I was really pleased to see climateprediction.net added to the projects list recently. However, I've got my first couple of jobs for them and both are huge - the initial prediction was 250 hours, but that may be an underestimate as they are completing at about 1% for every 8-10 hours of running. Given that other projects are running as well, at that rate they probably won't meet their deadline of 9th May. It would be a shame to spend so many cycles running the climate jobs, only for them to be abandoned as late!
I don't think the machine in question is particularly slow - 3GHz, running Win7?
if these tasks are impossible to complete intime then just abort them and go to the projects tab and click the "No new Tasks" Button so you wont get any further tasks from climateprediction.net
Hi Julia, climateprediction.net is known for its very large WUs (IIRC, they're the largest of any project) but you're correct of course; no point in doing them if they're going to miss the deadline.
We'll look into it and report back...
(*We probably won't be keeping climateprediction online long.)
As Matt states the climateprediction project probably won't keep online for long, but there also is no need to abort if it won't reach it's deadline as Darkknight900 suggested.
Even though climateprediction has huge tasks it implements an incremental update of the status of the computation called trickles that report back the status of the computation over time. Credits are awarded for these trickles when they are reported in the middle of the computation instead of at the end of the task... so if you like the project you are welcome to keep it and contribute until the deadline where the task will automatcially abort.
If/When Charity Engine removes the project automatically the task will also be aborted and any trickles will still be awarded... our whole design is let us handle the management so you can install and enjoy without need to make manual changes.
Thanks for that - I had a quick look at theie website and it had already reported credit for me. So I guess it's not nearly so much of a problem. I think it's such a worthwhile project I'd like to see it kept on (but then, aren't they all?)
I thought of another solution - on the machine concerned, throttle back the network communication by time of day severely, so that CE runs out of other jobs and then has to concentrate all the cpu on the climateprediction ones.
Thanks Jonathan for the explanation this was also new to me :) Great to hear :)
Julia thank you for the suggestion. The installed program actually learns over time how often work is completed and at what rate to estimate future requests. Thus adapting to reduce over requesting work that it will not be able to complete. This scheduling mechanism is ever evolving and we have some novel ideas on how to make it even better over time. :)
Darknight900 glad to have shared something new... trickles are a great feature.