climateprediction.net (CPDN) home page
Thread 'Time remaining and Percentage going backwards.'

Thread 'Time remaining and Percentage going backwards.'

Message boards : Number crunching : Time remaining and Percentage going backwards.
Message board moderation

To post messages, you must log in.

AuthorMessage
BugG

Send message
Joined: 24 Aug 08
Posts: 12
Credit: 663,890
RAC: 0
Message 36135 - Posted: 14 Feb 2009, 5:52:42 UTC
Last modified: 14 Feb 2009, 6:01:59 UTC

I noticed strange computing of hadsm3mh_kl5c_006010988_7、 Time remaining increasing and Percentage decreasing.
I am not sure but it occurred at about 78% point and after I had restarted BOINC.
Elapsed time is normally increasing and no error message is read.
Thanks.
ID: 36135 · Report as offensive     Reply Quote
ProfileIain Inglis

Send message
Joined: 9 Jan 07
Posts: 467
Credit: 14,549,176
RAC: 317
Message 36136 - Posted: 14 Feb 2009, 9:41:31 UTC

It looks rather like a Windows/Intel \"iceworld\", which slows down dramatically and shows a uniformly blue appearance in the temperature graphic. Some iceworlds eventually make progress but at a very slow rate; others just loop indefinitely.

There is another Windows/Intel model in the same work unit possibly stuck at the same point.

If the model is indeed an iceworld, then the usual advice is to abort it, which will be rather frustrating in your case as the model had made a lot of progress.

There is more discussion about this issue here.
ID: 36136 · Report as offensive     Reply Quote
BugG

Send message
Joined: 24 Aug 08
Posts: 12
Credit: 663,890
RAC: 0
Message 36145 - Posted: 14 Feb 2009, 23:22:34 UTC - in response to Message 36136.  

Hi Iain. Thank you for your quick response.

I mend to say about the percentage. It was not 78% but 84%.
There are new symptoms. Both Time remaining and Percentage are fluctuating.
There is BOINC logo but no graphics shown. I do not know when it disappeared.
I will abort the model. The pity is that only 16% are left to completion.

Thank you again.

ID: 36145 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 36146 - Posted: 15 Feb 2009, 0:30:14 UTC

Bearing in mind that the aim of these models is to see how far they run using the starting values for their parameters, whatever percentage they reach when they fail is OK. It all adds to the knowledge of what leads to a long time stable model.


Backups: Here
ID: 36146 · Report as offensive     Reply Quote

Message boards : Number crunching : Time remaining and Percentage going backwards.

©2024 cpdn.org