New hap ax s wifi

Hi,

I’ve got a new ax s. Happy to see new chipset. I’m testing wifi capabilities, and results are not good enough. Wifi client is connecting with 2.4gb/2.0gb tx/rx rates, and max speed is about 500 Mbit/s

Configuration is simple. Any suggestions?

/interface bridge
add name=bridge1 protocol-mode=none
/interface ethernet
set [ find default-name=sfp1 ] disabled=yes
/interface wifi
set [ find default-name=wifi2 ] channel.band=5ghz-ax .skip-dfs-channels=all .width=20/40/80/160mhz configuration.country=Russia .mode=ap .ssid=testAAAA disabled=no security.ft=no
/interface bridge port
add bridge=bridge1 interface=ether1
add bridge=bridge1 interface=wifi2
add bridge=bridge1 interface=ether2
/ipv6 settings
set disable-ipv6=yes
/ip dhcp-client
add default-route-tables=main interface=bridge1
/system clock
set time-zone-name=Europe/Warsaw

How do you test ?

Wifi connected to hap ax s and then where to ? How ?
External speed test server ? Then it might be normal.
Test results for hap AX S as router are about 500Mbps.

Try to test using internal iperf3 server connected to one of the LAN ether ports. See what speed you get then.

I know there's no HW offload on eth1, so I'm using WAN on eth2. Cable speeds are up to 1Gbps, but Wi-Fi still maxes out at 500Mbps.

Wifi test on pixel 8 pro, ookla speedtest app

Since you have 1 port as WAN, it's not part of your bridge (I would think) so it's not HW offloaded.

If you do that same Ookla speed test from a cabled computer, to you get 1Gb ? Or also 500Mbps ?

see cfg above, its simple ap bridge

Just to be 100% sure ...

Cabled computer, connected to one of the LAN ports of hAP AX S, same speed test. What do you get ?

up to 1GB/s

How is it doing 2.4Gbps if it’s triple chain 1.8Gbps ?

160Mhz

AX2400 refers to a

WiFi 6 (802.11ax) router's maximum theoretical speed, combining 574 Mbps on the 2.4GHz band and 1800 Mbps on the 5GHz band for a total of ~2400 Mbps, enabling smoother 4K streaming, faster gaming, and handling more devices, but actual speeds depend on distance, interference, and device

ahh ok

Nevermind...

I guess It allows setting 160mhz while the Data on the Mikrotik Website only mentions 80mhz.

Maybe the 160mhz just doesnt work and its a bug?

@kpxnt did you solve the problem? I have exactly the same problem with my hAP ax S device.

I think the bottleneck is CPU, at least in my case, as during WiFi transfer at ~500Mbps it’s load is ~90-95%. Based on block diagram ether1 is not hardware offloaded (not connected to switch chip) and all traffic in processed by CPU.

I will try to connect LAN to ether2 to make if HW but I will loose PoE capabilities.

BTW looks like a bug in OS or driver, that 500Mbps saturate CPU.

Problem still exist, on eth2 same problem. Probably it’s firmware (drivers) issue.

For each new case, create ticket with supout.rif.
The more data they have, the more they an find the root cause.

Ticket created, let’s see if support can find root cause.

Depending on your location (big town centre or remote in the country side) - translates on how congested the 5GHz spectrum is, the 160 MHz wide channels may be significantly impacted by interference, which will directly affect throughput

Moving wired connection from ether1 to ether2 (or any other switch-connected port) won't change CPU processing when passing traffic between wired and wireless sides. Every wireless-originating packet en route towards LAN will have to leave CPU via one of its interconnects ... either via ether1 or via switch chip and either way it's the same for CPU. Ditto for the opposite direction.

The only case where switch helps to reduce CPU load is when both ingress and egress port are connected to switch and no frame processing requires CPU (and that's true for both L2 and L3 HW offloading).

Support didn’t find any issues in configuration as well as support file. But issue in noted and will try to improve in the future.

here is what I get:
Thank you for the feedback! It was noted and passed down.

We will see how we can improve it if it is possible.

I’m just chiming in to say that I’m getting much better wireless performance when I set security.authentication-types=wpa2-psk instead of wpa2-psk,wpa3-psk.