climateprediction.net home page
No more WU's

No more WU's

Message boards : Number crunching : No more WU's
Message board moderation

To post messages, you must log in.

AuthorMessage
David A. Ort

Send message
Joined: 5 Jun 08
Posts: 1
Credit: 128,710
RAC: 0
Message 43986 - Posted: 9 Apr 2012, 21:42:23 UTC

I used to get lots of WU's - now I have not gotten in weeks. Am I doing something wrong? What is a BBCode tag? I guess I am pretty naive. Thanks. David
ID: 43986 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 43987 - Posted: 9 Apr 2012, 21:58:38 UTC - in response to Message 43986.  

The project doesn't have a continuous supply of work. It's released in batches, as and when the researchers decide. And then there'll be a gap while they analyse the results, and decide on what to look at next.

The researchers aren't at Oxford, but in various places around the world. The people at Oxford just set up the experiments according to instructions received.

Look at the blue menu to the left of here, 5 from the bottom to see the status of available work.

As for BBCode, click on the underlined line at the top of the list, and you'll get a list of codes, and their format.


Backups: Here
ID: 43987 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4472
Credit: 18,448,326
RAC: 22,385
Message 43988 - Posted: 9 Apr 2012, 22:32:58 UTC - in response to Message 43986.  

And a batch is unlikely tomorrow morning after a public holiday. No work for weeks makes me wonder if it is worth checking which model types you have allowed as there have been some work units going over the past week. - I picked up three a few days ago.
ID: 43988 · Report as offensive     Reply Quote
Profile Dave Jackson
Volunteer moderator

Send message
Joined: 15 May 09
Posts: 4472
Credit: 18,448,326
RAC: 22,385
Message 43993 - Posted: 10 Apr 2012, 11:15:28 UTC - in response to Message 43988.  

That will teach me - a batch of rapit models has been released!
ID: 43993 · Report as offensive     Reply Quote
old_user29078

Send message
Joined: 9 Nov 04
Posts: 11
Credit: 93,753
RAC: 0
Message 43997 - Posted: 11 Apr 2012, 22:37:30 UTC

I happened to check my status and I'd also noticed the same, that I didn't seem to have been given any work units in quite a while (and I'm still not getting any). While reviewing my stats, I also realised that many of my recent(ish) work units had in fact failed to complete, unfortunately.

I'm afraid I'm having to wonder (yet again) whether it is really worth my while trying to continue with climateprediction.net? While I don't have the most powerful computer, it is reasonably newish, and has a moderate amount of computing oomph to it (for a non-specialist desktop), but it's only ever switched on on average for about 2 - 4 hours per day. (Since cpn is supposed to be about countering climate change, for obvious reasons I'm not going to leave my computer on when I'm not actually using it myself.)

Also, the cpn work units are just so big and therefore it's very disheartening to see my computer crunch away for several weeks only for a work unit to often eventually fail. At least with my other BOINC projects, the little I can contribute does at least complete successfully.

cpn is the project that I'd absolutely most like to support, but if there is only a limited amount of work available, it's maybe better that I just pull out entirely, so that those with greater computing power and/or greater uptime can take the work instead, and I could then devote all of my spare CPU cycles to other BOINC projects instead, where I could be more sure that I was at least making a successful contribution?

I realise that breaking down the computational jobs into manageable chunks must be a difficult enough task in itself, but surely there must be some way to refactor the work so that there can be some smaller jobs available for those who can only contribute a little, otherwise I can't really see how it can be worthwhile for me, with my limited resources, to continue, I'm afraid?

It's a shame, because one of the things that appealed to me about cpn, as well as the importance of the issue, was that "everyone" would be able to contribute a little to help with the research, but it more seems to be that you really do need to have fairly substantial computing resources at your own disposal in order to make even the slightest contribution..
ID: 43997 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 43998 - Posted: 11 Apr 2012, 23:07:58 UTC - in response to Message 43997.  

Several years ago, the biggest of the models took 6-12 months to complete, even for those of us running 24/7. Since then a way was found to split up the models a bit, and use statistical means to join the bits together and analysis them.

But this is as small as they're realistically going to get.
2-3 hours a day on a desktop computer just isn't long enough for computer models that have been designed and written to run on the Met Office supercomputers.

The smaller, simpler models from the start of the project back in 2003-4 have outlived their usefulness, and the researchers have moved onto other areas of interest to them.

So, yes, it may be best to stick to other projects.
Sorry if this sounds harsh, but that's how it is with research that covers decades of weather.
And the researchers would ideally like models that run for centuries at much higher resolution than is used here.
This was once calculated to take over a thousand years of real time on today's fastest desktops, running without stop.

Your failures may be due to what appears to be a prefs setting; leaving the Suspend work if CPU usage is above to the default of 25%, which causes BOINC to keep stopping and starting every time the processor usage gets above that level.


Backups: Here
ID: 43998 · Report as offensive     Reply Quote
Dave Roberts

Send message
Joined: 15 Jan 11
Posts: 175
Credit: 6,242,691
RAC: 699
Message 44000 - Posted: 12 Apr 2012, 9:23:55 UTC

I think that Dave's comments about not keeping his computer running 24x7 will strike a chord with many people, especially those supporting CDPN.
I originally started with the BBC experiment but then signed up with CPDN and had the same thoughts and looked at my computer's power usage. I have a 27" IMac (twin core) and currently, running CPDN when I'm not using it, with the monitor disabled the usage is about 1.4/1.6 KW's per day. I think that is pretty reasonable and that by making sure that I am really careful in not using lighting etc. unnecessarily, I can justify 24x7 usage.
We desperately need this research and since the work has to be carried out eventually, my personal feeling is that we should really go ahead with 24x7 crunching. (Although I understand if others don't entirely agree.)
ID: 44000 · Report as offensive     Reply Quote
keputnam

Send message
Joined: 31 Aug 04
Posts: 26
Credit: 3,950,377
RAC: 630
Message 44051 - Posted: 20 Apr 2012, 0:17:56 UTC
Last modified: 20 Apr 2012, 0:18:36 UTC

I have noticed that there are very few WUs being created

My problem, is that I have set "all applications" and have not set any geographical limits (at least I don't think that I have - How can I tell??) and I've been on the Server Status page a couple of times, and have seen a few WUs available. If I immediately do an update, I get "Not reporting or requesting tasks" in the log

My last WU was reported more than two weeks ago, and I have a fairly large CPU percentage set for CPDN, so why would BoincMgr not be requesting more work?

TIA for any insight
ID: 44051 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 44053 - Posted: 20 Apr 2012, 0:56:59 UTC - in response to Message 44051.  

There are currently 65,177 work units of the Weather At Home Pacific North West variety available, so it's something to do with your computer and/or settings.
But the problem is:
"Not reporting or requesting tasks"
so your computer just doesn't want any more work at the moment.

Your computer is a Pentium 4, which only has 1 processor, plus about .3 of another processor contributed by the hyperthreading.
And you're running several other projects, which may be owed processor time, and one of your projects is T4T, which I think takes up a lot of computer resources, so it may just be a matter of waiting until your BOINC decides that it does need more work.

You've also had a lot of failures. You can minimise this by setting your preferences to 50% for the number of processors, so that you only get one work unit at a time. This may require you to use a separate "venue" for this project relative to your other projects.


Backups: Here
ID: 44053 · Report as offensive     Reply Quote
keputnam

Send message
Joined: 31 Aug 04
Posts: 26
Credit: 3,950,377
RAC: 630
Message 44054 - Posted: 20 Apr 2012, 5:28:26 UTC - in response to Message 44053.  

Yeah, it was just about the same time that I started T4T that I returned my last WU

OK, I'll just wait awhile more, then

And Yes, I am shopping around for a new computer. When I get one, I'll turn this old box into a dedicated cruncher for a couple of projects
ID: 44054 · Report as offensive     Reply Quote
keputnam

Send message
Joined: 31 Aug 04
Posts: 26
Credit: 3,950,377
RAC: 630
Message 44057 - Posted: 21 Apr 2012, 2:22:16 UTC - in response to Message 44054.  
Last modified: 21 Apr 2012, 2:22:43 UTC

I went ahead and reset the project, since CPDN had the highest CPU share of any of my projects, and was still not downloading after three weeks of inactivity

Within 2 minutes I had gotten a new WU

<shrug>
ID: 44057 · Report as offensive     Reply Quote

Message boards : Number crunching : No more WU's

©2024 cpdn.org