Long run Times ... !!!
26 Jan 2005 10:40:22 UTC
Topic 187360
(moderation:
The last 2 WU's I turned in on this Computer (Host 5404) took it about 14:30:00 when all the previous ones only took between 10:35:00 & 10:50:00 to run on that computer. The 2 WU's that it is running right now seem to be running the normal run time again. Could this have been a problem with the xml file not saving the the times right when I stop BOINC yesterday to do the upgrade to version 4.18 & v4.19 ... ???
The same thing happened on another computer of mine a few days ago but the WU"s have been granted credit already & wiped from the Database so I can't find them any more ...
1018753 313673 22 Jan 2005 15:01:44 UTC 26 Jan 2005 10:24:30 UTC Over Success Done 52,346.67 122.57 pending
1018748 313672 22 Jan 2005 15:01:38 UTC 26 Jan 2005 10:24:30 UTC Over Success Done 52,228.03 122.29 pending
1018744 313671 22 Jan 2005 15:01:32 UTC 25 Jan 2005 14:02:04 UTC Over Success Done 33,846.72 78.77 pending
1018736 313669 22 Jan 2005 15:01:26 UTC 25 Jan 2005 14:02:04 UTC Over Success Done 33,622.17 78.25 pending
1018727 313667 22 Jan 2005 15:01:20 UTC 25 Jan 2005 11:25:55 UTC Over Success Done 33,852.39 78.79 pending
1018723 313666 22 Jan 2005 15:01:14 UTC 25 Jan 2005 11:25:55 UTC Over Success Done 33,647.02 78.31 pending
1017186 313373 21 Jan 2005 15:44:48 UTC 25 Jan 2005 0:49:37 UTC Over Success Done 33,825.05 78.72 pending
1017182 313372 21 Jan 2005 15:44:42 UTC 25 Jan 2005 0:49:37 UTC Over Success Done 33,724.00 78.49 pending
1017178 313371 21 Jan 2005 15:44:36 UTC 24 Jan 2005 9:30:33 UTC Over Success Done 33,738.61 78.33 pending
1017174 313370 21 Jan 2005 15:44:36 UTC 24 Jan 2005 9:19:38 UTC Over Success Done 33,725.94 78.31 pending
1017170 313369 21 Jan 2005 15:44:30 UTC 24 Jan 2005 0:07:48 UTC Over Success Done 33,750.92 78.36 pending
1017166 313368 21 Jan 2005 15:44:24 UTC 23 Jan 2005 23:57:02 UTC Over Success Done 33,734.88 78.33 pending
1017162 313367 21 Jan 2005 15:44:18 UTC 23 Jan 2005 14:44:31 UTC Over Success Done 33,907.98 78.73 pending
Long run Times ... !!!
)
Please report the name of the Result/ Workunit. It might be one of those causing the problem described in the thread
WU not finishing
BM
BM
Hi Bernd, these are the 2
)
Hi Bernd, these are the 2 WU's that took a longer time than normal for that Computer ...
H1_0133.4__0133.5_0.1_T16_Test02_0
H1_0133.4__0133.9_0.1_T15_Test02_0
The WU's finished but just took about 4 hr's longer than it should have for them to do so ...
>
)
> H1_0133.4__0133.5_0.1_T16_Test02_0
> H1_0133.4__0133.9_0.1_T15_Test02_0
Thanks. These don't look suspicious to me.
> The WU's finished but just took about 4 hr's longer than it should have for
> them to do so ...
Well then - this might happen. The actual CPU time varies between the WUs in a way which is hard to predict, and it also depends on the state of the system it is running on (memory usage, other processes running (software updates, virus scanners etc)), so I would consider this to be within normal tolerances. Maybe we should raise the estimated times a bit.
BM
BM
Ok Bernd, I had a hard time
)
Ok Bernd, I had a hard time convincing myself to let the WU's finish running, I almost just reset the Project on that computer several times because at one point the WU's were telling me I had 60 hr's left to run them...D'oh
But I kept watching the times and they were dropping 30 sec's each time for 1 second of progress so I just figured they were like the other 2 I had a few days ago that took about 14 hr's to run ...
@ Bernd, ok I've got a couple
)
@ Bernd, ok I've got a couple more WU's that are running on the same computer that are acting real weird. The one WU got down to about 20 min's to go and
that was 1 hr & 30 min's ago and it's still showing 13 min left to run, for every 1 min of CPU Time it's only dropping 5 sec's off the completion time.
The other work unit is 1:31:00 into the run time & the completion time is telling me I have 62 hr & 13 min's left to run. It's been that way for an hour
now...
This computer normally run the Einstein WU's in about 9:30:00 time in HT Mode, in other words running 2 at a time. The computer is used for nothing but
crunching the WU's, no other activity is being used on the computer. The computer has the same Ram & Motherboard & CPU as several other Computers of mine that also crunch the WU's in about 9:30:00 time and I'm having no problems with them when it comes to taking to much time to run the WU's.
I checked the Task Manager for Windows and it shows 99% to 100% CPU usage for the 2 WU's, no other Processes are showing any activity in the Task Manager.
To be honest with you I think I have downloaded a bunch of goofy WU's onto this computer & I'm about to just reset it and shit can them & get some fresh WU's.
14 hr's of run time is just to long to be running the WU's in my opinion to be only getting a small amount of credit for them. It's just a waste of CPU time
as far as I'm concerned ...
This all seemed to start on this computer after I Upgraded to the new v4.19 so I don't know if it's the WU's or the new version causing the problems ...
PS: I for the heck of it I shut BOINC off and restarted it & the Times are dropping like flies now on both WU's now. The one should be done shortly & I'll
post it...H1_0133.4__0133.7_0.1_T16_Test02
I don't know how long the one one that was telling me it had over 60 hours left is going to take but I will post that WU later tonight for you to look at
again. I have to go to work later so I'm going to shut off Network access to that computer so it doesn't finish & upload it until I get back from work if it's even done by then yet.
2 things I would like to
)
2 things I would like to comment:
I have watched similar situations on an old WIN95-machine with PP@H in the past. Normally this happenes when the box is running a lot of days without restart. A restart solves this problem normally.
With BOINC-View, I can control all my results over a long time ago. I checked all E@H-results of the last 10 days, there is a difference between WUs on the same machine with round about 10%; this seems to be a normal factor.
All my boxes run BOIN 4.15 ...
Supporting BOINC, a great concept !
Please see the 'front page'
)
Please see the 'front page' news item that I posted. It's relevant to this thread.
Bruce
Director, Einstein@Home
OK Bruce, looks like you
)
OK Bruce, looks like you found the problem and hopefully it gets nipped in the bud shortly ... :)
3 of my AMD based machines
)
3 of my AMD based machines all take longer to process a EH w/u than predicted. I've been noticing this for a week or so. 2 of them are twin AMD 2000+ clocked as AMD 2600s, both with 256M of memory under XP Pro SP2. These two show 5:47 to complete before they start but take 8~9 hours to complete. These two are dedicated basket crunchers.
The 3rd machine is an AMD 2500+ clocked as a 3200+ with 1G of memory, same OS. Before stated an EH w/u says 5:26.43 to complete but they take nearly 6:30 or about 1 hour over the prediction to complete.
Due to this I have 1 w/u that is currently pushing the deadline to complete.
I just wanted to let ya'll know what I'm seeing. If this is the same problem as posted on the front page great... if not... oh well.
Skip
PS: I will take a look at my P4 1.6 Intel laptop later today and see if it's following this pattern or not. Also... all running 4.19