climateprediction.net home page
1424 hours to completion by Dec 2006

1424 hours to completion by Dec 2006

Questions and Answers : Preferences : 1424 hours to completion by Dec 2006
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user147411

Send message
Joined: 9 Jan 06
Posts: 1
Credit: 1,856
RAC: 0
Message 22300 - Posted: 22 Apr 2006, 15:21:58 UTC

My sulphur run is showing 1424 hours to completion by Dec 2006. There is no way I can dedicate 7 hours a day for the rest of this year to CPDN.

Should I simply kill it?

BOINC believes my pc is over committed and so isnt scheduling my seti and einstein projects.
ID: 22300 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 22304 - Posted: 22 Apr 2006, 16:01:03 UTC

Mike

The bad news:
Sulphur is now obsolete, as the project has moved on.
Sulphur models were a short, temporary, (4-5 month) project to get sulphate data for the start of the new models, TCM.

The \"good\" news:
1) Your model was probably one of a bad batch issued in December, that crash after the end of phase 1. (After 24 trickles).

2) There is extra data uploaded at the end of phase 1 that is /was very useful.

As you\'re halfway there, I\'d suggest that you let it continue until then.
Unless you\'re only giving cpdn a small amount of time, in which case, it may take too long.
The new models are a single phase of 160 model years.


ID: 22304 · Report as offensive     Reply Quote
ds50

Send message
Joined: 24 Nov 05
Posts: 30
Credit: 893,671
RAC: 0
Message 22353 - Posted: 24 Apr 2006, 8:25:41 UTC - in response to Message 22304.  



The bad news:
Sulphur is now obsolete, as the project has moved on.
Sulphur models were a short, temporary, (4-5 month) project to get sulphate data for the start of the new models, TCM.


Hi Les,

What do you mean exactly by \"obsolete\"? My sulphur model on my AMD64 3000+ will finish this week, but I\'ve got an other model running on my Athlon XP 1700+ at a progress of 57% done (almost end of phase 3) and is remaining approx. 1200h (approx. 6,89 s/TS).
Would you suggest to abort that model? It would somehow break my heart after 1414h of crunching without any problems... What scientistic value (if any??) would the final result have?

I\'m confused a bit...

(BTW: I know this is the wrong topic in the wrong forum, but your last post meets my interest)
ID: 22353 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 22355 - Posted: 24 Apr 2006, 9:11:01 UTC

Hmmm. Perhaps \'obsolete\' isn\'t the right word.

The sulphur models were a short term, (4-5 months starting late last year), project to obtain values of sulphate emmisions to include in the startup of Coupled Ocean models. The TCMs have now started, so there isn\'t much point in a suphur model that won\'t complete until next year. One that finishes in a month or two, can probably be justified on the grounds of \'just finishing it for the sake of finishing it\'.

What the value is of any models finishing, say, \"after now\", is something only the researchers know.
Sorry that I can\'t be more specific, but I just don\'t know. I guess that it\'s a matter of how people feel about using time and electricity on something that may or may not be usefull.

ID: 22355 · Report as offensive     Reply Quote
old_user186769

Send message
Joined: 14 May 06
Posts: 2
Credit: 15,582
RAC: 0
Message 22791 - Posted: 17 May 2006, 18:53:54 UTC - in response to Message 22304.  

I guess it doesn\'t matter what I run, if it will be obsolete in 10 months anyway.

The maximum time for climateprediction.net seems brutal compared to the other climate applications.

I get only 7 months to complete a 1683 hour workunit.

BBC Climate Change experiment gives me 11 months to complete 1610 hours.
CPDN Seasonal Attribution gives me 11 months to complete a 389 hour workunit.


ID: 22791 · Report as offensive     Reply Quote
old_user186769

Send message
Joined: 14 May 06
Posts: 2
Credit: 15,582
RAC: 0
Message 22792 - Posted: 17 May 2006, 19:13:16 UTC

If your computer is overcommitted, it might be worth going to the project screen to tell some projects not to accept new work. I have 4 projects running on this computer. Three will not accept new tasks, including this climate project. Two have sent things in and are empty. Tasks come in now and then for the remaining project. The computer seems to have stabilized, and dedicates a stable portion of time to each. I haven\'t been doing this enough to see how BOINC allocates multiple projects, but noticed that a lot of short workunits kept coming in and were running, while the climateprediction just sat there.

I noted elsewhere that climateprediction only gives me 7 months to complete a 1683 hour workunit. That means that I must dedicate a fairly large percentage of each day to make sure that climateprediction finishes. I\'m giving it 9.6 hours/day now (40% of my computer time), which means that it will take 175 days to run and has to finish in about 210 days. New tasks come in now and then for the other project. I\'ll check the percentages in another month to see whether I need to adjust anything.

Most of my effort is going to Vulture Central III at worldcommunitygrid.org, but my credit here will go to Vulture Central XVII.
ID: 22792 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 22800 - Posted: 17 May 2006, 21:33:30 UTC

retsof

The time limit only applies to other projects. Here at cpdn, the deadline is there because SOMETHING had to be there, but it\'s ignored.
As long as you trickle every week or so to let the server know that the model is still \'alive\', it will be OK.
Provided that it doesn\'t take, say, 5 years to complete, as the project may have finished by then. :)

ID: 22800 · Report as offensive     Reply Quote
old_user193179

Send message
Joined: 18 Jul 06
Posts: 6
Credit: 1,089
RAC: 0
Message 23683 - Posted: 21 Jul 2006, 7:11:39 UTC

I had a related question: my climate model will also require an average of seven hours per day to complete (in 2008), and there is no way I\'ll be able to complete it.

My general question: would partially-completed data and computations still be useful to the overall project? If so, I\'ll continue letting it run to the end of a model year. Otherwise, I\'ll find another project that my system can actually complete.
ID: 23683 · Report as offensive     Reply Quote
Profile MikeMarsUK
Volunteer moderator
Avatar

Send message
Joined: 13 Jan 06
Posts: 1498
Credit: 15,613,038
RAC: 0
Message 23684 - Posted: 21 Jul 2006, 7:22:05 UTC

Ideally the best time to abort a model is at the end of a model decade (you\'ll notice a data upload occurs then). As Les said, the deadlines are ignored with this project.
I'm a volunteer and my views are my own.
News and Announcements and FAQ
ID: 23684 · Report as offensive     Reply Quote
old_user193179

Send message
Joined: 18 Jul 06
Posts: 6
Credit: 1,089
RAC: 0
Message 23708 - Posted: 22 Jul 2006, 15:31:51 UTC - in response to Message 23684.  

Thanks..That\'s useful to know...two related questions:

* how long is a model \"decade\"? Same as a calendar decade, or a different unit
of time?
* if deadlines are currently being \"ignored\", does that mean that they can
still be worked on after the completion date referenced in the \"Tasks\" tab?
ID: 23708 · Report as offensive     Reply Quote
Profile MikeMarsUK
Volunteer moderator
Avatar

Send message
Joined: 13 Jan 06
Posts: 1498
Credit: 15,613,038
RAC: 0
Message 23709 - Posted: 22 Jul 2006, 18:05:13 UTC

If you look in the messages tab, you\'ll notice at the end of each 10 years modelled, an upload takes place. So if you can wait until you see this upload in the log, that would be the best thing to do (it\'s roughly 3/dec/1930, 1940, ... etc).
I'm a volunteer and my views are my own.
News and Announcements and FAQ
ID: 23709 · Report as offensive     Reply Quote

Questions and Answers : Preferences : 1424 hours to completion by Dec 2006

©2024 cpdn.org