MR44 WiFi speeds

[Iridium79]

Hey, just got an MR44 replaced the MR36 which was been outperformed by the MR33.   What speeds should I expect from the MR44 with a Wi-Fi 6 client.   I can’t seem to break the 400mbps down and 500mbps up, now this is with the client been literally 6 feet away from the access point.   I also tried my MR53 from the same distance and I am getting close to 520mbps down and 658mbps up.   I also have a beta unit from UniFi (UniFi 6Pro) and I am getting these results: 789mbps down and 831mbps up.   All tests have been run separately with no other access point in range and same location and client.  Any advice or help will be appreciated.  I can’t really compare the MR53 with the MR44 as the MR53 is a 4x4 on both 2.4ghz and 5ghz.   Now the UniFi 6 Pro is Wi-Fi 6 only on the 5Ghz which has a 4x4 and the 2.4ghz is only 2x2 which is not Wi-Fi 6 according to the spec sheet.   

[ww]

Does unify run 160mhz? 

Do you have the merak at 80Mhz?

Your client is 2x2? 3x3?

What is the client phy  rate when you test?

[Iridium79]

@ww 

I don’t have the UniFi at 160.

Meraki is at 80Mhz

client is a 2x2

physical rate 1200mbps according to the local status page of the meraki

[ww]

I know mr46 also was around 800Mbps.

Dont know about the mr44 .. but i would expect  also rates near 700-800 with those specs.

Phy rate you have to check at the client when testing. Meraki only report max capable

[GIdenJoe]

Since I have not yet heard of a 3x3 AX client we need to assume the client will be a 2x2 AX.

Looking at the MCS table for an 80 MHz wide channel, if that is how you are actually testing it you should have 3 possible PHY values:

Please check if your clients effectively says one of these data rates.

1200Mbps = 2x2 MCS 11, 80 MHz, 0.8µsec GI

1134Mbps = 2x2 MCS 11, 80 MHz, 1.6µsec GI

1020Mbps = 2x2 MCS 11, 80 MHz, 3.2µsec GI

Then should also have an Ekahau sidekick standing by to capture and live monitor the waves when you start your transfer and see if all channels are nicely used at 80% or a bit more.

If both of your scenario's are exactly right, then you should have a fair test between both access points.

Since your effectief throughput will be around 2/3rds to 3/4ths of the physical rate, this should possible give a 800 Mbps of throughput.

[NolanHerring]

What is the client in question? Is it an Intel card? I would try updating the driver if it is.

In addition, is your throughput test 'through' the AP to a hardwired iPERF server or are you access the local status page and running the test to that?  (try the local status page if not and see what you get).

What firmware version is the MR44 on and can you confirm your 80MHz channel is 'clean' or being used by other AP's nearby etc.

[Iridium79]

@NolanHerring 

Client in question is an iPhone 12 Pro and an a Lenovo X1 9th Gen with the intel Wifi 6 AX 201 160Mhz according to the properties.  On the Lenovo X1 the speed fluctuates from 861Mbps to 1.2Gbps.

On the iPhone 12 I cant tell.

Im running the Beta Firmware of 28.99 which is equivalent to the latest beta.   And def im on the 80MHz channel.  The channel im on is clear.

[NolanHerring]

Hi all, decided to lab this myself at home and see what results I got.

See below:

As a test, I accessed the local status page of the MR44 and only got ~300Mbps running that test. I'm under the impression the local throughput test is limited and best effort, and not really able to show you what it is actually capable of. See below.

Firmare Version - 28.2

Intex AX200 Driver Version 22.50.1.1

Laptop was connected at a rate of 1201Mbps (reference here: http://mcsindex.com/ - 2x2 client at 80MHz on AX using short guard interval (0.8)).

First I made sure my MR44 was on a clean channel, set to 80MHz (14dBm, laptop is only 5 feet away).

Then I ran the iPerf test on my laptop to a hardwired computer (iPerf server). The MR44 and the server computer are connected to the same switch, both at 1Gbps connections. Test ran for 5 minutes but here is a snippet.

Laptop NIC showing same results

The MR44 AP was also showing similar results on its Summary page. I also included the spectrum analysis from an MR32 nearby.

I also pulled out my Ekahau Sidekick to get a full spectrum capture:

Interesting note, the MR32 above showed 'correct' utilization on its spectrum page. The MR44 and another MR45 nearby, did not show as much and what they did show wasn't even on the correct UNII band. Pretty sure spectrum analysis is 'broken' on the Wi-Fi 6 models.

As far as I can see, the MR44 is performing as expected given that my port is 1Gbps, half-duplex nature of 802.11, and management overhead.

I hope this helps.

-Nolan

[NolanHerring]

I forgot to also upload the 3 different sources of reporting for channel utilization (around 80 to 85%). Might be a little more juice yet to be squeezed out of her.

[WB]

Any particular reason why you're running the beta firmware? Can this issue be replicated when running v27.7.1? 

[NolanHerring]

No reason, just happened to be on that. I'll downgrade to 27.7.1 and test again.

[redsector]

I did an testing with my iPhone8 (no AX) and the Fritz WLAN App. I checked with MR44, MR44E, MR46, MR46E, MR52 accesspoints in the same network: with firmware MR26.8 I got no more than 100MBit, with firmware 27.7.1 I get 50% more speed up to 150MBit.

[yepper]

In my scenrario I couldn't break 150mb on wifi when the AP itself could speedtest at 350+. Interestingly enough when I turned off meraki standard traffic shaping for the SSID I'm using as my test case, speeds doubled or came close to max for 2x2 streams. This includes testing at fast.com, speedtest.net and other major testing sites.

Hopefully that helps others as part of their troubleshooting...