WiFi radio test

Turris Omnia
1

I encourage readers to check my posts further down this thread, as the tests performed in this initial post were not as thorough as they should have been. The results do not accurately represent what the Omnia is capable of. - thanks

Hello all,
It’s me again back with some more information. This time, a WiFi throughput competition between the Compex cards provided with the Omnia. This was created partially in response to this post: https://discourse.labs.nic.cz/t/very-slow-wifi-to-wifi-speed-on-radio1-with-40-mhz/974

This competition is testing the two wifi cards (the Compex WLE900VX - based on the Qualcomm-Atheros QCA9880 chipset, and the Compex WLE200N2 - based on the Qualcomm-Atheros AR9287 chipset) in the 2.4GHz range. Both 20MHz and 40MHz are tested. It is a wifi-client to wifi-client test only.

Real quick for the TLDR’s out there: there’s little difference between the two, with the WLE200N2 providing slightly better speeds under my test setup on both 20MHz and 40MHz.

Now for the meat!

Test Setup
Laptop:
Windows 7 Pro X64
Intel Core 2 Duo @ 2.26GHz
4GB DDR2 RAM
Patriot Torqx2 64GB SSD
Intel WiFi Link 5300 AGN - Intel Driver version 14.3.2.1

Desktop:
Windows 7 Pro X64
Intel Core i5 @ 3GHz
8GB DDR3 RAM
Patriot Torqx2 64GB SSD
Dell DW 1501 Wireless N - Broadcom driver version 5.60.48.35

Other info
I live in an apartment building so wireless frequencies are fairly crowded. I chose channel 4 to dominate. The image below was taken on the laptop while sitting next to the Omnia, and gives you an idea of how crowded the 2.4GHz frequency is.

ZEAXGUggTKE7lZJB.jpg1396×906 245 KB

No other wifi connections to the Omnia were present during testing. The girlfriend was watching Netflix via PlayStation3 on Ethernet0 the entire time.

Settings changes were followed by device reboot.

IOMeter was used for network testing.

Please see my lower post for more accurate information. I have removed most material from the original post due to poor methodology and inaccurate information.

Conclusion
The Compex WLE200N2 provides slightly better performance between two wireless clients than the WLE900VX in the 2.4GHz frequency. This is most likely due to the WLE200N2 being a strict 2.4GHz card, while the WLE900VX is both 2.4 and 5 GHz (likely with emphasis on 5GHz capabilities).

Thoughts
On our 35Mbps down/5Mbps up cable line from TimeWarner/Charter, I regularly receive 25.5/3.5 on download tests over the wireless network. Even with a degraded signal of -70 to -75dBm. I have little doubt the Omnia is a capable wireless AP/router.

1 Like
2

QED !
i’m glad i’ve ordered no-wifi version

3

Bro, thanks for the detailed tests results.

However, you do understand that even using channel 4, the edges still overlap with other edges from other wireless ap/routers who operate on nearby channels like (1-3 and 5-7) or so. Besides that, you also are aware that other devices also operate on 2.4 Ghz, likeyou mentioned PS3…the Joystick of the PS3 using bluetooth, which uses 2.4 Ghz.

That is why you see channels 1, 6 and 11 widely being used, because those channels for sure do not overlap each other.

Could you do same test if it is possible with 5 Ghz?

4

Big_boss,
No problem. I do understand the channel overlap, the reason why channels 1, 6, & 11 are widely used, and that bluetooth is 2.4GHz as well. I did thorough testing with my previous wireless router to find the channel that worked best for me in my situation. I stress that because what works best for my situation won’t be the best for others. I tried default channels and other ‘off’ channels but none performed as well as channel 4 did for me. Wifi is a complicated technology, but most people are content with just plugging in a device, setting a name and password, and letting it auto-decide what channel to use. Which is why channels 1,6, & 11 are so frequently used.

I will do more testing with regards to 5GHz, but I will probably do some more with 2.4GHz range first.

5

You have the same source and target in all of the tests, while my tests revealed a problem when testing the speed in both directions.

This is a known bug. The real speed is much higher.

6

Back again with some additional testing. I realize my methodology may be a little flawed but besides providing info I am also trying to provide other testers a base off which to build.

I probably wasn’t pushing enough data with the 4KB size, so I upped the game to 256KB.

Test Setup
Pretty much the same as above except I did some gigabit to wifi tests as notated.

Test 0 & 1 settings
Wireless client to wired client.

Wifi:
Radio 1 - Compex WLE200N2
2.4GHz - 802.11n - 40MHz
Channel 4
Bitrate: 130 Mbps
2x antenna in vertical orientation

Laptop
RX rate 144.4Mbps
TX rate 130 Mbps
Signal strength -32 dBm

Desktop
1 Gbps Broadcom Netlink in Ethernet 1

Test 0
IOMeter running on laptop, Dynamo running on desktop.
Source desktop, target laptop.

256 KB 100% Read 0% Write 0 % Random
3 min length, 10 sec ramp up

Total IOs/s 64
Read MBps 8.3 = 66.4 Mbps
Total MBps 16.7 = 133.6 Mbps
Avg response time 31 ms
Max response time 173 ms

256KB 0% Read 100% Write 0% Random
3 min length, 10 sec ramp up

Total IOS/s 75
Write MBps 9.8 = 78.4 Mbps
Total MBps 19.7 = 157.6 Mbps
Avg response time 26 ms
Max response time 158 ms

Test 1
IOMeter running on laptop, Dynamo running on desktop.
Source laptop, target desktop

256 KB 100% Read 0% Write 0 % Random
3 min length, 10 sec ramp up

Total IOs/s 82
Read MBps 10.8 = 86.4 Mbps
Total MBps 21.7 = 173.6 Mbps
Avg response time 24 ms
Max response time 178 ms

256KB 0% Read 100% Write 0% Rad,om
3 min length, 10 sec ramp up

Total IOS/s 63
Write MBps 8.2 = 65.6 Mbps
Total MBps 16.5 = 132 Mbps
Avg response time 31 ms
Max response time 183 ms

Test 2 & 3 settings
Wireless client to wireless client

Wifi:
Radio 1 - Compex WLE200N2
2.4GHz - 802.11n - 40MHz
Channel 4
Bitrate: 130 Mbps
2x antenna in vertical orientation

Laptop
RX rate 144Mbps
TX rate 144 Mbps
Signal strength -51 dBm

Desktop
RX rate 72.2 Mbps
TX rate 72.2 Mbps
Signal strength -32 dBm

Test 2
IOMeter running on laptop, Dynamo running on desktop.
Source desktop, target laptop.

256 KB 100% Read 0% Write 0 % Random
3 min length, 10 sec ramp up

Total IOs/s 21
Read MBps 2.8 = 22.4 Mbps
Total MBps 5.6 = 44.8 Mbps
Avg response time 93 ms
Max response time 636 ms

256KB 0% Read 100% Write 0% Rad,om
3 min length, 10 sec ramp up

Total IOS/s 24
Write MBps 3.1 = 24.8 Mbps
Total MBps 6.3 = 50.4 Mbps
Avg response time 82 ms
Max response time 569 ms

Test 3
IOMeter running on laptop, Dynamo running on desktop.
Source laptop, target desktop.

256 KB 100% Read 0% Write 0 % Random
3 min length, 10 sec ramp up

Total IOs/s 24
Read MBps 3.1 = 24.8 Mbps
Total MBps 6.3 = 50.4 Mbps
Avg response time 82 ms
Max response time 672 ms

256KB 0% Read 100% Write 0% Rad,om
3 min length, 10 sec ramp up

Total IOS/s 20
Write MBps 2.7 = 21.6 Mbps
Total MBps 5.4 = 43.2 Mbps
Avg response time 96 ms
Max response time 613 ms

Conclusion
Utilizing a larger data size in the tests resulted in higher speeds that more closely match the published specifications. Using the 40MHz channel on the WLE200N2 is not problematic, but performs as expected.
Wireless to wired tests showed the closest match to the specs.
Wireless to wireless tests reveal some limitations of the 802.11 spec, specifically that it is a half-duplex communication standard, and thus can only transmit so much data at a time.

Thoughts
I still believe the Omnia is performing as it should with regards to wireless. I will continue tests into the 5GHz range sometime in the future.

7

Are you sure it is a 1501? This card is listed with a 1x1:1 config and limited to 150Mbps Wifi.

As on of your cards can only do 150Mbps there must be someone cheating. Transferring 157 Mbps over 802.11n 1x1:1 config is not possible. This is about the practical limit for 300 Mbps wifi with all nifty features available.

8

Could it be that he used some extra feature. You know that companies sometime give some “burst/boost/extra…whatever name” to give it an extra boost to it.

Although if this indeed is not a typo, i am very surprised and confused how this is possible.

9

File caching or overlapping writes. Both cases will merge multiple requests to one.

10

WiFi performance is below the capabilities of the hardware.
I bought no-wifi 1GB version and separately bought 2 quite powerfull cards:
MikroTiK R11e-5HacT (QCA9880 chipset based)
MikroTik R11E-2HPnD (AR9582 chipset based)
and sadly the ath10k opensource driver sucks big time.
Previously I used 2.4Ghz NEXX WT3020 with MediaTek (Ralink) proprietary driver and it was able to deliever ~70Mbps over 144Mbps (20Mhz) channel.
Now with Turris on 5GHz with R11E-5HacT (capable of 1300Mbps and 28dBm) I get merely ~280Mbps.
The only chance for this setup is the proprietary QCA-WIFI driver and I hope someone will release images for Omnia with this driver as I doubt the Omnia developers will use this driver although the sources for qca-wifi are not hard to find and available here: http://dl.openwrtdl.com

11

well, this highly depends on your setup…
a) first of all, this is a theoretical value and you’ll never see that in the real world
b) does your client have 3 antennas (3x3 mimo) as well?
c) how far is your client away from the router? if it’s really far away, does it have a similar powerful card, too? or if it’s located next to the router, does lowering the volume help?

12

I have an Intel 7260-AC 2x2, however, when connected to ASUS AC68U (1300Mbps AC capable Broadcomm based) I get around 450-500 Mbps. All numbers from tests done in the same room, no farther than 2 meters from the router.
I use 5 antennas, 2 for 2.4GHz and 3 for 5GHz.
When it comes to output power going down from 28dBm on channel 149 (80MHz bandwith) does not help, not even switching to lower frequency (channel 36) and 23-20dBm. And all that with not even one other 5GHz router in the neighbourhood. Like I said, my bet is on the ath10k driver.

13

well, i’ve got zero experience with ath10k yet, but imho there is no real opensource driver at all, as most stuff is handled by the firmware… maybe you could also try Ben Greear’s firmware, which LEDE is using, for example
https://www.candelatech.com/ath10k.php

14

You mean the firmware from http://www.candelatech.com, yeah tried that already. Seems Omnia developers did not compile the kernel with the support for the ath10k-ct firmware as it crashes the driver and 5GHz card does not come up.

15

just looked at their git repo, and it looks like they’re also using ath10-ct, now… so maybe it’s in the next update

16

Yes, that is the card, but that specific result is from when the Desktop was wired to its gigabit NIC. Only the laptop was wireless at that time. I tried to emphasize the test settings, maybe its not so clear.

I’m not sure when I’ll get to testing 5GHz on the WLE900VX, maybe within the next week.

17

Dude, I am no authority but it sure looks to me like your decision not to use one of channels 1, 6 or 11 is a mistake. Per your InSSIDer screen shot Wesley’s WiFi Network is stomping your signal big time. You really should be using channel 1.

1 Like
18

If these are the speeds that you are getting through wifi using the 5 GHz, well that makes me very excited indeed. This also gives me sort of a confirmation that when i receive my own Omnia, and testing it, my laptop still have to be connected through a cable. However, this is a BIG boost for handheld devices that have no other way then only through wifi or so.

So thanks or this information that you have shared. Although i am sure i will have less speed(because of other hardware), still expecting a bigger boost then the max 150 Mbps getting it right now.

19

You have got it all wrong. Those speeds, as I stated, were for my previous router (ASUS AC68U). I would not expect anything more than 280Mbps from Omnia’s stock 5GHz Wi-Fi card.

20

I see, i read it a bit hasty.

Well no matter how i look at it, double it’s speed still then a fact to my current situation to the handheld devices :).

Always look at the bright side of life…:).