The design of MMA memo 166 inherently performs fraction of a millisecond integrations. Hence extremely high dump rates have an effect only on relatively small parts of the system.
It has been suggested (Rupen 1997) that even faster dumps may be needed for the autocorrelations, to allow cancelling out the atmosphere without the need for a chopping secondary. The system can easily be designed to provide very fast dumps on the autocorrelation lags. Processing these fast dumps, and providing very fast dumps of more than just the autocorrelation lags (and processing them), requires a lot of study before any definite statement can be made but the problem is mainly the data rate and computation in a relatively small part of the total system.