Model reset to 1920 after requesting connection to Internet

[Anthony F. Withers]

Last time I checked on the experiment (Friday 30 June 2006), the run-time was showing 2%+. After requesting connection to the internet, it is now showing run-time 0.2%.
WHY?

[mo.v]

Hi Anthony, welcome to the forum. It looks as if your model may have crashed and a new model downloaded when you connected to the internet and your experiment contacted the server. Connecting to the internet is unlikely to have caused the model crash.

We need more details to try to find out what happened and help you avoid the same thing occurring again. Could you please copy & paste the contents of your boinc manager messages for us?

[Anthony F. Withers]

03/07/2006 17:04:10||Starting BOINC client version 5.3.19 for windows_intelx86
03/07/2006 17:04:10||libcurl/7.14.0 OpenSSL/0.9.8 zlib/1.2.3
03/07/2006 17:04:10||Data directory: C:\Program Files\Climate Change Experiment
03/07/2006 17:04:13||Processor: 1 GenuineIntel Intel(R) Pentium(R) 4 CPU 2.40GHz
03/07/2006 17:04:13||Memory: 255.48 MB physical, 619.39 MB virtual
03/07/2006 17:04:13||Disk: 32.70 GB total, 21.23 GB free
03/07/2006 17:04:14|BBC Climate Change Experiment|Computer ID: 155291; location: ; project prefs: default
03/07/2006 17:04:14||General prefs: from BBC Climate Change Experiment (last modified 2006-03-15 14:59:02)
03/07/2006 17:04:14||General prefs: using your defaults
03/07/2006 17:04:14||Local control only allowed
03/07/2006 17:04:14||Listening on port 31416
03/07/2006 17:04:14|BBC Climate Change Experiment|Resuming task hadcm3ln_4v01_00465975_1 using hadcm3l version 508
03/07/2006 17:04:14||Using earliest-deadline-first scheduling because computer is overcommitted.
03/07/2006 17:04:14||Suspending work fetch because computer is overcommitted.

[mo.v]

Hi again, thanks for that.

Using your computer ID number I've looked at your results so far. You can see all this for yourself by clicking on the Your Results button in the boinc manager then going through the links.
http://bbc.cpdn.org/show_host_detail.php?hostid=155291&PHPSESSID=a615a9727800411edbd21c10c3caa0a1

You've had 5 models so far - the current running model, one that was deliberately aborted from Oxford with the -99 killer code like everybody else's because it had the flawed sulphur file. One crashed with a -107 error/exit code and two with a -161 error code.

I think the basic problem is that your computer has a nice processor but is short on memory. The minimum for the BBC project is 512Mb of RAM but your computer only has half this, 256Mb. If I were you I'd consider adding an extra 256 Mb memory card - there's probably a second memory slot alongside the first. Spare cards aren't expensive (have a look at the Novatech website) and adding an extra card can be a DIY job for you or a friend. You just plug & clip the new card in, holding it by its edges, with everything turned off of course, then reboot. This would improve the performance of the computer across other applications too.

For the -107 error I'd also check that the graphics card has the latest update from the web. Thyme Lawn explains how:

http://bbc.cpdn.org/forum_thread.php?id=1038

For the -161 errors, Mike's post on avoiding these is very useful:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=4231

You also need to back up your running model so that if the worst happens again, you stand a chance of restoring most of the model. Les explains how:
http://bbc.cpdn.org/forum_thread.php?id=1952#8299

Hope that helps - let us know how you get on.

[microtoxic]

03/07/2006 17:04:13||Memory: 255.48 MB physical, 619.39 MB virtual

This is far below the minimum requirements. Running the project with success will most probably need a memory upgrade.