Message boards : climateprediction.net Science : LARGE project??
Message board moderation
Author | Message |
---|---|
Send message Joined: 5 Jun 07 Posts: 8 Credit: 78,491 RAC: 0 |
Since I am a bit of a newcommer to the cpdn program, I just wanted to verify whether or not it was normal to be assigned a relatively HUGE project(in comparison the the other project areas like seti)...I have about 2767 Hr. left...??? Greg |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
Yes, perfectly normal. This is the BIG DADDY of all the projects. Help, advice, hints, and tips available here, on the Community Board. If you want to know about the project, there\'s lots of info at the start of this web site, in the Climate Science section. And please! Do backup your model at least weekly if you want it to reach the end. Backups: Here |
Send message Joined: 29 Sep 04 Posts: 2363 Credit: 14,611,758 RAC: 0 |
It will probably be the biggest job your computer has ever handled. The models run for 160 simulated years. I see that your first trickle was doing about 2.7sec per timestep which is a good speed. The number of hours remaining that boinc quotes has to be taken with a pinch of salt particularly at the beginning. See how long it takes for your computer to reach each trickle point (Dec 5 of each year) then multiply by 160 for an estimate of total run time. I\'d guess up to 5 months running most of the time. You have masses of memory and if you want could try a seasonal model running in tandem on the second core. One of these might take you 3 weeks or a bit longer. These special models probably won\'t be available for very long. The seasonal models count as a separate project. http://attribution.cpdn.org/ To run these massive WUs you need to take a few precautions to make sure you complete it and aren\'t disappointed by a model crash. Through my sig look at the project READMEs. In the one called Running the model, look at the top tips. And in the README about avoiding crashes, look at item #5 by Mike and item #1 by Les about how to back up the boinc folder contents. So if you do have a crash, you restore and continue crunching. Edit - Les was quicker off the mark. Cpdn news |
Send message Joined: 5 Jun 07 Posts: 8 Credit: 78,491 RAC: 0 |
Thank you all, I see Im doing \"The Big Daddy\" project here... I also read the info about backing up the project in case of a crash. In this instance, Im assuming that the \"crash\" refered to is a system crash (a crash on my end rather than a crash on the cpdn server)?? If this is the case I feel relatively secure that my computer wont crash and that its quite stable (given the steps required for a backup - In my opinion it would be nice if coding for a BACKUP procedure or button were incorporated into the BOINC program)... I also followed the link - and checked out the \"seasonal climate model project\". I read there that it is recomended to \"wait to complete other cpdn projects before starting the seasonal weather mapping project\". I just wanted to make sure this was really necessary?? Greg |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
The reason for the \"wait\" related to Seasonal, is because most computers just don\'t have the memory neeeded for both. In your case, this isn\'t a problem. BOINC needs to be Exited from before a backup is made, to ensure that all files, including some that are locked by BOINC, are copied, otherwise the copied set won\'t work when copied back later. So there can\'t be a button in BOINC to do the copy for you. If an \"Exit, Copy, Paste\" is too much effort, then you could try the automated backup program written by a BBC cruncher, Richard Rodway. This is also documented in the Backups section. A crash of a model can be caused by many things, including a conflict with other programs. One of these is when another program is using ALL of the hardware resources, and the science app. is left idle for more than X amount of time. (I forget the figure.) In this case, Boinc will assume that the science app. has failed, and begin the process of closing it down and getting a new model. And if you run more than one work unit at a time, and need to restore a backup because of a climate model failure, then ALL of the work units have to be restored, unless you\'re prepared to do a fair bit of editing with xml files, etc. Backups: Here |
Send message Joined: 29 Sep 04 Posts: 2363 Credit: 14,611,758 RAC: 0 |
I also wish there were Backup and Restore buttons within the Boinc manager. This is unlikely to happen any time soon because cpdn is the only current boinc project with workunits so long that they need to be backed up. Another problem is that one needs to have exited from boinc before backing up because no boinc files must be in use....... In the Project READMEs in my sig, there\'s a whole compilation of backup methods. I use Les\'s easy manual instructions (the first method in the list), but as he says, there are other more sophisticated methods. Backing up regularly is a really good idea because the longer a workunit lasts, the more chances there are that something can go wrong. A model can crash even when everything else on the computer continues to behave normally. In the README about avoiding model crashes, item #5 by Mike suggests ways of minimising potential problems. And in the README about running the model, it\'s worth looking at the top tips. Cpdn news |
Send message Joined: 10 Jun 05 Posts: 10 Credit: 4,863 RAC: 0 |
Backing up is not the real problem, I am as capable as anyone else if it comes to copying the BOINC folder to another location. But I crunch for multiple projects, and restoring a backup without affecting other projects is not as simple as copying back the climate prediction folder. From the bit of reading I\'ve done it involves editing a number of files and selective copying. To be quite honest I can\'t be bothered to do that. So I\'ve stopped backing up and count myself lucky my last model crunched to completion without any problems. I think (hope) crashes are too infrequent for me to become proficient in the steps needed to restore a particular model. Anyway, I also would like a simple program which can backup CPDN and restore a backup for me. |
Send message Joined: 13 Jan 06 Posts: 1498 Credit: 15,613,038 RAC: 0 |
We\'ve asked the Boinc developers at Berkeley for that. I'm a volunteer and my views are my own. News and Announcements and FAQ |
Send message Joined: 1 Oct 04 Posts: 22 Credit: 14,464,707 RAC: 2,328 |
That would be a great help. Assuming it would be applicable to APS@Home (presuming they also have large multi-day, -week, -month wu\'s) as well as the CPDN family of projects. Have also seen Einstein wu\'s reaching into the 20-30 hour range, and QMC at 12-15 hours so likely to be of use on those projects also. Do more \"votes\" from other project help boost the priority? |
Send message Joined: 5 Aug 04 Posts: 1496 Credit: 95,522,203 RAC: 0 |
Probably a question better asked on the boinc Boards. OT -- APShome WU are short, 35 minutes to an hour for one version, two to three hours for another. (Nothing at all, while James is away from the UK.) "We have met the enemy and he is us." -- Pogo Greetings from coastal Washington state, the scenic US Pacific Northwest. |
©2024 cpdn.org