Have been running Climate Prediction on a number of different Windows systems for some time now. Have encountered problems on several systems running the Sulphur simulations. These simulations take a very long time (1,000 + hours) to complete, but in three instances my systems have had an unrecoverable error about half way through the computations (500 hours). On one system the error causes my entire system to slow down measurably as the process tries to recover by restarting itself constantly.

Is this a know bug, and if so is there a fix coming soon? I am going to have to stop running Climate Prediction as part of BOINC until I know that this problem is resolved.

All computers effected are running Windows XP Pro, using BOINC Manager 4.45.

---

Rather than type it all again, I\'ll point you to one of my posts on 4.45 <a href=\"http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=3764#18732\">
here.</a>

Also, in your general preferences, set \"Leave applications in memory while preempted? \" to yes.
Otherwise, I think, you will still have problems, because the timing wasn\'t changed, just that the model data is now left in memory if you ask for this, whereas with 4.45, the data was dumped regardless.

The \'problem\' you are having might be because your computer isn\'t quite stable enough for the intense, continuous, maths calcs involved.
It sounds as though the model is rewinding to try and recover from a problem.
It does this by rewinding a day and retrying, then a month, and then a year.

There are tips on hardware maintainence <a href=\"http://www.climateprediction.net/board/viewtopic.php?t=2124\"> here,</a>
and tests <a href=\"http://www.climateprediction.net/board/viewtopic.php?t=2126\"> here.</a>