I just wanted to share / query other peoples results with N in a PtMP enviroment.
In these tests we were running ROS 5.0, Single chain, 20mhz channel, SXT's running ROS 5.0 as clients, in nstream with 8 clients to the Omni. Total EiRP is around 800mW, furthest client is 4km.
So - under N, running MCS 5 we achieve approx 31mbit REAL TCP throughput running a Btest from an SXT to a PPC router in our core. Under this example the client has a signal of -59/-60 - i imagine we were maxing out the CPU on the SXT.
Also under N we have another SXT running with a signal of -74/-80 (not great), running the same Btest to the same PPC router, MCS 5 wont stay connected, when locked into MCS 4 - we get under 10mbits, MCS 3 we get 15mbits, but it is very unstable and CCQ's are jumpy. I didnt try going to any lower rates
When we lock this SXT into A rates - 24mbit/mode we get 16mbit/sec very stable with good CCQ's.
So i am wondering if other people have seen similar results - where clients with very strong signals get great throughput under N, but clients with less great signals get substantially worse throughput - which then is much better under A rates.