:05:03 UTC

So I have an 8t WU running which has now clocked up 16hrs but the deadline is 19/03/19 and it's been at 100% for quite some time now.
What happens if it gets past the deadline? Have I tied up 8 cores for 16hrs for nothing?
I won't be impressed if that's the case.

ID: 1373 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 734
Credit: 17,574,526
RAC: 510
Germany
Message 1374 - Posted: 18 Mar 2019, 17:16:33 UTC

You will get credits up to 5 days after the deadline.
If the process still consumes cpu, it is still running. In this case there should be also file modifications in the slot directory. At least the file nfs.dat is updated roughly every hour.

yoyo
ID: 1374 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Chooka

Send message
Joined: 4 Mar 19
Posts: 11
Credit: 28,616,045
RAC: 0
Australia
Message 1375 - Posted: 18 Mar 2019, 18:47:59 UTC - in response to Message 1374.  

Phew....thanks yoyo.
It's all good. Finished overnight. I was getting worried :)
ID: 1375 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Conan
Avatar

Send message
Joined: 5 Sep 11
Posts: 46
Credit: 7,102,856
RAC: 3,093
Australia
Message 1376 - Posted: 19 Mar 2019, 1:03:51 UTC - in response to Message 1104.  

If the files in the slot are still updating and still consuming CPU, than the jobs are still running.


Which files should be updating?

I have a small composite WU that has over 41 hours up now, using all 6 cores, but can't see any file that has updated in the Slot directory since the WU started.

Conan
ID: 1376 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile yoyo_rkn
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist

Send message
Joined: 22 Aug 11
Posts: 734
Credit: 17,574,526
RAC: 510
Germany
Message 1377 - Posted: 19 Mar 2019, 5:10:31 UTC - in response to Message 1376.  

You can oder them by time stamp and see which files are updating.
It should be factor.log and later nfs.dat.

yoyo
ID: 1377 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Conan
Avatar

Send message
Joined: 5 Sep 11
Posts: 46
Credit: 7,102,856
RAC: 3,093
Australia
Message 1378 - Posted: 19 Mar 2019, 8:47:16 UTC - in response to Message 1377.  

You can oder them by time stamp and see which files are updating.
It should be factor.log and later nfs.dat.

yoyo


OK, thanks yoyo,
but the only date stamp in the factor.log file is the original setup date and time, nothing has been added since the 16/3/19.
Even though it is using all 6 cores and has now reached over 49 Hours, I am going to abort it as it is not doing anything apparently, at least by the log file info anyway.
Sorry but this will have to be re-issued.

Conan
ID: 1378 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jozef J

Send message
Joined: 17 Apr 16
Posts: 1
Credit: 594,001
RAC: 0
Russia
Message 1379 - Posted: 24 Mar 2019, 11:07:32 UTC

hi, i have same problem , soo participate is imposible.
its horrible waste of time here soo i dont recomended this project .
ID: 1379 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : Slower and slower, then sticks at 100%




Datenschutz / Privacy Copyright © 2011-2024 Rechenkraft.net e.V. & yoyo