Thing is that most users want some form of qos/aqm to guard against load induced delay, which the apu and similar low power devices cannot handle above a few hundred mbps
The solution's not going to be found in the firmware, is it?
Enabling SQM [1] would probably mitigate the issue further, at the cost of requiring more processing power.
[1] https://openwrt.org/docs/guide-user/network/traffic-shaping/...