Whenever I start my computer, the work unit restarts at the beginning. My normal procedure when shutting down is to suspend the work unit using Commands -> Suspend, and then exit BOINC by going to File -> Exit. After restarting the computer, the work unit will show the work from before. However, when I go to Commands -> Run ..., it restarts the work unit rather than continuing from where it left off.

I\'ve also tested to see if this restart problem occurred when I: 1) Just suspended the work unit and then continued it. In this case it works fine.

2) When I suspend the work unit, exit BOINC, and restart BOINC, it also restarts the work unit.

Thanks for your help.

Below are the messages I get when starting up:
3/25/2006 11:41:51 AM||Starting BOINC client version 5.2.13 for windows_intelx86
3/25/2006 11:41:51 AM||libcurl/7.14.0 OpenSSL/0.9.8 zlib/1.2.3
3/25/2006 11:41:51 AM||Data directory: C:\\Program Files\\BOINC
3/25/2006 11:41:51 AM||Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.80GHz
3/25/2006 11:41:51 AM||Memory: 766.33 MB physical, 1.83 GB virtual
3/25/2006 11:41:51 AM||Disk: 24.40 GB total, 6.78 GB free
3/25/2006 11:41:52 AM|climateprediction.net|Computer ID: 184027; location: home; project prefs: default
3/25/2006 11:41:52 AM||General prefs: from climateprediction.net (last modified 2006-03-24 19:11:29)
3/25/2006 11:41:52 AM||General prefs: no separate prefs for home; using your defaults
3/25/2006 11:41:53 AM||Remote control not allowed; using loopback address
3/25/2006 11:41:53 AM||Suspending network activity - user request
3/25/2006 11:42:00 AM|climateprediction.net|Deferring computation for result hadcm3lb_5bql_05036473_0
3/25/2006 11:42:00 AM||Resuming computation and network activity
3/25/2006 11:42:00 AM||request_reschedule_cpus: Resuming activities
3/25/2006 11:42:00 AM||Suspending work fetch because computer is overcommitted.
3/25/2006 11:42:00 AM||Using earliest-deadline-first scheduling because computer is overcommitted.
3/25/2006 11:42:01 AM|climateprediction.net|Restarting result hadcm3lb_5bql_05036473_0 using hadcm3lb version 508

It restarted because the models crashed, possibly as a result of using an old version of BOINC.
As you have now upgraded, the models may now work.

It restarted because the models crashed, possibly as a result of using an old version of BOINC.
As you have now upgraded, the models may now work.

This problem occurs when running the new version (just downloaded it yesterday or the day before). It doesn\'t seem like the model is crashing. It\'s running fine for as long as I keep my computer on, but when I shut it down and re-start, or if I exit BOINC and restart that, then the model will start over rather than continuing where it left off.

The new model takes a lot longer to reach a checkpoint. If you don\'t let it run for this long, then the only restart point that it has is the beginning.

Checkpoints are done every 432 Time steps (six Model days). So, wait until just after: 0030 on the 1st, 7th, 13th, etc, of any Model month to minimise lose.

You can see where you\'re up to, on the graphics display. Use Z to get rid of the top layer of info, and 8 to get a countdown to the next save.

Some people find it helps if they \'exit\' from the bonc manager before shutting down windows.

---

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

Another thing that may be slowing the computations down is if your are running the screen saver side as well. That takes up quite a bit of the CPU capacity, slowing the computation down. A lot of people just use a \"Blank\" screen saver.

I also click on \"run always\", not just based on preferences, which I don\'t know if it really makes any difference.

Dave

The new model takes a lot longer to reach a checkpoint. If you don\'t let it run for this long, then the only restart point that it has is the beginning.

Checkpoints are done every 432 Time steps (six Model days). So, wait until just after: 0030 on the 1st, 7th, 13th, etc, of any Model month to minimise lose.

You can see where you\'re up to, on the graphics display. Use Z to get rid of the top layer of info, and 8 to get a countdown to the next save.

Thanks. I\'ll make sure to run it far enough to reach a checkpoint, and post here to let you know if that was it or not.