climateprediction.net home page
CPDN badly behaved

CPDN badly behaved

Questions and Answers : Unix/Linux : CPDN badly behaved
Message board moderation

To post messages, you must log in.

AuthorMessage
Richard Kimber

Send message
Joined: 17 Oct 06
Posts: 5
Credit: 209,102
RAC: 0
Message 42227 - Posted: 21 May 2011, 14:37:21 UTC

I'm running CPDN and other projects on Ubuntu 11.04 with Boinc 6.10.59. The machine is AMD Phenom II x4 945.

My preferences are that up to three CPUs may be used by Boinc projects.

I have just had a CPDN work unit that wanted to run high priority. By doing this it took over the entire Boinc allocation, presumably for the next couple of months, but only ran in one CPU, thus preventing other projects from working at all and leaving two CPUs unused.

Is there a way of getting CPDN to work in a more friendly fashion? I have aborted the work unit concerned.
ID: 42227 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 42232 - Posted: 21 May 2011, 20:14:05 UTC - in response to Message 42227.  

There are two ways to "get cpdn to behave":

1) Run the project on a computer by itself. Causes split cpids, and therefore credits.
2) Stop being a back seat driver, and just let BOINC get on with it and learn how to run a mix of projects with mostly very short completion times, along with the climate models with their very long completion times.

Apparently BOINC needs, (or used to need), to complete several work units of a given type to 'learn' the amount of time it takes for each, given the user's computer usage habits.
I think that the latest versions 'learn' faster, due to different ways of using the various estimate numbers provided by projects with each work unit.

But it will still take a week or so, without constant fiddling of BOINC settings, for it to 'learn' the actual time-to-completion of climate models. This value should decrease fairly quickly, and when BOINC is confident about the new mix of models, it'll drop out high priority and go back to round robin processing.

But if this isn't suitable for you, the best thing to due is to NOT run climate models. Aborting them just wastes them.


Backups: Here
ID: 42232 · Report as offensive     Reply Quote
Profile astroWX
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1496
Credit: 95,522,203
RAC: 0
Message 42233 - Posted: 22 May 2011, 4:45:32 UTC - in response to Message 42227.  

I'm running CPDN and other projects on Ubuntu 11.04 with Boinc 6.10.59. The machine is AMD Phenom II x4 945.

My preferences are that up to three CPUs may be used by Boinc projects.

I have just had a CPDN work unit that wanted to run high priority. By doing this it took over the entire Boinc allocation, presumably for the next couple of months, but only ran in one CPU, thus preventing other projects from working at all and leaving two CPUs unused.

Is there a way of getting CPDN to work in a more friendly fashion? I have aborted the work unit concerned.

To help us understand your complaint, please list the resource share assigned to your projects.

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.
ID: 42233 · Report as offensive     Reply Quote

Questions and Answers : Unix/Linux : CPDN badly behaved

©2024 climateprediction.net