climateprediction.net (CPDN) home page
Thread 'Why can\'t to use GPU?? (Winxp)'

Thread 'Why can\'t to use GPU?? (Winxp)'

Message boards : Number crunching : Why can\'t to use GPU?? (Winxp)
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user501019

Send message
Joined: 9 Feb 08
Posts: 2
Credit: 59,769
RAC: 0
Message 38629 - Posted: 4 Jan 2010, 16:22:35 UTC

project is need a long time. What have you to use a method, who can to accelerate project?

thanks

my english is bad, sorry
ID: 38629 · 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 38631 - Posted: 4 Jan 2010, 19:39:39 UTC

Hi Skymp

You speak much better English than most of us speak Chinese!

It is not possible at the moment to accelerate the processing of ClimatePrediction models. They are not adapted for GPU; this may be possible in the future, but certainly not yet.

There are no optimised applications for CPDN models. Everyone uses the standard application from the CPDN server. The programmers don\'t want optimised applications because the researchers want stable results all processed in the same way. The researchers do not care if a model takes a day, week, month or year to process because for them the important thing is the quality of the results.

But you may be able to process models a little faster:

* HadAM3P models are shorter. You can select them in the ClimatePrediction section of your account. But I think the server has none for us at the moment.

* If HadAM3P is available, do not process a lot of them together on a multicore computer because they slow each other down.

* For all model types it\'s better NOT to run the Boinc screensaver which includes the CPDN model. The dynamic model screensaver uses CPU cycles and slows the processing of the model. Look at the graphics in Boinc manager View Graphics when you want to.

* If you have a HadSM or HadSM MH model, look at the graphics every day or two to check that your globe looks normal. A small proportion of these model types develop an \'iceworld\' problem (you would see monochrome graphics). You would need to abort the model and not waste time on it.
Cpdn news
ID: 38631 · Report as offensive     Reply Quote
old_user501019

Send message
Joined: 9 Feb 08
Posts: 2
Credit: 59,769
RAC: 0
Message 38648 - Posted: 7 Jan 2010, 16:13:03 UTC - in response to Message 38631.  
Last modified: 7 Jan 2010, 16:13:49 UTC

Can we to see results??
ID: 38648 · Report as offensive     Reply Quote
ProfileastroWX
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1496
Credit: 95,522,203
RAC: 0
Message 38649 - Posted: 7 Jan 2010, 18:56:20 UTC - in response to Message 38648.  

Can we to see results??

Is this what you\'re looking for?

http://climateprediction.net/content/experiments
http://climateprediction.net/content/results

Note that, on the right side list, is a link to publications, some derived from CPDN.

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.
ID: 38649 · Report as offensive     Reply Quote

Message boards : Number crunching : Why can\'t to use GPU?? (Winxp)

©2024 cpdn.org