climateprediction.net home page
Time to completion increasing

Time to completion increasing

Questions and Answers : Windows : Time to completion increasing
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user166565
Avatar

Send message
Joined: 20 Feb 06
Posts: 7
Credit: 52,516
RAC: 0
Message 21957 - Posted: 9 Apr 2006, 10:17:51 UTC

I\'ve done a search and can\'t find this problem listed. If this is a duplication, I apologise. I have two BBC models running. Whilst checking resource share yesterday, I noticed that the two models were showing larger times to completion than the previous day. One model had 19:02:09 of CPU time during the 24 hours, the other 7:45:23. But as of 10:46 BST today each is showing almost an hour more to comletion than it was the previous day. Can someone tell me what is going on, please? The message file doesn\'t show any problems, nor does it show any extra download of work.
ID: 21957 · Report as offensive     Reply Quote
Profile Andrew Hingston
Volunteer moderator

Send message
Joined: 17 Aug 04
Posts: 753
Credit: 9,804,700
RAC: 0
Message 21960 - Posted: 9 Apr 2006, 12:07:37 UTC

Experience with the 5.2 versions of BOINC suggests that they are a poor indicator of time to completion. Whether 5.3 and later will prove much better I\'m not sure but the best estimate will be the one you do yourself, taking % complete and actual CPU time as the basis.

The only reason to worry about time to completion increasing would be if the percentage complete actually fell - this can happen if your experiment hits a problem and has to go back to an earlier point, but there is nothing you can do about that.

If you are worried about how long the experiments wil take, you can set no new work in the projects tab and then suspend one of the tasks, letting the other run to completion.
ID: 21960 · Report as offensive     Reply Quote
old_user166565
Avatar

Send message
Joined: 20 Feb 06
Posts: 7
Credit: 52,516
RAC: 0
Message 21962 - Posted: 9 Apr 2006, 13:25:07 UTC - in response to Message 21960.  

Experience with the 5.2 versions of BOINC suggests that they are a poor indicator of time to completion. Whether 5.3 and later will prove much better I\'m not sure but the best estimate will be the one you do yourself, taking % complete and actual CPU time as the basis.

The only reason to worry about time to completion increasing would be if the percentage complete actually fell - this can happen if your experiment hits a problem and has to go back to an earlier point, but there is nothing you can do about that.

If you are worried about how long the experiments wil take, you can set no new work in the projects tab and then suspend one of the tasks, letting the other run to completion.



Thank you. I hadn\'t noticed (but should have) the disparity between CPU times & percentages. Using those times as a measure, my two models have 4000hrs and 4562 hrs to run (using 23 hrs per day to allow for backup etc.) instead of the 1800hrs and 1799 hrs listed. Quite a disparity!
ID: 21962 · Report as offensive     Reply Quote

Questions and Answers : Windows : Time to completion increasing

©2024 climateprediction.net