hAP be lite performance issues

I just recently bought 2 hAP be lite to act as a simple AP to replace my 2 Unifi U6 PROs (because they have that stupid bug where clients randomly lose connectivity in multi-AP setup).

And boy, I was not really happy. I tried literally every configuration possible - standalone/CAPsMAN, with/without MLO, with/without 802.11k/v/r, wpa2/wpa3 settings, superchannel test, ether1 or other ethernet ports, Netinstall 7.24.1...

But one thing persisted, I simply couldn't get anything more than ~600mbps download on wifi7 nor wifi6 clients (the CPU went to 100%) with our without MLO. The upload peaked at about 874Mb/s (CPU 66%).

There is also a problem with other clients. Same config (Tx power, channel, bandwidth...) and position as my u6 PRO, I couldn't get more than ~320Mb/s on my 5GHz clients literally 1.5m from the device (and this number constantly changes up and down, while the CPU is around 35-40%), while U6 PRO provided stable ~490Mb/s. 2.4GHz 802.11n clients in some places where U6 PRO gave stable 70-80Mb/s, hAP be lite barely gets ~10Mb/s (which also oscillates a lot)

Does anyone else have these performance problems with this device? Is there some specific fix, or are my 2 units simply from some bad batch? A friend of mine bought this same WiFi chip (MT7991) on a m.2 slot, and was able to get phenomenal performance with anything connected to it.

Here is a sample config, but don't take this config as a current setup, as I tried literally to change everything in here:

# 2026-08-31 12:56:04 by RouterOS 7.24.1
/interface bridge
add name=bridge1
/interface ethernet
set [ find default-name=ether1 ] l2mtu=2048
set [ find default-name=ether2 ] l2mtu=2026
set [ find default-name=ether3 ] l2mtu=2026
set [ find default-name=ether4 ] l2mtu=2026
/interface wifi channel
add band=2ghz-be name=ch-2g width=20/40mhz
add band=5ghz-be frequency=5180-5320,5500-5640 name=ch-5g width=\
    20/40/80/160mhz
/interface wifi datapath
add bridge=bridge1 disabled=no name=DpLocal
/interface wifi security
add authentication-types=wpa2-psk,wpa3-psk disabled=no ft=no ft-over-ds=no \
    name=sec-roam
/interface wifi steering
add 2g-probe-delay=yes disabled=no name=steer-roam neighbor-group=\
    dynamic-XXXXXXX-abc77de7 rrm=yes wnm=yes
/interface wifi configuration
add channel=ch-2g country=Superchannel datapath=DpLocal disabled=no \
    interworking.realms-raw="" mode=ap name=cfg-2g security=sec-roam ssid=\
    XXXXXXX steering=steer-roam
add channel=ch-5g country=Superchannel datapath=DpLocal disabled=no \
    interworking.realms-raw="" mode=ap name=cfg-5g security=sec-roam ssid=\
    XXXXXXX steering=steer-roam
/interface wifi
add configuration=cfg-5g configuration.mode=ap interworking.realms-raw="" \
    mld-name=mld-d0ea11c8d9b6 name=mld1 \
    steering.neighbor-group=dynamic-XXXXXXX-abc77de7
# managed by CAPsMAN YY:YY:YY:YY:YY:YY%bridge1, traffic processing on CAP
# mode: AP, SSID: XXXXXXX, channel: 2437/be/eC
set [ find default-name=wifi1 ] configuration=cfg-2g configuration.manager=\
    capsman .mode=ap disabled=no interworking.realms-raw="" mld-interface=\
    mld1
# managed by CAPsMAN YY:YY:YY:YY:YY:YY%bridge1, traffic processing on CAP
# mode: AP, SSID: XXXXXXX, channel: 5320/be/eeeeeeeC
set [ find default-name=wifi2 ] channel.frequency=5180-5320,5500-5640 \
    configuration=cfg-5g configuration.manager=capsman .mode=ap disabled=no \
    interworking.realms-raw="" mld-interface=mld1
/interface bridge port
add bridge=bridge1 interface=ether1
add bridge=bridge1 interface=ether2
add bridge=bridge1 interface=ether3
add bridge=bridge1 interface=ether4
add bridge=bridge1 interface=wifi1
add bridge=bridge1 interface=wifi2
add bridge=bridge1 interface=dynamic
/interface wifi cap
set discovery-interfaces=bridge1 enabled=yes
/ip dhcp-client
add default-route-tables=main interface=bridge1 name=client1
/system clock
set time-zone-name=Europe/Zagreb
/system identity
set name=hAP_BE_lite_1

At least make sure there are no "" set, like in /interface wifi.

interworking reals-raw="" is caused by using Winbox 3.

I see roughly the same performance on my be lite with 7.24.1.
In the rc versions 7.24rc3 and 7.24rc4, performance was better with around 800mbit/s download, sometimes peaking higher.
I hope performance will be increased with future software updates. At least it is stable with 7.24.1 - earlier versions of router OS had issues with sporadic kernel crashes and subsequent device reboots.

Yes, but this doesn't really make any difference. I did this final config yesterday around midnight, and simply didn't clean it up.

There are also some other things wrong, for example I can't access anything via layer 2, for some reason my printer connects to the wifi and gets an IP address via DHCP, but I can't access it.

Hopefully future updates will bring this device to its potential.

If wifi cannot exceed gigabit speed, the 2.5G port is kind of useless.

You could exceed 1g by using wifi+wired clients in parallel. I tried that, and achieved something slightly over 1gbit/s in total. The problem is that high wifi load leads to high cpu load, which has an impact on throuput of wired devices as well.
So yes, in current status of routerOS performance, the 2.5gbit/s port seems pretty useless. It only really has an effect when multiple wired clients need the full 1gbit/s performance in parallel, while wifi is not too loaded such that there are enough cpu ressources. Sounds like a rather theoretical benefit...

have you experience any problem like random packet loss on 5ghz on hap be lite?

How do you notice this? If you briefly explain how you experience the issue, I could try to reproduce it.

I didn't really notice anything, but then again I didnt really check for it specifically nor do I play online games where I would really see it.

Same issue here. Tried different configs, I get around 500Mbps down max and 900+Mbps up an all WiFi devices

Which routerOS version do you use?

have 2 of these for testing, they're definitely not ready for production

One of the issues is power related. At face value USB-C is great but the problem is how do I know if the power supply and cable are up to the task? I've tried a few that definitely aren't and causes the router to crash and reboot. But how do I know for certain? I don't. And it absolutely will have future problems of people swapping in random cables or power adapters. This wasn't as big of a deal with 2.1mm DC plugs because you can read the label, but USB-C almost nobody does they just plug things in and expect it works. The big hidden variable is cable quality because a lot of them are utter shit and cannot supply enough current

We really need a way to see voltage/current and have things logged. Warning log messages during operation as well as upon a crash/reboot that alert if the was voltage sag or not enough current to keep it stable. If we can't get this data due to lack of sensors, then MikroTik should implement a stress test feature that causes it to draw absolute maximum power for a few minutes to see if it crashes

The other thing is wifi is pretty hit and miss. Apple devices in particular seem to really struggle with this, they work ok for a few minutes or even a couple hours but then they either disconnect or just stop passing data. I don't know if its an apple or mikrotik issue but it remains unusable for them

In regards to wifi7, i'm still learning this. It's early days but is there currently any way to see actual packet throughput of each radio when they are in the MLD group? I want to know whats happening with MLO but currently I have no idea if a device is actually using 2.4ghz, 5ghz or both. I can see if they are connected, but I can't see which interfaces are actually passing data so I have no visibility

image

So you need a 12W 5V Power Supply.
Better leave some headroom so lets say max. 15W.
And cables already come rated with "Wattage" like 60W/80W/100W.

And luckily this results in a Raspberry Pi Power Supply. Or just use the included one.

If you get long USB-C Cables buy quality ones :slight_smile:

I'm on stable 7.24.5, both ROS and firmware

Interesting. I get ~700mbit/s down with iperf3 on the same routeros Version.

I’ve got a hAP be lite running RouterOS v7.25rc1 at home for testing and everything works great. I don’t use MLO and my Wi-Fi is set to WPA2-PSK AES only, but… IoT devices work without any issues, as does everyday use. I even tried playing Counter-Strike 2 over the 2.4 GHz Wi-Fi just for a test, and there wasn’t a single lag in a couple of hours.

I upgaded to v7.25rc1 too (both ROS and firmware) with a clean config: a bridge, internet from other Mikrotik on eth1, wifi2 (5GHz) only at WPA2-PSK CCMP. No firewall, no queue, nothing extra.

All wifi devices (some WiFi 7 some WiFi 6) reach 300Mbps download max and CPU at that time is 70-90%. Tested more channels, more configs (WiFi 6, 20/40), all same result.

I've opened a support ticket, but so far, my experience has been pretty disappointing and I can't return the device.

ROS 7.26beta3

I am considering buying 3 hAP be lite to use as cAPs, can someone please confirm if the wireless performance has improved?

We tried some more hap be lite configurations at work, tried different ROS versions (7.25rc1, 7.26beta3).

But results are bad. We can't get any meaningful throughput on any config iteration, and even found some weird bugs like the one where only one client is able to connect, or when the throughput on ~2gb/s datarate is <15Mb/s

So for now, I would say this device is still in its pre-alpha state and not ready for production.

The device has wifi7 but has a sub wifi6 performance as it seems.

I got a hAP ax2 yesterday and the best Wi-Fi download speed (fast.com) I could see was 830-840 MBit/s (edit: peak 860Mbit/s on a MbPro M1 and 920 MBit/s with an iPhone 18 Pro) which makes me think that device is much more powerful than the be lite.

The ax2 also has a 64 bit cpu and twice the ram. I see no reason to get a be lite over an ax2, the price is also almost identical.