Oh, I should add that the number for "max" is often grossly out of proportion. This is because at least once per tick, the operating system takes control and makes sure the right process is running (this is called scheduling). Because we simply read the TSC and then measure the difference, we can't account for this operating system overhead that occasionally happens in the middle of a rule evaluation. So, unfortunately, this max number is actually quite useless (originally, I had hoped it would show the longest path through the rule). Eventually, we are planning on adding more detailed statistics to further help diagnose and improve rule performance.