[tomegun]

The overhead due to the message ordering is constant, irrespective of the number of cores/peers. See the link posted above for an explanation.