climateprediction.net (CPDN) home page
Thread 'Past Due Data'

Thread 'Past Due Data'

Message boards : Number crunching : Past Due Data
Message board moderation

To post messages, you must log in.

AuthorMessage
user

Send message
Joined: 21 Jul 09
Posts: 1
Credit: 69,696
RAC: 0
Message 41902 - Posted: 2 Apr 2011, 17:14:33 UTC

On climateprediction.net I have 2 tasks that were due 2/20/11 and still have 300+ hours to go each. Will that data still be useful? If I abort as the BOINC manager suggests will the 600+ hours already done be sent and used, or simply deleted? I care only about the data, not credits etc. Thanks for any help/advice.
ID: 41902 · Report as offensive     Reply Quote
ProfileIain Inglis
Volunteer moderator

Send message
Joined: 16 Jan 10
Posts: 1084
Credit: 7,827,799
RAC: 5,038
Message 41903 - Posted: 2 Apr 2011, 17:47:13 UTC - in response to Message 41902.  
Last modified: 2 Apr 2011, 22:59:44 UTC

The due dates are a BOINC feature that CPDN ignores, though it will affect the scheduling of tasks if you are running more than one project. CPDN accepts any data submitted after the due date and credits remain intact.

However, the two tasks are from the HADSM3 family, which do sometimes suffer from a slowdown that does render subsequent data worthless. You should check the graphics display for each of the models: if the temperature display is blue for a couple of timesteps then the model is an 'iceworld' and should be aborted. The 'u59g' model has already completed on another Windows/Intel computer, so yours should finish too. For some reason I can't get to the Web page for the other model, 'u49s', so it would be a good idea to check whether that is a possible iceworld - these slab models should take a matter of weeks rather than a year.

[Edit: Actually the second one is fine too. They're probably running slowly because there are two models running hyperthreaded on one Pentium 4.]
ID: 41903 · Report as offensive     Reply Quote

Message boards : Number crunching : Past Due Data

©2024 cpdn.org