That's because the results are for an approximately 24 hour day. BTW, when the next day starts seems to vary depending on which stats site you use and when they grab the XML file with the credit from that particular project. That's why you see a different amount reported for different BOINC statistics sites for the same day. Also, if a project site is down like SIMAP was for about a 12 hour period last week, it caused boincstats and boincsynergy both (IIRC) to show a day with zero credits followed by a day that had the credits for a 48 hour day.
I'm not sure if it's QMC you're talking about but, on QMC, the stats are all over the place because many WU take longer than 24 hours on slower machines. I've seen them take over 30 hours of
CPU time on a Sempron 2500+ and
actual clock time can be days if you have other projects going on the machine. The long WU can be worth 300+ to 600+ cobblestones so you may get daily credit results like:
DAY : Creditday 1 : 0
day 2 : 0
day 3 : 640
day 4 : 0
day 5 : 0
day 6 : 350
day 7 : 0
Since the length of WU varies, you will see this to some extent on all projects. Also, doing something like a virus scan can cause a WU to take a much longer clock time since it takes the CPU and disk away from the program for long blocks of time. Then, you have projects like Climate Prediction that can take days between trickles on slower machines.
Also, you may not get credit on the day you turn a WU in if it has to wait for another machine working on the same WU so it can validate them against each other for a quorum.
Yes, I do tend to go overboard in my explanations
-- David
