climateprediction.net home page
Unusual report deadlines

Unusual report deadlines

Questions and Answers : Windows : Unusual report deadlines
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Duane Marble

Send message
Joined: 31 Jan 05
Posts: 3
Credit: 12,789,788
RAC: 0
Message 31728 - Posted: 15 Dec 2007, 19:51:18 UTC

I am currently running two HadCM3 coupled models. Both
are showing report deadlines of August 2012. The deadlines have always been greater than the actual time required but these seem too distant to be real.
ID: 31728 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 31729 - Posted: 15 Dec 2007, 20:41:20 UTC


This deadline was set because of a number of complaints from people who run lots of projects, and said that 1 year was too short.

However, it turns out that there is a limitation in \'earlyish\' versions of BOINC which resets deadlines over 1000 days to (I think), 2002.
So the latest batch of models is set for less than 1000 days.


Backups: Here
ID: 31729 · Report as offensive     Reply Quote
Profile astroWX
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1496
Credit: 95,522,203
RAC: 0
Message 31730 - Posted: 15 Dec 2007, 20:47:50 UTC
Last modified: 15 Dec 2007, 20:48:48 UTC

They were. The latest batch, of which I started one yesterday, has a deadline of June 2009.

The deadlines are long for a couple of reasons (at least):
- Accommodate slower machines that aren\'t run continuously
- Keep CPDN from going into boinc high priority and hogging the machine while boinc tries to meet the deadline (EDF mode). BOINC works it all out in the long term but folks tend to get bent out of shape when it happens.

CPDN doesn\'t care about the deadline and only uses it to satisfy boinc; boinc uses it for other Projects.


Edit: I come in a distant second again, Les!

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.
ID: 31730 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 31731 - Posted: 15 Dec 2007, 22:47:43 UTC

Not to worry. You covered other ground that I only thought of after posting. (EDF)

ID: 31731 · Report as offensive     Reply Quote
old_user116529

Send message
Joined: 25 Nov 05
Posts: 1
Credit: 140,178
RAC: 0
Message 31928 - Posted: 31 Dec 2007, 16:06:05 UTC - in response to Message 31730.  

They were. The latest batch, of which I started one yesterday, has a deadline of June 2009.

The deadlines are long for a couple of reasons (at least):
- Accommodate slower machines that aren\'t run continuously
- Keep CPDN from going into boinc high priority and hogging the machine while boinc tries to meet the deadline (EDF mode). BOINC works it all out in the long term but folks tend to get bent out of shape when it happens.

CPDN doesn\'t care about the deadline and only uses it to satisfy boinc; boinc uses it for other Projects.


Edit: I come in a distant second again, Les!


Why are the packets so large?
Seti and others use packets which take some hours of calculation and so they are shure, to get the results back.
I have one climate project running on a 3.2GHz Intel, deadline is over and so your project now looses 962 hour of calculation at 29.475% finished on my machine. Deadline was not even a year.

For this reason, I am NOT able and willing TO SUPPORT CPDN PROJECT IN THE FUTURE.

CPDN IS A WASTE OF CPU POWER!!!





ID: 31928 · 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 31930 - Posted: 31 Dec 2007, 17:10:01 UTC
Last modified: 31 Dec 2007, 17:56:51 UTC


As stated above (in the post you quoted), deadlines are not enforced here, so you don\'t need to worry about tasks which have gone over the deadline. As also stated above it may however affect your scheduling of other Boinc tasks if the Boinc manager thinks the deadline is under threat.


The climate models send their results back piece-by-piece (via \'trickles\'), so the project gets useful climate data from models which only part-complete. Hence no time lost.


There are three different types of climate model available, which take varying amounts of time and resources. Details in the \'running the model\' readme. The README posts are well worth reading (the link is in my signature).


What model are you referring to? The only one I can see in your computer listings which have done any significant processing is this one on an Athlon XP dating back to 25th November 2005, and is about 40% complete, and 238 hours of CPU time? The last trickle uploaded from it was dated 21st December, also 2005. If it had run to completion I estimate it would have been finished near the end of Jan 2006 at just under 600 hours CPU time, with ten months (80%) to spare on the deadline.


http://climateapps2.oucs.ox.ac.uk/cpdnboinc/result.php?resultid=1287533

Other than that, there are four started a couple of days ago on a Xeon 2.8GHz which wouldn\'t have had time to accrue any significant processing time. I don\'t see a 3.2GHz intel box in your computer list.

I'm a volunteer and my views are my own.
News and Announcements and FAQ
ID: 31930 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 31931 - Posted: 31 Dec 2007, 19:37:42 UTC


Why are the packets so large?

Because they\'re climate models.
If they were as short as you wanted, they would only have time to be weather models. Even the Met Office\'s supercomputers take nearly a day to run these big models. And these big models are what the researchers want, to explore the distant possibilities.


ID: 31931 · Report as offensive     Reply Quote

Questions and Answers : Windows : Unusual report deadlines

©2024 cpdn.org