Hi all,i\'m new in climateprediction! I have a problem,to complete my task BOINC say me 3232:29:13 hh:mm:ss.
Is this a PC problem or what? Thank to all

---

No, your computer is correct. The models crunches from 4 months to a year. I have models running with times like your to one that says 7740 hours to completion. Crunch away...Don\'t forget to read about backing up the results.

It\'s also worth noting that deadlines aren\'t enforced here. As long as you keep trickling, your model will be considered active even if it goes over the deadline.

I want to say that you need a trickle every 6 weeks to be considered active, but I could be wrong.

---

Kathryn :o)
The BOINC FAQ Service
The Unofficial BOINC Wiki
The Trac System
More BOINC information than you can shake a stick of RAM at.

Ciao Utente, benvenuto al progetto.

The essential cpdn info for newbies is in the project READMEs. Get to them thro the link in my sig. Don\'t try to read it all!

*In the README about running the model, look at the top tips.

*In the README about avoiding model crashes, I recommend item #5 by Mike and item #5 by Les (how to back up regularly).

*If you want a more sophisticated backup method, there\'s a whole backup README offering a selection.


Edit:

I\'ve now looked at your computer details and see a problem. You have a fast dual-core computer but it only has 512Mb memory/RAM. You need 512Mb for ONE climate model. If you simultaneously process cpdn and a second workunit from a different project, the memory may not be sufficient for the two workunits (plus you using the computer). The climate model could crash.

As a precaution you should make backups. You should also consider adding extra memory to your computer. It\'s not very expensive and you can add it yourself.

Don\'t abort models - you\'ll have to start from zero with a new model. And don\'t reset the project - this would also abort your model.

---

Cpdn news

Ciao Utente, benvenuto al progetto.

The essential cpdn info for newbies is in the project READMEs. Get to them thro the link in my sig. Don\'t try to read it all!

*In the README about running the model, look at the top tips.

*In the README about avoiding model crashes, I recommend item #5 by Mike and item #5 by Les (how to back up regularly).

*If you want a more sophisticated backup method, there\'s a whole backup README offering a selection.


Edit:

I\'ve now looked at your computer details and see a problem. You have a fast dual-core computer but it only has 512Mb memory/RAM. You need 512Mb for ONE climate model. If you simultaneously process cpdn and a second workunit from a different project, the memory may not be sufficient for the two workunits (plus you using the computer). The climate model could crash.

As a precaution you should make backups. You should also consider adding extra memory to your computer. It\'s not very expensive and you can add it yourself.

Don\'t abort models - you\'ll have to start from zero with a new model. And don\'t reset the project - this would also abort your model.

Ciao e grazie del benvenuto!

I know i have only 512 MB ram and i\'m thinking to upgrade to 1GB,but for now i have to use that memory. i didin\'t know about crashes,i hope climateprediction won\'t crash but i\'m going to read about backups for security and your guide for newbies! thank you

These hours to completion are completely ridiculous! I have to crunch for another 2453 hours. Forget it. Did the scientist who make the models ever consider that people want some kind of satisfaction? We (as I speak for all crunchers) want to have the feeling that we are actual helping and making progress. This way I only think that I\'m burning kWh\'s for warming up the earth even more. I quit.

Can\'t you split up the models, and make it a more distributed experience?

I think cpdn including BBC members is the second biggest boinc project after SETI, so a lot of members are obviously finding that these massive workunits are satisfying as well as challenging.

The question of sharing the climate models between different members has been considered by the programmers on a number of occasions, but at the moment at least, it\'s just not possible.

There\'s a lot of support on the forum and in the project READMEs to help members actually complete the models. In some ways, the harder the task, the more satisfying it is when you succeed.

The Oxford research team has already made a large collection of completed models freely available to any researchers worldwide, even if they haven\'t got the resources to run climate models themselves. So running the models can be very rewarding.

Fortunately the cpdn servers don\'t enforce the completion deadline, so if you run over the theoretical time limit your work will still be accepted and used.

---

Cpdn news

You get credits during crunching, unlike other projects, so you do see your work doing something. It\'s not like you have to wait the whole 2500+ hours to get credited, etc. Each year of work provides data to the team, and the further along they go, the better the information.

---

I get satisfaction from the fact that it is a big job. My P4 2.5 processor is taking nearly 3900 hours. I started in November 2006 and I’m looking forward to finishing this work unit some time next month. I do feel that I am helping and making progress. The whole point is that the project is big. The processing requirement for the entire cpnd project is huge, but there are a lot of us and I feel that I am doing something worthwhile if I can give the researchers the data they need to predict the climate into the future. Together, the processing power that we are making available to the project is massive. There is satisfaction from the long runs and every decade a data file is uploaded with the results of the last ten year’s modelling. This is the data that the researchers need, so view it as 16 ten-year runs.

Andy

hendrik, pity you went off the deep end without checking the water\'s depth and then not attempting to swim. It\'s obvious you didn\'t look into the Project first. Unfortunate.

No satisfaction? Not in running a million-plus-line Fortran super-computer Model ported to run on PCs?
Not in running an atmosphere and ocean Model for 160 model-years, at 30-second time slices, for 19 levels of atmosphere times 73x96 global grid squares? For ~22 levels of ocean time-sliced each minute? Where every iteration depends on what went before?

How would you break it down? How would you handle the enormous communication load? How would you handle maintaining consistency for CPU types for each unique Model [to maintain computational integrity]?

If your thing is huge numbers of completed work Units, this is not the place for you.
If you want credits, this Project awards them as the Run progresses, 160 times in all. It\'s as lucrative as any, at least theoretically, because all boinc Projects are supposed to award equal credit for equal work.

I, for one, derive satisfaction from contributing to a worthwhile undertaking, and from helping as many Work Units through to completion as possible. If you look at the \"Top Participants\" list (link on left of page) you\'ll see that many participants apparently derive enough satisfactions to make continuing major contributions -- and that sentiment surely pervades on down the list.

Hope you reconsider. There\'s plenty of information in the Sticky ReadMe section, and willing participants here to help when other issues arise.

[Edited for typo.]

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

Hi, I have a similar problem/question as Mattia.
I am new as well and started running climateprediction on a quite new PC. But then I also added my older laptop and it wouldn\'t download a slab model (although at the time I had specified that in preferences). Finally I let it download a combined model, but I\'m afraid that it\'s too big for that computer (says only 383.47 MB of RAM). Now should I cancel the project or let it run for more than a year? Due to heat problems I can only let it use about 45% of CPU time. It takes about an hour to complete 0.01% of the task.
So is that of any use, or not?

---

You\'d definitely be better off with a slab HADSM model. They take less than half the computing time and can manage with less RAM. I\'ve checked the server status page and plenty of slab models are available for download. I think the best thing will be to abort the HADCM you\'ve started.

But before you do abort it, we need to work out why you didn\'t get the type of model you wanted. Check your cpdn preferences page again in your account to make sure only HADSM is selected. If you have to change what it says, make sure that at the bottom of the preferences page you click the Update Preferences button.

Then, on the old laptop, open boinc manager and click the Update button while you\'re connected to the internet and you have boinc network activity allowed. This should tell the server to apply your preferences to this laptop.

What I don\'t know is whether when you\'ve clicked Update the effect will be instantaneous or not. Because it\'s important for you to get the right sort of model, it might be worth waiting an hour or two or until tomorrow to abort the model you have and get a slab.

Let us know please whether that works.

By the way, as an additional precaution to keep laptops cool while running a model, it\'s a good idea to raise the entire laptop slightly above the desk surface - not just the little feet at the back.

---

Cpdn news

There was a recent period where no Slabs were available, Mo. If a specification of Model type is made and none of that type are available, nothing else is substituted.

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

(...)I think the best thing will be to abort the HADCM you\'ve started. (...)

By the way, as an additional precaution to keep laptops cool while running a model, it\'s a good idea to raise the entire laptop slightly above the desk surface - not just the little feet at the back.

Thanks mo.v, I\'ll try to do exactly as you specified. Actually, after the laptop had downloaded the first combined model, I had aborted it, then verifed the preferences (on the webserver) again. What I did not do was to click \"update\" in BOINC. Anyway, while HADSM was checked, every time BOINC requested a model, I got the message that no models conform to my prefs were available. That was yesterday (while the server stats showed slab models available). After I checked HADCM, I got a combined model right away.
Does this maybe have to do with the fact that the HADCM3 models are optimised (Disk I/O) for laptops and the slab models are not? Maybe the software distributing the packages is configured to distribute only disk optimised models to laptops (which it recognises by the cpu type).

OK I\'ll abort then. If I don\'t get a slab model with the new try I\'ll run some other, smaller project on that laptop. I use the laptop as a light local server for design purposes and it\'s running most of the time anyway.

Your suggestion to keep the laptop cool: I already had placed it on two slats so that it gets air from the bottom.

---

Hi Batan

This thread on the independent forum describes the models available and the second post explains how the selection in one\'s preferences is implemented by the server. The info was given to us by Tolu who\'s the chief programmer in Oxford. He said nothing about the server detecting which computers were laptops and therefore overriding preferences and forcing them to accept an optimised HADCM model. In any case, I don\'t think that\'s an idea he\'d particularly want to implement because the optimised models, while reducing the strain on the computer\'s memory, still run at 100% CPU unless the member changes the % in their preferences. It\'s the 100% CPU usage that can cause overheating on some computers, particularly laptops.

http://www.climateprediction.net/board/viewtopic.php?p=66108#66108

(Nor is the server forcing computers without enough RAM for the HADCM models to accept HADSM slabs.)

If the model allocation works exactly as Tolu explained to the mods, and we have had no reports to the contrary, the member\'s model selection is never overridden by the server.

Your mistake, very easily made, was to forget to click Update in boinc manager after changing your preferences.

Why you got a message saying that no models conforming to your preferences were available when the server presumably thought you wanted HADCM or either type, is a complete mystery to me.

As I type, slabs are available, but do first check what is called the \'Server Stats\' page in the menu to the left of here under Taking part in CPDN. And please let us know whether you get what you want.

---

Cpdn news

This time it worked as expected; somehow I must have proceeded in a wrong way last time.
I aborted the HADCM model now which would have taken almost 2 years to compute, and this time BOINC downloaded a HADSM model. Still takes a lot of time, but at least with a perspective of actually finishing one day. So what I suspected of disk I/O optimised models does not apply.
Thanks.

Hi mo.v,
I just saw your last message which you probably posted while I was typing mine.
As I said, it worked fine this time and I got a slab model. Thanks.

As of a few months ago the Slab model has also been optimised to reduce disk I/O, and work is underway to do the same with the SAP (HadAM3) model. The SAP model isn\'t available here yet, but once the testing is complete I imagine it will appear at this project as well as the SAP project.

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

Hi Batan

You did well to persevere and get what you knew was best for the laptop.

As even the slab will take a long time, during which all sorts of things can happen, I\'d advise you to back up the contents of the boinc folder regularly. Through my sig you can get to the project READMEs where several methods of backing up are described.

---

Cpdn news

Why my Model date keeps on going back to 25/08/1812 00:30 when ever it reach 10:00????