Well I'm out of ideas. Try a different channel? The 23% packet error rate isn't great. Still makes me think the signal's too hot.
Signal may be too hot, but in this case it seems that this is caused somewhere else.
Is there any other devices in between PCs and Wireless Wire dishes?
My first guess was that traffic generation was bottle necking Mac mini single core as well, adding -P argument and multiple parallel streams should give some improvement.
based on this:
it seems that -P isn't much help.
As a workaround try opening multiple servers and multiple tests in parallel
You can also set-up traffic-generator (more resource friendly UDP test)
if addresses are still - master (192.168.88.2) and slave (192.168.88.3)
in master device terminal paste:
/tool traffic-generator packet-template
add ip-dst=192.168.88.2 ip-gateway=192.168.88.3 udp-dst-port=0-100 name=test1
/tool traffic-generator stream
add mbps=900 name=str0 packet-size=1500 tx-template=test1
before each test refresh template by running:
/tool traffic-generator packet-template set [find]
and run test:
/tool traffic-generator quick mbps=900
this will create bi-directional (loopback) UDP packet stream.
More information available here:
https://wiki.mikrotik.com/wiki/Manual:T ... _Generator
Documentation is written by following KIS (Keep It Simple) design principles, it may not include in depth tutorials, but will be based on absolute minimum information to get users started.