climateprediction.net (CPDN) home page
Thread 'When to abort 160 year coupled model'

Thread 'When to abort 160 year coupled model'

Message boards : Number crunching : When to abort 160 year coupled model
Message board moderation

To post messages, you must log in.

AuthorMessage
staffann

Send message
Joined: 23 Oct 05
Posts: 22
Credit: 526,746
RAC: 0
Message 32531 - Posted: 8 Feb 2008, 17:15:45 UTC
Last modified: 8 Feb 2008, 17:21:26 UTC

My computer has crunched two 160 year models for quite some time, but since it is only on every once in a while I\'d like to abort one of them.

My question is when it is a good time to abort it. Does it matter? Should I wait until a certain model time? The model is just past 50%.

I think I\'ve read the answer some time very long ago, but I cannot find the info now so please help.

/Staffan
ID: 32531 · Report as offensive     Reply Quote
ProfileIain Inglis

Send message
Joined: 9 Jan 07
Posts: 467
Credit: 14,549,176
RAC: 317
Message 32533 - Posted: 8 Feb 2008, 17:33:36 UTC
Last modified: 8 Feb 2008, 17:39:11 UTC

Models of the generation you\'re running trickle every model year, make a larger upload every 10 model years and a full restart dump every 40 model years.

Your two models are at 2001 and 2054 and the machine is a fast one. If you\'re going to abort one and leave the other running, then I would abort the 2001 model and let the other finish.

However, you have a dual-core machine so they will each be running at nearly full speed - so aborting one will not speed the other one up by much, in case that\'s what you were hoping.
ID: 32533 · Report as offensive     Reply Quote
staffann

Send message
Joined: 23 Oct 05
Posts: 22
Credit: 526,746
RAC: 0
Message 32538 - Posted: 8 Feb 2008, 18:32:49 UTC - in response to Message 32533.  

Models of the generation you\'re running trickle every model year, make a larger upload every 10 model years and a full restart dump every 40 model years.

Your two models are at 2001 and 2054 and the machine is a fast one. If you\'re going to abort one and leave the other running, then I would abort the 2001 model and let the other finish.

However, you have a dual-core machine so they will each be running at nearly full speed - so aborting one will not speed the other one up by much, in case that\'s what you were hoping.


The idea was indeed to abort the one at 2001. The problem is not the speed of the computer as much as the fact that I cannot leave it on. Its low uptime means that it takes forever for it to finish a model. These models have crunched since 2006 (!) and I want to be able to do some other projects as well.

Since a full restart dump is uploaded every 40 years, the model I intend to abort has only run 1 year after that. It seems to me that I might as well abort it where it is.

I\'ll finish my other model and let the other core crunch some more WCG and Rosetta workunits. I\'ll also make sure my CPDN preferences are to download only shorted models (maybe even slab only).

Thank you!

ID: 32538 · Report as offensive     Reply Quote
Profilemo.v
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 2363
Credit: 14,611,758
RAC: 0
Message 32539 - Posted: 8 Feb 2008, 20:24:53 UTC

Some unfinished models are later completed on other members\' computers from the restart dumps so the model you abort may still reach 2080!
Cpdn news
ID: 32539 · Report as offensive     Reply Quote

Message boards : Number crunching : When to abort 160 year coupled model

©2024 cpdn.org