One of my machines, a new notebook with the slowest processor available for that model but otherwise OK specs (Dell XPS L702X, Win64, i3 2310M, 8GB RAM, NVidia 550M) is currently processing 4 workunits:
2 setiathome 6.03 which take between 7 and 9 hours
1 seti@home 6.10 (cuda_fermi) which take about 12 minutes
1 Astropulse v505 5.05 which is estimated to take about 55 hours
The thing is, the Astropulse workunit has been processing for 6 hours and 22 minutes (estimated by the application to be at 11.8 GFLOPs/sec) and still has 48 hours 22 minutes to go but the deadline is 11 July.
I usually hibernate my notebook when I know I'm going to be away for a while but I still manage to process, I estimate, about 2.5 seti@home hours a day. So I reckon that by the time the Astropulse workunit fails to make it's deadline this machine will have dedicated over 40 hours of processing time at 11.8 GFLOPS/sec for no purpose. Shouldn't the deadlines be less tight for workunits of this size? It seems very wasteful of processing time that could be dedicated to workunits that will actually be used. The seti 6.03 workunits have deadlines a few weeks longer despite taking much less time to process.
Time to completion is only a 'best guess' based upon certain variables BOINC takes on your system. In fact, most AstroPulse workunits need only about 15 to 20 hours of processing on a system such as yours.
Have you noticed the Time to Completion estimate drops at a faster rate than 1 second per wall clock second?