[RT-AX88U] Wifi performance test (80 vs 160 MHz)

[RMerlin]

Just a quick test I ran tonight with iperf. Server was on a Huawei P30 phone, with the client on my desktop (connected over Ethernet).

The phone is in the same room as the RT-AX88U (maybe about 5 meters away).

5 GHz 80 MHz channel:

P:\Applications>iperf -c 192.168.10.113 -M 1400 -N -t 30
------------------------------------------------------------
Client connecting to 192.168.10.113, TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[316] local 192.168.10.100 port 2170 connected with 192.168.10.113 port 5001
[ ID] Interval       Transfer     Bandwidth
[316]  0.0-30.0 sec  1.91 GBytes    546 Mbits/sec

5 GHz 160 MHz channel

P:\Applications>iperf -c 192.168.10.113 -M 1400 -N -t 30
------------------------------------------------------------
Client connecting to 192.168.10.113, TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[316] local 192.168.10.100 port 1835 connected with 192.168.10.113 port 5001
[ ID] Interval       Transfer     Bandwidth
[316]  0.0-30.0 sec  3.06 GBytes    875 Mbits/sec

Pretty good performance difference.

 

[L&LD]

Would be interesting how this translates and correlates into transferring an actual file of 1GB or larger, in the exact same conditions.

 

[doczenith1]

@RMerlin have you tried using the iperf equivalent of the iperf3 option -w 4M? I think it may be -w 4000? When I did my testing I got the better results using -w 4M than I did using the default over wifi.

 

[RMerlin]

Would be interesting how this translates and correlates into transferring an actual file of 1GB or larger, in the exact same conditions.

Just do the math - 10240/875 = around 11 seconds.

 

[RMerlin]

@RMerlin have you tried using the iperf equivalent of the iperf3 option -w 4M? I think it may be -w 4000? When I did my testing I got the better results using -w 4M than I did using the default over wifi.

I'm sticking to this older iperf2 build and my current settings because I've been using the exact same settings for years. They might not show ideal performance (i.e. it's not multithreaded), but consistency is more important if I want to compare results over time.

 

[L&LD]

Just do the math - 10240/875 = around 11 seconds.

The math I can do, but what I wanted to see is if the actual file downloaded matched the iperf indicator.

Trying to transfer the 'science' numbers into the actual, real world.