Advanced search

Message boards : Multicore CPUs : Run time too long

Author Message
zombie67 [MM]
Avatar
Send message
Joined: 16 Jul 07
Posts: 207
Credit: 1,668,326,456
RAC: 4,342,873
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 215 - Posted: 18 Jul 2007 | 14:20:03 UTC

I am in the middle of crunching my first result:

http://www.ps3grid.net/PS3GRID/result.php?resultid=6002

I went to bed last night, and it was a little over 50% complete, on track to be complete just over 22 hours. That matches what I have seen with other results. I expected to see it completed by the next time I looked.

This morning, I check in on it, and it shows 25:40 elapsed time, 53% complete, and 20 more hours to go. Is this normal? Or do I have a problem?
____________
Reno, NV
Team: SETI.USA

zombie67 [MM]
Avatar
Send message
Joined: 16 Jul 07
Posts: 207
Credit: 1,668,326,456
RAC: 4,342,873
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 217 - Posted: 18 Jul 2007 | 15:00:31 UTC

Now that I confirmed we have checkpointing, I decided to reboot to see what happens with this result.

It now shows as 11:50 elapsed time, 53% complete, and just over 13 hours left to go. If I look at the elapsed time, and the percentage, it looks like it will complete in about 22 hours, which is normal, I think. But this thing has been running for over a day now.

Opinions? Comments? Suggestions?
____________
Reno, NV
Team: SETI.USA

zombie67 [MM]
Avatar
Send message
Joined: 16 Jul 07
Posts: 207
Credit: 1,668,326,456
RAC: 4,342,873
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 229 - Posted: 19 Jul 2007 | 1:49:16 UTC

Okay, that WU finally completed and returned successfully. Here is the result if anyone wants to look at it (I listed the wrong one in the the original post):

http://www.ps3grid.net/PS3GRID/result.php?resultid=6000

It says that it ran for 80261.98 seconds, or 22.3 hours. In reality, it ran for about 36 hours.


Now there is a second oddity with my second WU. It has been crunching for about 20 minutes now, but the numbers are funny:

00:20:00 elapsed time
50.625% complete
08:42:29 remaining time

Do WUs always take about 80k seconds? Or could this one really be shorter?
____________
Reno, NV
Team: SETI.USA

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 231 - Posted: 19 Jul 2007 | 7:35:24 UTC - in response to Message 229.

Okay, that WU finally completed and returned successfully. Here is the result if anyone wants to look at it (I listed the wrong one in the the original post):

Do WUs always take about 80k seconds? Or could this one really be shorter?



The workunit progression number does not work for restarted workunits.
We still have to fix this. Only time is reliable. I does not seem there was a problem with your previous workunit. Let\'s see if it happens again.

Profile Krunchin-Keith [USA]
Avatar
Send message
Joined: 17 May 07
Posts: 512
Credit: 111,288,061
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 247 - Posted: 19 Jul 2007 | 20:39:58 UTC

Thr run time should always be 22hrs and a few minutes minutes, of CPU time.

If you have other stuff running, especially a screensaver, this takes CPU time too. The run time of the work will still be 22 CPU hours, but wall time will be longer.

If yours took 36 wall clock hours, you have something else slowing things down.

Look at my account and my system, you will see the run times are realativly all close. They vary by a few minutes.

zombie67 [MM]
Avatar
Send message
Joined: 16 Jul 07
Posts: 207
Credit: 1,668,326,456
RAC: 4,342,873
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 249 - Posted: 20 Jul 2007 | 1:11:09 UTC

The second result took just over 80k seconds, both in wall clock and CPU clock. I noticed that the Time setting on the PS3 was *way* off. I am guessing that something screwy happened during the first result that caused both the delay in crunching and the Time to change.

In any case, it seems a-ok now.
____________
Reno, NV
Team: SETI.USA

Fred
Send message
Joined: 27 Jun 07
Posts: 4
Credit: 1,003,357
RAC: 0
Level
Ala
Scientific publications
wat
Message 268 - Posted: 28 Jul 2007 | 16:10:01 UTC - in response to Message 249.

The second result took just over 80k seconds, both in wall clock and CPU clock. I noticed that the Time setting on the PS3 was *way* off. I am guessing that something screwy happened during the first result that caused both the delay in crunching and the Time to change.

In any case, it seems a-ok now.


Yes, I have seen this happen. Right now, a WU has been running for 1.5 days and will soon complete. Did notice when I checked the TV that the mouse had been bumped and the view was in desktop 1,0 instead of the normal 0,0. Maybe this put the BOINC task in the background?

Post to thread

Message boards : Multicore CPUs : Run time too long

//