PDA

View Full Version : New project!! - WWF / CPDN Seasonal Attribution Project



NeoGen
03-11-2006, 02:43 AM
New Climate Prediction project people!! :D
Website: http://attribution.cpdn.org/

Let me tell you guys, this is similar in aspect, but way different than the BBC one! (And for better in my oppinion). If you tried BBC and quit because of the errors and taking too much time, you definetly have to try this one! :)
I'll tell you all about it, but first, an introduction...

The climateprediction.net Seasonal Attribution Project uses computing time donated by the general public to run state-of-the-art high-resolution model simulations of the world's climate. These simulations are used to determine the extent to which the risk of occurrence of extreme weather events is attributable to human-induced climate change.

We focus on extreme weather events that occur on a seasonal timescale, and in our current project we focus specifically on the United Kingdom floods of Autumn 2000 (http://attribution.cpdn.org/floods.php) which occurred during the wettest autumn ever recorded, causing widespread damage and an estimated insured loss of £ 1.3 billion.
This project is backed by the World Wildlife Foundation (http://www.panda.org/) (WWF) and a UK Natural Environment Research Council (http://www.nerc.ac.uk/) studentship, and based at the Atmospheric, Oceanic and Planetary Physics (http://www-atm.physics.ox.ac.uk/) department of Oxford University.

System requirements
It uses a much higher resolution model data for the calculations, which makes it much heavier than the BBC Climate Prediction. In fact, these are the advised minimum specs:

CPU: Pentium 4, 2.4GHz
RAM: 1GB
Disk space: 500MB

On a machine with these specs, the model reaches the end in about 1 month! :D Note that the 1GB of RAM is very important. It can run with 512MB too, but if you use the machine on a daily basis you will notice a considerable loss of performance.

Here's a screenshot of it running on my machine :)
http://img150.imageshack.us/img150/131/cpdnsap1ew.th.png (http://img150.imageshack.us/my.php?image=cpdnsap1ew.png)
That was taken before the actual calculations started. Don't think it runs at 2.70 s/TS, it's far from that really.:lol:
The high resolution model takes its toll so much that it runs at around 20 s/TS really on a decent machine.
But don't let that scare you! The full simulation only takes 51,984 Timesteps. (Unlike BBC that takes hundreds of thousands)

Checkpoints
Checkpointing happens every 144 Timesteps (1 model days). At an average speed of 20 s/TS that means it saves a checkpoint every 48 minutes. That's the maximum of work you might lose in case of shutdown. :)

Trickles
Trickles are sent every 720 Timesteps (5 model days). At a rate of 20 s/TS that's about 4 hours. Each trickle gives 72 credits I think, so in a full day (24 hours) you should get 432 credits for each model you're running. :D

How to join
Attach through the boinc wizard, using the site provided on the top of the post. Detailed instructions can be found here (http://attribution.cpdn.org/join.php)
And afterwards, don't forget to join our Team here (http://attribution.cpdn.org/team_display.php?teamid=16) ;)

A note from the project admins

If you are already running one of climateprediction.net's experiments, or the new BBC Climate Change Experiment, then PLEASE allow your current experiment to finish before starting your Seasonal Attribution experiment.

Have fun!! :D

drezha
03-11-2006, 10:12 AM
Are you sure thats right an average os 20s/TS? :?

Because I might be doing something very wrong...
My old XP used to run about 3.5s/TS (before the sulphur cycle)
My A64 3500+ runs thesulphur cycle with about 3.5s/TS as well :shock:

NeoGen
03-11-2006, 11:27 AM
Yep, I'm sure. :)
Check this host for example:
http://attribution.cpdn.org/show_host_detail.php?hostid=73

But it is not bad because the whole simulation takes very few steps compared to other projects.

drezha
03-11-2006, 11:33 AM
:shock: :D

Thats not bad. Unfortunatly the large requirements put's me off. This is my main rig...I'd rather not have it slowing down :-(

NeoGen
03-11-2006, 11:41 AM
You mean, you only have 512Megs of RAM in the machine, right?

I don't advise to run it either on less than 1Gb of RAM, unless you don't use that machine for anything else. (a pure cruncher) :roll:

drezha
03-11-2006, 01:15 PM
No. I have a gig but I'd rather not want it slowing down at all really ;)

Thats why I stopped Predictor many moons ago..because the thing would slow my machine down no end with no reason...

vaughan
03-16-2006, 06:53 AM
Bok's stats:

http://stats.free-dc.org/new/projpage.php?proj=att

Ototero
03-16-2006, 11:33 AM
I'll add it to the stats :D