For people wanting to take advantage of the 2.5G ports but "only have CAT 5E" wiring, fear no more (or at least fear less). I got the same throughput when I swapped to the CAT5E port in my office and it even goes through a telco punch down in my wiring closet with a rat tail connect to the RT-BE86U. SAME NUMBERS, both direction to the RT-BE96U! Good thing running the CAT6 wasn't that much effort and it IS a homerun at least!.
@CollinTaylor got me curious about the CPU loading. I also have a GT-AXE11000 which is a lower speed CPU (1.8 vs 2GHz), less memory and one less radio 3.0.0.4 version of the GT-AXE16000.
After seeing that the CAT 5E run worked for the GT-AXE16000 (prolly 100ish feet) I connected my GT-AXE11000 CAT5E termination to the other 10GW/L port on the RT-BE96U and validated the PHY between the RT-BE96U and the GT-AXE110000 was 2.5G
I then ran a series of iperf tests between GT-AXE11000 and the GT-AXE16000 with the RT-BE96U as the switch in the middle.
With the GT-AXE16000 as the server I got the same 2.35GB ish numbers. With the GT-AXE11000 as the server it dropped to just over 2GBish. I ran the tests with a single thread, 2 threads, and 4 threads over 30 seconds. No statistically significant changes were observed.
So as theorized, the 1.8GHz processor in the GT-AXE11000 appears to get challenged vs the 2.0GHz version in the GT-AXE16000
The GT-AXE16000 were on sale at less than $250 out the door last weekend on the ASUS site. I lost a 24 port Gb switch and an RT-AX88U in our thunder storms a couple weeks ago.