waveform bufferbloat score of A+ wired and A via wifi
computer 1 is in an online game with stable ping of <10ms
computer 2 starts a large file download
computer 1 ping spikes and has packet loss
computer 1 ping settles after about a minute or so of struggling
this repeats itself whenever a computer or device on the network downloads or streams
sqm is on and set to 900/900 via ethernet running cake/piece of cake without link layer adaptation.
i also tried standard qos with max priority to/from the gaming computer and wan which also did not make much of a difference.
the one thing that did work is if i drastically reduce the sqm ingress/egress to less than half of the ethernet throughput ex: 400/400, and then the ping spike is minimal.
this is not ideal of course because it limits over half the download speed available to the network.
speedtest via wifi generally runs around 950/950 which is great, but it seems that latency suffers greatly if more than one client is pulling data.
is this normal behavior? of course it would be ideal to run the gaming computer via ethernet but for now due to where things are located i cannot run a wire. my previous tplink ax1500 didn't have 160mhz but never had ping spike issues via 80mhz and qos turned on.
Wifi is not full duplex like ethernet but half duplex.
This essentially mean between the router and the device doing wireless only one can talk, so either the router is silence or the device they cannot cross talk at the exact same time which with full duplex can be, that is already one drawback
if a packet is big especially if mtu was high set, then it takes also longer for the router to respond, this delay causes also bufferbloat like symptoms.
you can lower the sqm speed or choose fq_codel (often works better on mediatek based socs), or try lower the mtu and re-optimize sqm.
But other than that, wifi never could substand the same exact stability as ethernet, therefor wifi has also more priority than lan but the negative side is that it can also bring down lan if a client was far away and started downloading alot of data, its not priortized because it is 'better' more to survive performance with these drawbacks.
i see! so do you think the ax1500 is doing some sort of wifi qos that the mt6000 is not? i'm mainly curious why the 'older inferior' router is exhibiting better performance within the same parameters
thank you! the oem was what i was using initially and it had the same behavior, possibly worse, which is why i decided to try the op24 driver. are there any particular settings you suggest i try? i love everything else about this router and the throughput speeds are much nicer than my ax1500, but if i can't reliably use it for wifi gaming then i may switch back to my old setup