Hi Timoleon
Mo here from CPDN. We are aware that CPDN puts very little news on the project website, but that's a problem we moderators can't solve. So we put the news on the forum at the top of the Number Crunching section here:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_forum.php?id=2
And we really make a big effort to keep that properly up-to-date. CPDN server status problems are also posted on this BoincStats forum in the Project News section.
Your climate model may well have crashed with a 'visual Fortran runtime error'. I suggest you post about it on the CPDN forum in that same Number Crunching section. Then several moderators etc will be able have a look at your computer details, the model crash details and so on. On the project forum we have direct links to members' project web pages with these details.
Forget about Predictor@Home for the time being. They are in the process of moving from San Diego to Michigan and won't be issuing workunits for at least another month as far as I know. In your Boinc Manager Projects tab, set Predictor to No New Tasks, which will stop it constantly trying to get new work. I would advise against newbies starting off with Predictor even when they have work available because it's difficult to get a quick response on their forum.
I know that Primegrid and QMC (Quantum Monte Carlo) currently have workunits available. From my (limited) experience, the workunits from these 2 projects run very stably.