Zmeg --

It is very difficult to tell just by looking at the number of work units processed. The work units do vary enough in size that it is entirely possible to experience a period when the number of unit processed will be much higher than normal. A better measure would be to look at what the device is accumulating for run time and points on those instances when it appears to be sending in more than normal. If it is showing that it has accumulated about a day's worth of run time and is getting about the same number of points as it would when processing fewer, then things are okay. If, however, your run time is substantially less than a day and the points are down, you are probably having work units crash and would want to check out your system using the tools discussed in the FAQ here, or comparable utilities for Linux if that is what you are running.