When it comes to recording data, we usually avoid resetting the original value on the object. It tends to cause pain down the road. Instead, we basically make a copy, increment it whenever the throughput should increment, and then reset that value.
From your question, it sounds like you want a chart like this:

You can see that once the throughput gets to the limit (100 in this case) then counting starts back at one. I did this with a statistics collector, that listens for objects leaving the processors. It remembers how many times that event has fired, and keeps a count for each processor (on a row label). If that count gets above the threshold, it gets set back to one.
ThroughputReset.fsm
.
Jordan Johnson
Principal Software Engineer
>