I've been fighting with a situation where CPDN tasks show no progress, despite as much troubleshooting as I (and others on these forums) can think of. This is the computer:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=755074

...and this post is from someone who seemed to have a similar problem (my comment is at the end):
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=7167

At this point, I think that maybe the best thing to do is not run CPDN on this machine. =-(
The question is, what's the best way to exit? For example, would detaching permanently "kill" the workunit vs. aborting it allow it to be sent to someone else? Or is detaching the thing to do? Or maybe it doesn't matter?

Or-- (fingers crossed) has someone found a good way to deal with this sort of thing & rescue the result?

Both options will abort the model. In such situations I generally abort then update (with 'no new tasks' set) to make sure everthing has been synchronised with the server, and when the model is marked as aborted on the Web site then finally detach.