Thanks guys. I think I got it now.
The explanation with the calculation helped very much.

Hello Sprigo,

Can you go to the Result Status page, find those middle 2 results and click on the link in the Status column. Probably if you compare those Result logs with the 2 at bottom/top of your 4 posted here, you'll see some difference. Probably something was recorded towards the later part of the log, suggesting only the last X seeds/structures were counted for credit.

ttyl

Hi sprigo,
A while back a member had his result file written over toward the end of the 8 hours. As a result, only a few structures (seeds) were returned and he only got credit for those structures, so he lost a lot of points. This gives you something to look for in those 1100+ lines. There may be a line reporting a problem with the work file.

Lawrence

sprigo, can you remember if your machine was doing something which would have caused a high system activity (e.g. much paging because of another process needing much more RAM than available) on this machine while those two WUs were running?
Since Boinc counts CPU time (vs UD which was counting wallclock time) that should theoretically have no importance. But it is obvious when you monitor these things long enough that Windows is "charging" much of its activity to the running processes, as opposed to good old mainframes of the past which were more carefully accounting CPU time between user processes and system tasks.

If that was the case those two WUs have reached the 8-hour limit before having been able to do as much useful work as their benchmark would let Boinc expect.

Edit: If you still have the logs could you please tell us the start and end times for those two WUs.

Cheers. Jean.

Didactylos:

Do points awarded indicate the approximate amount of work processed? If computer "A" usually averages 136 points and computer "B" 272 ... does this mean that "B" averages twice as much work per eight hour unit?

fkehoetx
New Cruncher