R7800 performance

There is no performance difference. (or there should not be...)

My currently latest 19.07 build (of commit e7f1313b) is a few commits ahead of 19.07.2 but nothing related to performance.
https://git.openwrt.org/?p=openwrt/openwrt.git;a=shortlog;h=e7f1313bbbf44e529e5de40045b191e026af6b7b

And there are no performance tweaks in my build.

1 Like

With Samsung EVO850 SSD I got read speed about 25MB/s, write speed about 20MB/s. ntfs-3g process takes about 20% cpu load. Maybe with other FS speed will be better because NTFS makes more load than others AFAIK.

I have used this router on a 300/300 link for a few months now and based on what I am observing, this is its ceiling for real life use cases: it can only deliver ~300Mbps combined throughput WIFI to WAN (I ran several tests downloading lots of data from one cloud and immediately uploading the files to another to confirm that). Both router cores are at ~95% while only uploading or only downloading or combined. Needless to say, that the router from my ISP can handle bi-directional traffic at full speed.
No QoS or SQM and software flow offload is enabled. Does anyone observe faster speeds?

I don't know if you have tried balancing the irq's, this make a noticable improvement on the linksys series. You could also try observing the throughput by forwarding packets from the lan port to the wan port bidirectionally using iperf and two computers if possible via ethernet.

EDIT: I looked over the part where you mentioned wan throughpout to be expected.

The CPUs are already evenly loaded without that, but I will try balancing just in case.

This is a not a real life use case, but LAN-to-WAN without NAT, etc can get close to 800Mpbs. But that is not useful.

Someone does...

I edited my post with a notation, but yes I did overlook that part of just wifi > wan. You can't expect much from wifi, about ~60% over tcp. Blast it with udp and see the difference.

Speed tests have changed over the last couple weeks due to COVID, speedtest has been tracking the changes, most places are still seeing degradation in service. Speed tests have been varying every day. I wouldn’t be too frustrated by speed tests for the next couple weeks until things get a little more normal.

Wifi speed test from 2 weeks ago:

http://www.dslreports.com/speedtest/61437431

Recently I’ve been running the following tweaks on the latest master build (I’ve been tweaking CPU, Irqbalance, beacon interval, etc) - see if they change your results:


uci set network.globals.packet_steering=1; uci commit network


uci set irqbalance.irqbalance.enabled=1; uci commit


echo 35 > /sys/devices/system/cpu/cpufreq/ondemand/up_threshold; echo 10 > /sys/devices/system/cpu/cpufreq/ondemand/sampling_down_factor

Here are how my APs are set up currently for their wifi settings:

1 Like

I just recently did tests with a R7800 and a AX200 card, wifi to wan was about 600Mbps with qos enabled.
Not using a hnyman build though.

Well, I am using 4.14 vs 4.19 in the test, but for me SQM is not even an option: it tops up <200Mbps. I will try snapshot again, but the wireless stability was poor last time I tried.

I am not using speediest: I am just observing the download/upload rate while running the test: only download, only upload, or both. Each of the three show the same combined throughput.

Thx, I have tried them all.

Post up your cat /etc/config/wireless and a speedtest

Are you using different 2.4ghz and 5 ghz SSIDs? What type of wifi client are you testing with? Maybe there is something subtlety different.

config wifi-device 'radio0'
	option type 'mac80211'
	option hwmode '11a'
	option path 'soc/1b500000.pci/pci0000:00/0000:00:00.0/0000:01:00.0'
	option htmode 'VHT80'
	option legacy_rates '0'
	option country 'US'
	option channel 'auto'
	option short_preamble '1'
	option txpower '23'
	option channels '36 40 44 48 149 153 157 161 165'
	option beacon_int '101'
	option disabled '0'

config wifi-device 'radio1'
	option type 'mac80211'
	option hwmode '11g'
	option path 'soc/1b700000.pci/pci0001:00/0001:00:00.0/0001:01:00.0'
	option htmode 'HT20'
	option country 'US'
	option legacy_rates '0'
	option channel 'auto'
	option short_preamble '1'
	option txpower '21'
	option channels '1 6 11'
	option beacon_int '103'
	option disabled '0'

No, but I make sure that the client is connected to the 5GHz AP before testing.

A few years old MacBook Pro.

Some of the older macOS clients don’t support AC. I recently retired a Mac laptop with a wireless N adapter (don’t know the age of yours. For me the wireless speed...slowness...was starting to become very noticeable):

Apple has a good write up on macOS clients. I’d evaluate your environment settings to get the best balance for your clients.

Especially: “macOS always defaults to the 5 GHz band over the 2.4 GHz band. This happens as long as the RSSI for a 5 GHz network is -68 dBm or better.”

Try this:

  1. Doesn’t take many obstacles for 5ghz to drop to -68 dBm. I’ve found it better to split the 5 and 2.4ghz SSIDs (and have my higher performance clients forget the 2.4ghz SSID) so that I retain the higher performance of 5ghz. There is much more bandwidth and less interference from household items on 5ghz.

  2. do a survey of your local area and pick a channel rather than going with Auto. Pick a 80Mhz wide non DFS channel - 36 or 161.

  3. if you have other 5ghz clients try fast.com speedtest.net or http://www.dslreports.com/speedtest to compare against your macOS device

you should max out your 300/300 connection easily via wifi.

Thx, but I already did all these basics steps without any improvements. My laptop does support AC and the auto channel works great and does pick the best channel, even though I tried setting different channels manually.

As far as I know, there is no real automated channel selection in OpenWrt at this time. The first allowed channel in your regulatory country setting is selected in "auto" e.g. channel 36.

Also the number of 80 MHz channels is even more limited. See the non DFS 80 MHz 11ac channels: https://www.smallnetbuilder.com/wireless/wireless-features/33210-160-mhz-wi-fi-channels-friend-or-foe

That is not correct: I have it enabled and it picks other channels when the conditions are right. I just added DFS and it picked the channel form the DFS range.

That is why I added DFS channels to experiment.

What are people's performance comparisons between OEM stock firmware and the latest OpenWRT?

For stock OEM I get wired to wired performance of ~940mbps.
For wireless I get 629mbps from an ax200 client to wired server on lan switch.

Any numbers for OpenWRT?

A half of that...

I really don't understand?
Wired to r7800 (serving as a dumbap with activity) as iperf3 server(no -P):

[ ID] Interval           Transfer     Bitrate
[  5]   0.00-1.00   sec   111 MBytes   932 Mbits/sec
[  5]   1.00-2.00   sec   111 MBytes   933 Mbits/sec
[  5]   2.00-3.00   sec   111 MBytes   929 Mbits/sec
[  5]   3.00-4.00   sec   111 MBytes   933 Mbits/sec
[  5]   4.00-5.00   sec   111 MBytes   931 Mbits/sec
[  5]   5.00-6.00   sec   111 MBytes   933 Mbits/sec
[  5]   6.00-7.00   sec   111 MBytes   929 Mbits/sec
[  5]   7.00-8.00   sec   111 MBytes   927 Mbits/sec
[  5]   8.00-9.00   sec   111 MBytes   927 Mbits/sec
[  5]   9.00-10.00  sec   111 MBytes   927 Mbits/sec
[  5]  10.00-10.00  sec   141 KBytes   910 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate
[  5]   0.00-10.00  sec  1.08 GBytes   930 Mbits/sec                  receiver

-P 8 on my android phone gets:

[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec   115 MBytes  96.9 Mbits/sec  777             sender
[  5]   0.00-10.01  sec   115 MBytes  96.7 Mbits/sec                  receiver
[  7]   0.00-10.00  sec  95.3 MBytes  80.0 Mbits/sec  650             sender
[  7]   0.00-10.01  sec  95.3 MBytes  79.8 Mbits/sec                  receiver
[  9]   0.00-10.00  sec   113 MBytes  94.4 Mbits/sec  736             sender
[  9]   0.00-10.01  sec   113 MBytes  94.2 Mbits/sec                  receiver
[ 11]   0.00-10.00  sec   104 MBytes  87.0 Mbits/sec  337             sender
[ 11]   0.00-10.01  sec   104 MBytes  86.7 Mbits/sec                  receiver
[ 13]   0.00-10.00  sec   104 MBytes  86.9 Mbits/sec  742             sender
[ 13]   0.00-10.01  sec   103 MBytes  86.5 Mbits/sec                  receiver
[ 15]   0.00-10.00  sec  97.7 MBytes  82.0 Mbits/sec  305             sender
[ 15]   0.00-10.01  sec  97.4 MBytes  81.6 Mbits/sec                  receiver
[ 17]   0.00-10.00  sec   105 MBytes  87.8 Mbits/sec  949             sender
[ 17]   0.00-10.01  sec   104 MBytes  87.4 Mbits/sec                  receiver
[ 19]   0.00-10.00  sec   106 MBytes  88.9 Mbits/sec  387             sender
[ 19]   0.00-10.01  sec   105 MBytes  88.4 Mbits/sec                  receiver
[SUM]   0.00-10.00  sec   839 MBytes   704 Mbits/sec  4883             sender
[SUM]   0.00-10.01  sec   837 MBytes   701 Mbits/sec                  receiver

Now try it with NAT, and watch your numbers be magically sawed in half like a magicians assistant.

2 Likes