Ath79 builds with all kmod packages through opkg [flow offloading]

dalanmiller · August 16, 2018, 12:33pm

Curious what speeds people are getting from this build on their Archer C7?

My setup as follows:

EdgeRouter X => Archer C7 => Macbook Pro

Signal:

And then this is the iperf3 results from my Macbook to the EdgeRouter X.

iperf3 -c 192.168.1.1
Connecting to host 192.168.1.1, port 5201
[  5] local 192.168.1.89 port 53298 connected to 192.168.1.1 port 5201
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec  33.7 MBytes   283 Mbits/sec
[  5]   1.00-2.00   sec  33.2 MBytes   278 Mbits/sec
[  5]   2.00-3.00   sec  34.2 MBytes   286 Mbits/sec
[  5]   3.00-4.00   sec  35.0 MBytes   293 Mbits/sec
[  5]   4.00-5.00   sec  34.3 MBytes   288 Mbits/sec
[  5]   5.00-6.00   sec  34.3 MBytes   287 Mbits/sec
[  5]   6.00-7.00   sec  29.7 MBytes   249 Mbits/sec
[  5]   7.00-8.00   sec  30.2 MBytes   253 Mbits/sec
[  5]   8.00-9.00   sec  31.5 MBytes   264 Mbits/sec
[  5]   9.00-10.00  sec  31.6 MBytes   265 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-10.00  sec   328 MBytes   275 Mbits/sec                  sender
[  5]   0.00-10.00  sec   327 MBytes   275 Mbits/sec                  receiver

It seems like I should be seeing 3X the bitrate from my laptop to my EdgeRouter, curious if there's anything people have done to increase their speed?

Pilot6 · August 16, 2018, 1:48pm

This looks OK, you can read in any wiki that the channel 702 speed is not very much related to a real speed.

kofec · August 16, 2018, 2:08pm

Hi,
In ath79 there are: TP-Link TL-WR740N v1/v2 and TP-Link TL-WR741N/ND v1/v2 . I have such device but a little worry before upgrade. Anyone has any experience with ath79 for tiny device ?
@juppin
Is there any reason that you didn't build it ? You just focus on "normal" not "tiny" ?

Pilot6 · August 16, 2018, 2:12pm

I have with MR3040v2. Works well.
These builds are with many extra packages, that is overkill for tiny devices.

frauhottelmann · August 16, 2018, 8:05pm

The build for the Unified AP AC Lite is working nicely

Thanks

maxspeed3006 · August 16, 2018, 10:31pm

Any plans on supporting the Archer C7 V4?

djgreg13 · August 17, 2018, 6:11pm

Very great passthrouth
Source IP : 192.168.10.103 NAT masquerade into 192.168.1.21
$ iperf3 -c 192.168.1.17
Connecting to host 192.168.1.17, port 5201
[ 5] local 192.168.10.103 port 55753 connected to 192.168.1.17 port 5201
[ ID] Interval Transfer Bitrate
[ 5] 0.00-1.00 sec 110 MBytes 923 Mbits/sec
[ 5] 1.00-2.00 sec 108 MBytes 904 Mbits/sec
[ 5] 2.00-3.00 sec 113 MBytes 948 Mbits/sec
[ 5] 3.00-4.00 sec 110 MBytes 921 Mbits/sec
[ 5] 4.00-5.00 sec 110 MBytes 923 Mbits/sec
[ 5] 5.00-6.00 sec 112 MBytes 941 Mbits/sec
[ 5] 6.00-7.00 sec 110 MBytes 926 Mbits/sec
[ 5] 7.00-8.00 sec 112 MBytes 941 Mbits/sec
[ 5] 8.00-9.00 sec 111 MBytes 928 Mbits/sec
[ 5] 9.00-10.00 sec 110 MBytes 922 Mbits/sec

[ ID] Interval Transfer Bitrate
[ 5] 0.00-10.00 sec 1.08 GBytes 928 Mbits/sec sender
[ 5] 0.00-10.00 sec 1.08 GBytes 928 Mbits/sec receiver

iperf Done.

With 18.06.0 official, paththrouth at 600 Mbit/s

With a speed test from Orange FR Gigabit fiber (1000/200)

Great Archer C7, old but stay in the place

Edit:

From WAN to LAN with port forwarding and INPUT rules
Accepted connection from 192.168.1.17, port 51521
[ 5] local 192.168.10.103 port 5201 connected to 192.168.1.17 port 51522
[ ID] Interval Transfer Bitrate
[ 5] 0.00-1.00 sec 99.7 MBytes 836 Mbits/sec
[ 5] 1.00-2.00 sec 102 MBytes 858 Mbits/sec
[ 5] 2.00-3.00 sec 103 MBytes 868 Mbits/sec
[ 5] 3.00-4.00 sec 103 MBytes 867 Mbits/sec
[ 5] 4.00-5.00 sec 104 MBytes 871 Mbits/sec
[ 5] 5.00-6.00 sec 104 MBytes 870 Mbits/sec
[ 5] 6.00-7.00 sec 103 MBytes 865 Mbits/sec
[ 5] 7.00-8.00 sec 104 MBytes 872 Mbits/sec
[ 5] 8.00-9.00 sec 104 MBytes 872 Mbits/sec
[ 5] 9.00-10.00 sec 92.5 MBytes 776 Mbits/sec
[ 5] 10.00-10.03 sec 3.24 MBytes 907 Mbits/sec

[ ID] Interval Transfer Bitrate
[ 5] 0.00-10.03 sec 1023 MBytes 856 Mbits/sec receiver

Server listening on 5201

dalanmiller · August 18, 2018, 12:29pm

Is that directly from a client machine via wifi to the AP? Mind sharing what your firewall settings are?

juppin · August 18, 2018, 1:18pm

I don´t think that here is a wireless connection involved...
Have you ever seen such real speeds on a wireless link also if the theoretical throughput is 1300 Mbit/s?

djgreg13 · August 18, 2018, 2:02pm

I did the test with RJ45 cable

For the first test :
iperf -s wan side
iperf -c lan side

For WAN to LAN test :
iperf -c wan side
iperf -s lan side

wan zone is drop by default

config redirect
	option target 'DNAT'
	option src 'wan'
	option dest 'lan'
	option proto 'tcp udp'
	option dest_ip '192.168.10.103'
	option dest_port '5201'
	option name 'test-iperf'
	option enabled '1'

config defaults
	option syn_flood '1'
	option input 'ACCEPT'
	option output 'ACCEPT'
	option forward 'REJECT'
	option flow_offloading '1'
	option flow_offloading_hw '1'

config zone
	option name 'wan'
	option output 'ACCEPT'
	option forward 'REJECT'
	option masq '1'
	option network 'wan wan6 WWAN6 wwan'
	option input 'DROP'

psyborg · August 18, 2018, 6:34pm

should be possible. killer guys say they achieve 1200Mbps on 1700Mbps PHY

Gigabit · August 18, 2018, 8:29pm

I don't think these builds will improve/change WiFi performance, so I don't think those who are complaining have issues with the firmware but perhaps the router they are using or their environment, etc.?

gr1m · August 18, 2018, 10:53pm

Mostly enviroment, unless they came from a build with reghack like patches. Then again, it shouldn't change max throughput.

juppin · August 18, 2018, 10:59pm

Not directly, but with enabled flow offloading there is at least more available cpu time for the cpu hungry hostapd service.

Gigabit · August 18, 2018, 11:15pm

@juppin Is there any chance of a light build with only the essentials and none of the extras added, basically what a trunk build would be? Maybe LuCI added?

Sorry if that's a lot of work but I prefer to add packages from a baseline of nearly nothing, just how I roll!

chros · August 19, 2018, 7:51am

I compiled ath79 for Archer C7 v2 from master 2 weeks ago and had just time time play with it.
Software flow offloading doesn't work with sqm (cake) + pppoe (checking with htop):

if I disable sqm flow offloading works

Any ideas? Thanks

slh · August 19, 2018, 8:03am

Neither (software-) flow-offloading nor sqm are overly hardware specific (beyond endianess), so these issues are probably not ath79 specific and better covered in a generic thread.

In general, "Software flow offloading doesn't work with sqm (cake) + pppoe" isn't really a bug report - what (exactly) doesn't work, how do you determine that it doesn't work. Enabling both SQM and software flow-offloading (this is different with hardware flow-offloading, but that's only available on ramips for now) should 'work' in the sense that you should get a functional internet connection, but the whole concept of SQM relies on it keeping close taps on the individual packets. This basically means that SQM 'wins' and that flow-offloading won't have much of an effect anymore (it's still enabled, it technically still works, but SQM won't provide it with a whole lot of flows to offload).

chros · August 19, 2018, 8:16am

Thanks, that's what I meant, having 9 MB/s download the CPU usage is ~96%. So it's expected, thanks.

juppin · August 19, 2018, 8:22am

9 MByte/s or 9 Mbit/s?

I´ve using a non PPPoE connection in combination with SQM and never reach more than 50% cpu utilization on ~28 Mbit/s download rate in my girlfriends flat...

chros · August 19, 2018, 8:28am

9 MByte/s (~74Mbit/s), so it's on pair with yours, thanks for the clarification!

Ath79 builds with all kmod packages through opkg [flow offloading]

[ ID] Interval Transfer Bitrate [ 5] 0.00-10.03 sec 1023 MBytes 856 Mbits/sec receiver

Server listening on 5201

[ ID] Interval Transfer Bitrate
[ 5] 0.00-10.03 sec 1023 MBytes 856 Mbits/sec receiver