Problem with work

I've just had a look at the single task from your machine still visible in the online database. It shows indications that the processing stopped and restarted a couple of times. This is quite normal since the restart was always from a saved checkpoint and not from 0% done. This is exactly the behaviour to be expected if you stopped and restarted your machine.

So why do you think the processing was constantly resetting to 0% done? There doesn't seem to be any indication of that in the task output. The task seems to have got to about 25-30% completed when you aborted it.

Deadkines are something that everyone has to live with. The deadline for current work is 18 days and your machine should take rather less than 40 hours to crunch a task. Based on a similar speed machine that I'm running (P4 2.4 GHz) you should expect crunch times around 24 hours on average for the version of the science app you are running. So if your machine can spend 2 hours per day on E@H tasks, you can easily meet the deadline. If you set the resource share too low for E@H (ie 10% or less), you might have a problem.

This really isn't possible. BOINC will insist that your other projects will get their fair share of your computer's resources. A single project may hog the CPU in the short term in order to meet a deadline. Afterwards BOINC will ensure that the other projects are given more time to make up for this.

What was it on previously?

Your version of BOINC, although "old" shouldn't be an issue. However, the more recent versions have more features and probably do a better job of managing deadlines for multiple projects. You could consider upgrading BOINC. It's actually very easy to do and would be worthwhile. Let us know if you need any assistance.