SQM/QoS can saturate the CPU/is this expected or can the code be improved?

[darksky]

No idea... as an update, I am finding that the NSS code @quarky @Ansuel (and others?) made available to the community preforms as-good-as SQM with the endpoint of bufferbloat numbers yet allows for faster download speeds on WiFi devices.

@darksky

Pretty decent difference in terms of CPU% and download throughput.

iphone7:
Without NSS download speed was 325 Mbps with NSS it was 445 Mbps. Difference was seen in htop. Without NSS it was saturating one of the cores. With NSS it was 20-25% across both cores at the peak.

With an older Macbook Pro:
Without NSS download speed was 282 Mbps with NSS it was 415 Mbps. Difference was seen in htop. Without NSS it was saturating one of the cores. With NSS it was 78% at its peak on a single core (both in use).

Hardware is R7800 with QoS (fq_codel/simplest-tbf.qos).

See here for key steps to build the NSS code (assuming you have a compatible device). See here for setup of fq_codel on it. You do not use the LuCI package. That thread is pretty large.