A Wireguard comparison DB

[evs]

Three older AP/devices plus a rtl8380 switch. I think the only other thing I have that is a different SOC to what's on the list is like a dgnd3700v1 haha.

MSM460/MSM466: p1020 @ 800MHz

Routers details:
{
	"kernel": "5.15.162",
	"hostname": "OpenWrt-J9622-testap",
	"system": "e500v2",
	"model": "Hewlett-Packard MSM460",
	"board_name": "hpe,msm460",
	"rootfs_type": "squashfs",
	"release": {
		"distribution": "OpenWrt",
		"version": "23.05.4",
		"revision": "r24012-d8dd03c46f",
		"target": "mpc85xx/p1020",
		"description": "OpenWrt 23.05.4 r24012-d8dd03c46f"
	}
}
Connecting to host 169.254.200.2, port 4242
[  5] local 169.254.200.1 port 32906 connected to 169.254.200.2 port 4242
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.00   sec  8.25 MBytes  69.1 Mbits/sec    0    246 KBytes       
[  5]   1.00-2.00   sec  8.75 MBytes  73.3 Mbits/sec    0    286 KBytes       
[  5]   2.00-3.00   sec  8.25 MBytes  69.3 Mbits/sec    0    305 KBytes       
[  5]   3.00-4.00   sec  8.62 MBytes  72.3 Mbits/sec    0    321 KBytes       
[  5]   4.00-5.00   sec  9.00 MBytes  75.3 Mbits/sec    0    378 KBytes       
[  5]   5.00-6.00   sec  8.38 MBytes  70.4 Mbits/sec    0    378 KBytes       
[  5]   6.00-7.00   sec  8.38 MBytes  70.3 Mbits/sec    0    378 KBytes       
[  5]   7.00-8.00   sec  8.50 MBytes  71.3 Mbits/sec    0    401 KBytes       
[  5]   8.00-9.00   sec  8.00 MBytes  67.1 Mbits/sec    0    401 KBytes       
[  5]   9.00-10.00  sec  8.00 MBytes  67.1 Mbits/sec    0    486 KBytes       
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec  84.1 MBytes  70.6 Mbits/sec    0             sender
[  5]   0.00-10.04  sec  83.2 MBytes  69.5 Mbits/sec                  receiver

iperf Done.
4242/tcp:             2573
root@OpenWrt-J9622-testap:~# 

HiveAP 330
P1020 @ 533MHz:

{
	"kernel": "5.15.167",
	"hostname": "OpenWrt",
	"system": "e500v2",
	"model": "Aerohive HiveAP-330",
	"board_name": "aerohive,hiveap-330",
	"rootfs_type": "squashfs",
	"release": {
		"distribution": "OpenWrt",
		"version": "23.05.5",
		"revision": "r24106-10cc5fcd00",
		"target": "mpc85xx/p1020",
		"description": "OpenWrt 23.05.5 r24106-10cc5fcd00"
	}
}
Connecting to host 169.254.200.2, port 4242
[  5] local 169.254.200.1 port 34408 connected to 169.254.200.2 port 4242
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.00   sec  8.62 MBytes  72.3 Mbits/sec    0    262 KBytes       
[  5]   1.00-2.00   sec  8.88 MBytes  74.4 Mbits/sec    0    326 KBytes       
[  5]   2.00-3.00   sec  8.75 MBytes  73.4 Mbits/sec    0    367 KBytes       
[  5]   3.00-4.00   sec  8.50 MBytes  71.3 Mbits/sec    0    432 KBytes       
[  5]   4.00-5.00   sec  8.38 MBytes  70.3 Mbits/sec    0    476 KBytes       
[  5]   5.00-6.00   sec  9.00 MBytes  75.4 Mbits/sec    0    476 KBytes       
[  5]   6.00-7.00   sec  8.62 MBytes  72.5 Mbits/sec    0    476 KBytes       
[  5]   7.00-8.00   sec  8.38 MBytes  70.2 Mbits/sec    0    476 KBytes       
[  5]   8.00-9.00   sec  8.75 MBytes  73.4 Mbits/sec    0    476 KBytes       
[  5]   9.00-10.00  sec  8.50 MBytes  71.3 Mbits/sec    0    476 KBytes       
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec  86.4 MBytes  72.4 Mbits/sec    0             sender
[  5]   0.00-10.04  sec  85.5 MBytes  71.4 Mbits/sec                  receiver

iperf Done.
4242/tcp:             5885

Aruba AP105: ar7161 @680MHz

Routers details:
{
	"kernel": "5.15.167",
	"hostname": "OpenWrt-AP105",
	"system": "Atheros AR7161 rev 2",
	"model": "Aruba AP-105",
	"board_name": "aruba,ap-105",
	"rootfs_type": "squashfs",
	"release": {
		"distribution": "OpenWrt",
		"version": "23.05.5",
		"revision": "r24106-10cc5fcd00",
		"target": "ath79/generic",
		"description": "OpenWrt 23.05.5 r24106-10cc5fcd00"
	}
}
Connecting to host 169.254.200.2, port 4242
[  5] local 169.254.200.1 port 43094 connected to 169.254.200.2 port 4242
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.00   sec  3.62 MBytes  30.4 Mbits/sec    0    154 KBytes       
[  5]   1.00-2.00   sec  3.50 MBytes  29.4 Mbits/sec    0    202 KBytes       
[  5]   2.00-3.00   sec  3.62 MBytes  30.4 Mbits/sec    0    235 KBytes       
[  5]   3.00-4.00   sec  3.12 MBytes  26.2 Mbits/sec    0    235 KBytes       
[  5]   4.00-5.00   sec  3.38 MBytes  28.3 Mbits/sec    0    235 KBytes       
[  5]   5.00-6.00   sec  3.50 MBytes  29.4 Mbits/sec    0    255 KBytes       
[  5]   6.00-7.00   sec  3.38 MBytes  28.3 Mbits/sec    0    255 KBytes       
[  5]   7.00-8.00   sec  3.50 MBytes  29.4 Mbits/sec    0    255 KBytes       
[  5]   8.00-9.00   sec  3.38 MBytes  28.3 Mbits/sec    0    347 KBytes       
[  5]   9.00-10.00  sec  3.88 MBytes  32.5 Mbits/sec    0    347 KBytes       
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec  34.9 MBytes  29.3 Mbits/sec    0             sender
[  5]   0.00-10.03  sec  34.0 MBytes  28.4 Mbits/sec                  receiver

iperf Done.
4242/tcp:             3674

JG921A, RTL8380. mips4kc?

Routers details:
{
	"kernel": "5.15.162",
	"hostname": "OpenWrt-JG921A",
	"system": "RTL8380",
	"model": "HPE 1920-8G-PoE+ 65W (JG921A)",
	"board_name": "hpe,1920-8g-poe-65w",
	"rootfs_type": "squashfs",
	"release": {
		"distribution": "OpenWrt",
		"version": "23.05.4",
		"revision": "r24012-d8dd03c46f",
		"target": "realtek/rtl838x",
		"description": "OpenWrt 23.05.4 r24012-d8dd03c46f"
	}
}
Connecting to host 169.254.200.2, port 4242
[  5] local 169.254.200.1 port 59738 connected to 169.254.200.2 port 4242
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.00   sec   896 KBytes  7.33 Mbits/sec    0   44.1 KBytes       
[  5]   1.00-2.01   sec  1.00 MBytes  8.35 Mbits/sec    0   49.4 KBytes       
[  5]   2.01-3.00   sec   896 KBytes  7.37 Mbits/sec    0   53.4 KBytes       
[  5]   3.00-4.00   sec  1.00 MBytes  8.39 Mbits/sec    0   60.1 KBytes       
[  5]   4.00-5.00   sec   896 KBytes  7.34 Mbits/sec    0   60.1 KBytes       
[  5]   5.00-6.00   sec  1.00 MBytes  8.39 Mbits/sec    0   60.1 KBytes       
[  5]   6.00-7.00   sec  1.00 MBytes  8.39 Mbits/sec    0   93.5 KBytes       
[  5]   7.00-8.00   sec  1.00 MBytes  8.39 Mbits/sec    0   93.5 KBytes       
[  5]   8.00-9.00   sec   896 KBytes  7.34 Mbits/sec    0   93.5 KBytes       
[  5]   9.00-10.00  sec  1.00 MBytes  8.38 Mbits/sec    0   93.5 KBytes       
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec  9.50 MBytes  7.97 Mbits/sec    0             sender
[  5]   0.00-10.05  sec  9.38 MBytes  7.82 Mbits/sec                  receiver

iperf Done.
4242/tcp:            17989

[fakemanhk]

The last one is a switch??

[evs]

https://openwrt.org/supported_devices/openwrt_on_switches_faq

Yeah. CPU is weak per this article =)

[fakemanhk]

I saw that someone managed to swap the WiFi module in HiveAP 330 then you'll be able to make it an AP with 5GHz capability (maybe time to source a MT7915E)

[evs]

@fakemanhk

make it an AP with 5GHz capability

Hive-AP 330 does have 5ghz. It's a dual radio 2.4 / 5 with two 3x3 cards?

Hahaha. I would like to test that at some point. When I first got them I bought the cheap brcmfmac 4x4 wifi ac cards you could get on aliexpress with that intent. I run 40mhz channels on my AC gear anyway so it's not much of an upgrade.

My main issue at the moment before I explore using my HiveAP 330's over my AP105's in production is that you can't run the PoE and the 12v without browning out the supply causing a reboot if one of them goes. Probably would require hardware hacks. i.e. the idea was cross stack link aggregation with a poe to 12v adapter. (These were by far the cheapest dual ethernet PoE AP's i could find when I was exploring)

I ended up just doubling up the AP's and going with the Aruba AP 105 per power domain as they were even cheaper.

Recently I saw there's the fortiap 421. Issue is non local supply, but that looks to have dual PoE supplies.

Regarding Wifi 6, I have heaps of 802.11n and '1ac clients still. My design is 4x4 wifi 5 ap's in every room at home. Lounge room has the 4x4 wifi 6 AP and a 2x2 wifi 6 ap just for the TV which sadly is wifi only and I can't get a USB adapter to work with it.

I've cabled everything else. Plus there's always the fact I'm on VDSL2 and fibre upgrade involves redoing my entire concrete driveway or digging up the garden. So getting better than 70mbit at home is a way away.

[fenrirkun]

so RTL8380 get around 9Mbps for wireguard, and around 30Mbps for routing/nat, not bad, for backup tunnel with ssh/webadmin access is more than enough.

really curious about RTL930x switches wireguard performance

[evs]

Yeah. Not good enough to send a ludicrous amount of logs through though =P

[evs]

@fenrirkun

really curious about RTL930x switches wireguard performance

MISP34Kc @ 800MHz?
Probably could approximate its performance on an RTL839x too as they have 700MHz MIPS34Kc. Perhaps ask someone in the realtek support thread?

[antonlindb]

Significant performance decrease jumping to the newest SNAPSHOT and 6.6 kernel for the ASUS TUF 6000 compared to latest stable release:

Latest stable release: 841 Mbits
Latest SNAPSHOT: 732 Mbits
—23% decrease

(Used exact same settings/packages)

Anyone with an hypothesis why this is happening for some devices? With my x86 N100 device I got a significant performance increase with new kernel.

Router details:
{
        "kernel": "6.6.56",
        "hostname": "OpenWrt",
        "system": "ARMv8 Processor rev 4",
        "model": "ASUS TUF-AX6000",
        "board_name": "asus,tuf-ax6000",
        "rootfs_type": "squashfs",
        "release": {
                "distribution": "OpenWrt",
                "version": "SNAPSHOT",
                "revision": "r27751-f8c22c9bff",
                "target": "mediatek/filogic",
                "description": "OpenWrt SNAPSHOT r27751-f8c22c9bff"
        }
}
Connecting to host 169.254.200.2, port 4242
[  5] local 169.254.200.1 port 55988 connected to 169.254.200.2 port 4242
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.00   sec  89.4 MBytes   749 Mbits/sec    0    505 KBytes       
[  5]   1.00-2.00   sec  88.2 MBytes   740 Mbits/sec    0    558 KBytes       
[  5]   2.00-3.00   sec  87.0 MBytes   730 Mbits/sec    0    558 KBytes       
[  5]   3.00-4.00   sec  86.1 MBytes   722 Mbits/sec    0    558 KBytes       
[  5]   4.00-5.00   sec  87.8 MBytes   736 Mbits/sec    0    558 KBytes       
[  5]   5.00-6.00   sec  87.0 MBytes   730 Mbits/sec    0    558 KBytes       
[  5]   6.00-7.00   sec  86.1 MBytes   722 Mbits/sec    0    558 KBytes       
[  5]   7.00-8.00   sec  87.8 MBytes   736 Mbits/sec    0    558 KBytes       
[  5]   8.00-9.00   sec  86.4 MBytes   724 Mbits/sec    0    558 KBytes       
[  5]   9.00-10.00  sec  86.9 MBytes   729 Mbits/sec    0    558 KBytes       
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec   873 MBytes   732 Mbits/sec    0             sender
[  5]   0.00-10.00  sec   871 MBytes   731 Mbits/sec           

[fakemanhk]

Maybe you can report to the relevant device thread (or GitHub?) and see if some others observing the same issue.

[remittor]

@antonlindb

Latest stable release: 841 Mbits
Latest SNAPSHOT: 732 Mbits

Please read:

A Wireguard comparison DB

I can no longer reproduce this result. I install the same image (my build) from March 7, do the same steps and get only 800 Mbps. I'm very surprised. UPD.: On March 12, in the morning, I turned on the router and ran test on snapshot 6.1.79 again: got the same 800 Mbps. I decided to install the release version (5.15.137) and ran a test on it: got 838 Mbps. I installed the snapshot 6.1.79 again (with save configs) and ran the same test: got 972 Mbps.

972 Mbps (mt7986a)

root@OpenWrt:~/wg-benc…

A Wireguard comparison DB

I figured out the reason for this behavior. My builds contain package luci-app-xray, which incorrectly installs nft rules. Which causes the nft list ruleset command to return an empty string. So, when command nft list ruleset returns emptiness, the test gets a significant speedup. Clearing all nft rules can be achieved with the following command: nft flush ruleset After completing this command, the WG test boost occurs on any build.

A Wireguard comparison DB

ASUS TUF-AX6000

ubus call system board

| OpenWrt 23.05-SNAPSHOT, r23780-6f70e09a00 | | | | Build time: 2024-03-09 13:45 CET | | Cezary Jackiewicz, https://eko.one.pl | ----------------------------------------------------------------------------- | Machine: ASUS TUF-AX6000 …

A Wireguard comparison DB

after nft flush ruleset

./benchmark.sh

root@AsusWRT:/wg-bench# nft flush ruleset root@AsusWRT:/wg-bench# ./benchmark.sh Connecting to host 169.254.200.2, port 5201 [ 5] local 169.254.200.1 port 40826 connected to 169.254.200.2 port 5201 [ ID] Interval Transfer Bitrate Retr Cwnd [ 5] 0.00-1.00 sec 111 MBytes 927 Mbits/sec 0 708 KBytes [ 5] 1.00-2.00 sec 108 MBytes 907 Mbits/sec 0 819 KBytes [ 5] 2.00-3.00 sec 109 MBytes 913 Mbi…

[systemcrash]

Useful to know. Although the result shows an overall performance regression, and is not irrelevant: it indicates that there is a regression somewhere, or some new CPU consumption vector was introduced. The test is overall still (quite) objective.

Edit: is there some 'link speed' to the CPU? Given that some devices I've tested are DSA, some ports are the CPU. And some speeds were suspiciously close to 1Gbit limit.

[thedude]

@systemcrash

Although the result shows an overall performance regression, and is not irrelevant: it indicates that there is a regression somewhere, or some new CPU consumption vector was introduced. The test is overall still (quite) objective.

I haven't checked if things have got worse, but I noticed a regression months ago. And I didn't report it since nobody else shared comparisons and this is a synthetic benchmark.

[evs]

@fenrirkun

Here's the data for rtl8393M which is mips34kc @ 650MHz, ddr3 @ 200MHz and the closest I can get you to an rtl930x platform which also has MIPS-34kC?
(that would be 800MHz. IDK about comparing DDR3 speed or width? I'd need to check datasheet?)

Router details:
{
	"kernel": "6.6.57",
	"hostname": "OpenWrt",
	"system": "RTL8393",
	"model": "HPE 1920-48G (JG927A)",
	"board_name": "hpe,1920-48g",
	"rootfs_type": "initramfs",
	"release": {
		"distribution": "OpenWrt",
		"version": "SNAPSHOT",
		"description": "OpenWrt SNAPSHOT",
		"revision": "r27893-b6bbc76c0b",
		"target": "realtek/rtl839x",
		"builddate": "1729635115"
	}
}
Connecting to host 169.254.200.2, port 4242
[  5] local 169.254.200.1 port 57656 connected to 169.254.200.2 port 4242
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.00   sec  2.38 MBytes  19.8 Mbits/sec    0   77.5 KBytes       
[  5]   1.00-2.00   sec  2.50 MBytes  21.0 Mbits/sec    0   94.9 KBytes       
[  5]   2.00-3.00   sec  2.38 MBytes  20.0 Mbits/sec    0    100 KBytes       
[  5]   3.00-4.01   sec  2.50 MBytes  20.7 Mbits/sec    0    106 KBytes       
[  5]   4.01-5.00   sec  2.62 MBytes  22.3 Mbits/sec    0    106 KBytes       
[  5]   5.00-6.00   sec  2.38 MBytes  19.9 Mbits/sec    0    106 KBytes       
[  5]   6.00-7.00   sec  2.75 MBytes  23.1 Mbits/sec    0    160 KBytes       
[  5]   7.00-8.01   sec  2.50 MBytes  20.9 Mbits/sec    0    160 KBytes       
[  5]   8.01-9.00   sec  2.50 MBytes  21.1 Mbits/sec    0    160 KBytes       
[  5]   9.00-10.01  sec  2.50 MBytes  20.9 Mbits/sec    0    160 KBytes       
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.01  sec  25.0 MBytes  21.0 Mbits/sec    0             sender
[  5]   0.00-10.02  sec  24.8 MBytes  20.7 Mbits/sec                  receiver

iperf Done.
4242/tcp:             7303

[fakemanhk]

Is this a switch???

[frollic]

@evs

HPE 1920-48G (JG927A)

Yes, it is

[evs]

Fortinet fortiwifi fwf-50e-2r.
Cortex A9 @ 1.6GHz ~400Mbps

Router details:
{
        "kernel": "6.6.58",
        "hostname": "OpenWrt",
        "system": "ARMv7 Processor rev 1 (v7l)",
        "model": "Fortinet FortiWiFi 50E-2R",
        "board_name": "fortinet,fwf-50e-2r",
        "rootfs_type": "squashfs",
        "release": {
                "distribution": "OpenWrt",
                "version": "SNAPSHOT",
                "description": "OpenWrt SNAPSHOT",
                "revision": "r27917-3cdab2ad44",
                "target": "mvebu/cortexa9",
                "builddate": "1730035524"
        }
}
Connecting to host 169.254.200.2, port 4242
[  5] local 169.254.200.1 port 42908 connected to 169.254.200.2 port 4242
[ ID] Interval           Transfer     Bitrate         Retr  Cwnd
[  5]   0.00-1.00   sec  47.9 MBytes   401 Mbits/sec    0   1.03 MBytes       
[  5]   1.00-2.00   sec  46.2 MBytes   388 Mbits/sec  180    590 KBytes       
[  5]   2.00-3.00   sec  42.8 MBytes   357 Mbits/sec    0   1.24 MBytes       
[  5]   3.00-4.00   sec  49.4 MBytes   414 Mbits/sec    0   1.35 MBytes       
[  5]   4.00-5.00   sec  49.2 MBytes   414 Mbits/sec    0   1.42 MBytes       
[  5]   5.00-6.00   sec  48.0 MBytes   403 Mbits/sec  103   1.07 MBytes       
[  5]   6.00-7.00   sec  49.1 MBytes   412 Mbits/sec    0   1.13 MBytes       
[  5]   7.00-8.00   sec  49.1 MBytes   412 Mbits/sec    0   1.18 MBytes       
[  5]   8.00-9.00   sec  48.5 MBytes   407 Mbits/sec    0   1.20 MBytes       
[  5]   9.00-10.00  sec  48.5 MBytes   406 Mbits/sec    0   1.22 MBytes       
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Retr
[  5]   0.00-10.00  sec   479 MBytes   401 Mbits/sec  283             sender
[  5]   0.00-10.02  sec   477 MBytes   399 Mbits/sec                  receiver

iperf Done.
4242/tcp:             4471
root@OpenWrt:~#

[fakemanhk]

Comparing with Linksys WRT1900ACv2 which has the same SoC and same clock speed, with snapshot 6.6 LTS kernel it's also showing ~10% increase WG speed than 23.05.x.

[evs]

I'm slowly working on backporting 23.05.5 to it but no 23.05.5 build to confirm your conjecture yet =)

[fakemanhk]

If reverting to 23.05.x can show the same magnitude of speed reduction then there is good amount of cases here showing that the 6.6 LTS has improved WG performance!