Advanced search

Forums : General Topics : dropping to 60%
Message board moderation

To post messages, you must log in.

AuthorMessage
Paul Forsdick

Send message
Joined: 4 Apr 09
Posts: 7
Credit: 30,165
RAC: 0
Message 8779 - Posted: 20 Dec 2009, 11:12:51 UTC

I have been crunching for 26 hours and it reached 93% with 1 hr 33 minutes left now it has decided it is really 60% done and 16 hours left so this looks it will run for 42 hours for 420 points or 10 points an hour whereas most other projects are much better than this
ID: 8779 · Report as offensive     Reply Quote
Paul Forsdick

Send message
Joined: 4 Apr 09
Posts: 7
Credit: 30,165
RAC: 0
Message 8780 - Posted: 20 Dec 2009, 16:30:49 UTC - in response to Message 8779.  

now it has done 31 hours with 19 still left

I am going to abort it so 31 hours wasted

time to detach from the projrct
ID: 8780 · Report as offensive     Reply Quote
kararom

Send message
Joined: 9 Jan 09
Posts: 69
Credit: 29,506,700
RAC: 0
Message 8781 - Posted: 20 Dec 2009, 17:26:41 UTC - in response to Message 8780.  

If capacity of your computers does not suffice for reception more than 10 credit in an hour are does not say that the project is worse than others...
ID: 8781 · Report as offensive     Reply Quote
Profile Misfit
Volunteer tester
Avatar

Send message
Joined: 9 Jun 07
Posts: 150
Credit: 237,789
RAC: 0
Message 8788 - Posted: 26 Dec 2009, 22:29:11 UTC - in response to Message 8779.  

I've seen this many times before. If I recall correctly that 60% mark is a checkpoint. So for whatever reason your WU stopped and restarted it continued from there. Increased checkpointing would fix this issue.
me@rescam.org
ID: 8788 · Report as offensive     Reply Quote
.clair.

Send message
Joined: 4 Nov 07
Posts: 604
Credit: 10,881,302
RAC: 0
Message 8792 - Posted: 27 Dec 2009, 21:06:33 UTC

Because cosmo checkpoints seem to be far apart make shure
that `Leave applications in memory when suspended` is ticked for `yes`
otherwise you will loose work after a checkpoint restart.
Its one possible fix for the problem.
ID: 8792 · Report as offensive     Reply Quote
steveplanetary

Send message
Joined: 19 May 09
Posts: 18
Credit: 24,360
RAC: 0
Message 8795 - Posted: 28 Dec 2009, 8:01:50 UTC - in response to Message 8792.  

Even with leaving apps in memory I have seen C@H drop abruptly from 25 hours remaining to 7+ hours remaining, and continue from there. Go figure.
ID: 8795 · Report as offensive     Reply Quote
Profile Misfit
Volunteer tester
Avatar

Send message
Joined: 9 Jun 07
Posts: 150
Credit: 237,789
RAC: 0
Message 8800 - Posted: 30 Dec 2009, 3:42:53 UTC - in response to Message 8792.  

Because cosmo checkpoints seem to be far apart make shure
that `Leave applications in memory when suspended` is ticked for `yes`
otherwise you will loose work after a checkpoint restart.
Its one possible fix for the problem.

Now there's one way to max out your memory quickly.
me@rescam.org
ID: 8800 · Report as offensive     Reply Quote
Eugene Stemple

Send message
Joined: 8 Dec 09
Posts: 2
Credit: 264,180
RAC: 0
Message 8837 - Posted: 21 Jan 2010, 6:18:29 UTC - in response to Message 8779.  

I observed this phenomenon soon after I joined Cosmo group. Some WU's were completing o.k. but one in particular had gone through 14 suspend/resume iterations and was always resuming at the 60% progress (check)point. My dual CPU system works on Seti (66%) and Cosmo (33%) so the BOINC rescheduling algorithm inherently suspends WU's at intervals far shorter than their run times. For the WU in question the elapsed time was >28 hours when I stumbled upon the utility of the "leave app in memory" setting. After updating BOINC preferences with that option the WU finished in ~7 hours. Otherwise I believe that WU would never have finished, and certainly not before its deadline.
Yeah, it does use lots of memory (typical Cosmo 300MB to 700MB) but in my 2 GB RAM system I don't have any problems even when 2 Seti's (~40MB each) and 2 Cosmo's are either running or "waiting" to run. I presume, but have not tested it, that should more RAM be needed by no-BOINC processes their RAM contents would be saved in swap disk space and restored when needed.

ID: 8837 · Report as offensive     Reply Quote

Forums : General Topics : dropping to 60%