When using the real-time mode of the broker, it's not sufficient to report the total wall clock time versus the total simulated time as an average. There could be "bursty" behavior that is lost; certain intervals could take significantly longer to process and therefore real-time mode was not actually achieved. Report the real-time efficiency as an optional CSV output of the broker.