Why is ethernet traffic slower than wifi on Asus RT-AC68U?

[Patryk]

I've been using Asuswrt-Merlin for a couple of months now and it is only recently that I've noticed that my Ethernet traffic is slower than my wifi traffic.

My setup:
192.168.1.1 - router
192.168.1.6 - pc connected via ethernet (iperf3 server)
192.168.1.48 - macbook pro connected to 5Ghz 802.11ac wifi (1300Mbps link speed)

I start iperf3 server on my pc with default settings and run a test with macbook as client:

Accepted connection from 192.168.1.48, port 50145
[  5] local 192.168.1.6 port 5201 connected to 192.168.1.48 port 50146
[ ID] Interval           Transfer     Bandwidth
[  5]   0.00-1.00   sec   102 MBytes   855 Mbits/sec
[  5]   1.00-2.00   sec   105 MBytes   881 Mbits/sec
[  5]   2.00-3.00   sec   105 MBytes   879 Mbits/sec
[  5]   3.00-4.00   sec   105 MBytes   878 Mbits/sec
[  5]   4.00-5.00   sec   101 MBytes   844 Mbits/sec
[  5]   5.00-6.00   sec  89.9 MBytes   754 Mbits/sec
[  5]   6.00-7.00   sec  95.0 MBytes   797 Mbits/sec
[  5]   7.00-8.00   sec  93.7 MBytes   786 Mbits/sec
[  5]   8.00-9.00   sec  98.1 MBytes   823 Mbits/sec
[  5]   9.00-10.00  sec  66.6 MBytes   558 Mbits/sec
[  5]  10.00-10.01  sec   711 KBytes   825 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  5]   0.00-10.01  sec  0.00 Bytes  0.00 bits/sec                  sender
[  5]   0.00-10.01  sec   961 MBytes   806 Mbits/sec                  receiver

while doing the same with router as client

Accepted connection from 192.168.1.1, port 43406
[  5] local 192.168.1.6 port 5201 connected to 192.168.1.1 port 43407
[ ID] Interval           Transfer     Bandwidth
[  5]   0.00-1.00   sec  40.9 MBytes   343 Mbits/sec
[  5]   1.00-2.00   sec  44.2 MBytes   370 Mbits/sec
[  5]   2.00-3.00   sec  37.5 MBytes   315 Mbits/sec
[  5]   3.00-4.00   sec  38.8 MBytes   325 Mbits/sec
[  5]   4.00-5.00   sec  39.8 MBytes   333 Mbits/sec
[  5]   5.00-6.00   sec  42.0 MBytes   352 Mbits/sec
[  5]   6.00-7.00   sec  43.0 MBytes   361 Mbits/sec
[  5]   7.00-8.00   sec  50.8 MBytes   426 Mbits/sec
[  5]   8.00-9.00   sec  59.5 MBytes   499 Mbits/sec
[  5]   9.00-10.00  sec  63.8 MBytes   535 Mbits/sec
[  5]  10.00-10.07  sec  4.83 MBytes   541 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bandwidth
[  5]   0.00-10.07  sec  0.00 Bytes  0.00 bits/sec                  sender
[  5]   0.00-10.07  sec   465 MBytes   387 Mbits/sec                  receiver

When running the test via Wifi with macbook I can observe that softirq cpu usage is maxed on both cores nearly 100% but while doing the same on pc I can only see 1 core being maxed on softirq.

-----------------------------------------------------------------------------------------------------

On the other hand when running iperf3 server on the router I get the following results with 1 core softirq maxed in both cases:
* when launching test via wifi with macbook

Accepted connection from 192.168.1.48, port 50188
[  5] local 192.168.1.1 port 5201 connected to 192.168.1.48 port 50189
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec  25.6 MBytes   214 Mbits/sec
[  5]   1.00-2.00   sec  26.7 MBytes   224 Mbits/sec
[  5]   2.00-3.00   sec  27.2 MBytes   227 Mbits/sec
[  5]   3.00-4.01   sec  26.3 MBytes   219 Mbits/sec
[  5]   4.01-5.00   sec  26.9 MBytes   227 Mbits/sec
[  5]   5.00-6.00   sec  25.6 MBytes   214 Mbits/sec
[  5]   6.00-7.00   sec  27.0 MBytes   227 Mbits/sec
[  5]   7.00-8.01   sec  27.0 MBytes   225 Mbits/sec
[  5]   8.01-9.00   sec  27.4 MBytes   231 Mbits/sec
[  5]   9.00-10.01  sec  27.9 MBytes   234 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-10.01  sec   268 MBytes   224 Mbits/sec                  receiver

* when launching the test with pc

Accepted connection from 192.168.1.6, port 54326
[  5] local 192.168.1.1 port 5201 connected to 192.168.1.6 port 54328
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.01   sec  36.7 MBytes   306 Mbits/sec
[  5]   1.01-2.00   sec  36.1 MBytes   304 Mbits/sec
[  5]   2.00-3.01   sec  36.6 MBytes   306 Mbits/sec
[  5]   3.01-4.01   sec  35.1 MBytes   295 Mbits/sec
[  5]   4.01-5.00   sec  32.7 MBytes   275 Mbits/sec
[  5]   5.00-6.00   sec  36.5 MBytes   306 Mbits/sec
[  5]   6.00-7.01   sec  36.7 MBytes   307 Mbits/sec
[  5]   7.01-8.01   sec  37.5 MBytes   313 Mbits/sec
[  5]   8.01-9.00   sec  37.2 MBytes   315 Mbits/sec
[  5]   9.00-10.00  sec  36.3 MBytes   305 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-10.00  sec   362 MBytes   303 Mbits/sec                  receiver

I do have to note that I have NAT Acceleration on auto

Any idea why do I get this limited bandwith on ethernet?

 

[coxhaus]

You might try it with Jumbo frames turned off. I would never run jumbo frames but I have never run an ASUS router.

 

[ColinTaylor]

Yes, I noticed exactly the same thing with my RT-AC68U. I even had a discussion about it with john9527 last year.

Unfortunately there's nothing you can do about it as it's limited by the router's CPU.

Do note however that this only applies to traffic going between the router and the LAN. LAN to LAN traffic isn't effected because it's switched.

LAN to WAN traffic can take advantage of hardware acceleration which means that it doesn't suffer the same CPU limitations.

 

[coxhaus]

Wow, wire is faster than wireless and has more bandwidth. They must of got something wrong.

 

[ColinTaylor]

Wow, wire is faster than wireless and has more bandwidth. They must of got something wrong.

No that's not what it says.

Traffic that stays on the router's switch, either wired or wireless (because the APs are connected directly to the switch), is fast.

Traffic that leaves the switch and has to be processed by the router's CPU (i.e. router-to-switch or switch-to-router) is CPU-limited.

 

[Patryk]

You might try it with Jumbo frames turned off. I would never run jumbo frames but I have never run an ASUS router.

I've just turned jumbo frames on today. Without them the results were similar.

 

[L&LD]

I've just turned jumbo frames on today. Without them the results were similar.

I don't see this as limited bandwidth on Ethernet. I see this as the router not being powerful enough to give valid results.

Running speed tests on/from the router to a device is a flawed process to begin with because the router hardware (CPU, RAM) is extremely limited when it also has to do everything else too (like routing, for example). As I found out with a 1Gbps up/down ISP Fibre connection and trying to use the spdMerlin script by Jack Yaz.

You need to test from between two wired computers instead. Not from wired to wireless (and vice versa), or, from the router to a wired or wireless client.

 

[sfx2000]

You might try it with Jumbo frames turned off. I would never run jumbo frames but I have never run an ASUS router.

Jumbo shouldn't matter here... if anything, it would improve bandwidth.

 

[sfx2000]

You need to test from between two wired computers instead. Not from wired to wireless (and vice versa), or, from the router to a wired or wireless client.

Yes...

MacMini 2012 over ethernet to NUC7i5 running Ubuntu... seeing expected results here...

MacMini 2012 is basically a MacBook Pro without a screen... same wifi as MacBook Pro 2012 (802.11n)

949 Mbits/sec - sender
941 Mbits/sec - receiver

builder@builder:~/builds/openwrt$ iperf3 -c 192.168.1.20 -Z -P10
Connecting to host 192.168.1.20, port 5201
[  4] local 192.168.1.113 port 33988 connected to 192.168.1.20 port 5201
[  6] local 192.168.1.113 port 33990 connected to 192.168.1.20 port 5201
[  8] local 192.168.1.113 port 33992 connected to 192.168.1.20 port 5201
[ 10] local 192.168.1.113 port 33994 connected to 192.168.1.20 port 5201
[ 12] local 192.168.1.113 port 33996 connected to 192.168.1.20 port 5201
[ 14] local 192.168.1.113 port 33998 connected to 192.168.1.20 port 5201
[ 16] local 192.168.1.113 port 34000 connected to 192.168.1.20 port 5201
[ 18] local 192.168.1.113 port 34002 connected to 192.168.1.20 port 5201
[ 20] local 192.168.1.113 port 34004 connected to 192.168.1.20 port 5201
[ 22] local 192.168.1.113 port 34006 connected to 192.168.1.20 port 5201
[ ID] Interval           Transfer     Bandwidth       Retr  Cwnd
[  4]   0.00-1.00   sec  12.2 MBytes   102 Mbits/sec    0    161 KBytes       
[  6]   0.00-1.00   sec  12.0 MBytes   101 Mbits/sec    0    134 KBytes       
[  8]   0.00-1.00   sec  12.0 MBytes   101 Mbits/sec    0    134 KBytes       
[ 10]   0.00-1.00   sec  12.0 MBytes   101 Mbits/sec    0    129 KBytes       
[ 12]   0.00-1.00   sec  12.1 MBytes   101 Mbits/sec    0    163 KBytes       
[ 14]   0.00-1.00   sec  12.0 MBytes   101 Mbits/sec    0    126 KBytes       
[ 16]   0.00-1.00   sec  12.0 MBytes   100 Mbits/sec    0    132 KBytes       
[ 18]   0.00-1.00   sec  12.1 MBytes   101 Mbits/sec    0    144 KBytes       
[ 20]   0.00-1.00   sec  12.0 MBytes   100 Mbits/sec    0    126 KBytes       
[ 22]   0.00-1.00   sec  12.0 MBytes   100 Mbits/sec    0    140 KBytes       
[SUM]   0.00-1.00   sec   120 MBytes  1.01 Gbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   1.00-2.00   sec   113 MBytes   952 Mbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   2.00-3.00   sec   113 MBytes   946 Mbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   3.00-4.00   sec   112 MBytes   937 Mbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   4.00-5.00   sec   113 MBytes   946 Mbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   5.00-6.00   sec   112 MBytes   941 Mbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   6.00-7.00   sec   112 MBytes   941 Mbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   7.00-8.00   sec   112 MBytes   940 Mbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   8.00-9.00   sec   112 MBytes   942 Mbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   9.00-10.00  sec   112 MBytes   939 Mbits/sec    0             
- - - - - - - - - - - - - - - - - - - - - - - - -
[SUM]   0.00-10.00  sec  1.11 GBytes   949 Mbits/sec    0             sender
[SUM]   0.00-10.00  sec  1.10 GBytes   941 Mbits/sec                  receiver

 

[coxhaus]

Jumbo shouldn't matter here... if anything, it would improve bandwidth.

Jumbo frames are going to make the router internally buffer and fragment frames because it is feeding 9000 bytes frames into a bunch of 1518 frames unless it is all local traffic which I kind of doubt.

 

[ColinTaylor]

Jumbo frames are going to make the router internally buffer and fragment frames because it is feeding 9000 bytes frames into a bunch of 1518 frames unless it is all local traffic which I kind of doubt.

It is all local traffic. This is clearly explained in post #1.

Regardless, this has nothing to do with the original question which I answered in post #3 and #5. Maybe I wasn't clear enough.

 

[sfx2000]

Jumbo frames are going to make the router internally buffer and fragment frames because it is feeding 9000 bytes frames into a bunch of 1518 frames unless it is all local traffic which I kind of doubt.

Most, if not all, switches can deal with this - and if OP is on the wire, the Broadcom chip can work it...