Wi-Fi 6 Performance Roundup: Five Routers Tested

[avtella]

Is that 1.7Gbps Only if 160 MHz bandwidth and 1024 QAM modulation are used?

There is two sorts of Intel Wi-Fi 6 AX200 (No vPRO and vPRO), what does mean?

1.7 Gbps is HT160 with 256QAM.

Broadcom is the only one that had 1024QAM on AC as far as I know.

vPRO enables some remote management features, not that important for a regular consumer. It’s more useful for business and corporate use.

 

[pege63]

1.7 Gbps is HT160 with 256QAM.

Broadcom is the only one that had 1024QAM on AC as far as I know.

vPRO enables some remote management features, not that important for a regular consumer. It’s more useful for business and corporate use.

Thank you very much for the answers.

 

[Razor512]

I wonder, has there been any thorough benchmarking of the Intel ax200 yet at 160MHz?
Purchased one a little while back, but can't really verify ax speeds since I lack a router with a 10GbE port or something that will allow me to test the full speed of a single ax client in more ideal conditions.

Just want to know how close it can get to delivering on that high PHY rate (or if it will disappoint with a rediciously high overhead).

 

[Adamm]

I wonder, has there been any thorough benchmarking of the Intel ax200 yet at 160MHz?
Purchased one a little while back, but can't really verify ax speeds since I lack a router with a 10GbE port or something that will allow me to test the full speed of a single ax client in more ideal conditions.

Just want to know how close it can get to delivering on that high PHY rate (or if it will disappoint with a rediciously high overhead).

A few months ago I was able to hit 1.4 - 1.5Gbps, with driver / firmware updates since then you can probably push this alittle higher.

So my AX200 wifi card for my finally laptop came today (only took three months ). Did a few quick benchmarks and was able to push throughput to my NAS to around 1.4 - 1.5Gbps for a 36% increase over my previous tests with a 160MHz AC client. These results fit in pretty accurately with the link speed increase (38%).

 

[sm00thpapa]

Great article Tim. I been gone awhile. Guess I'll be staying on AC for quite some time.

 

[sm00thpapa]

eero Pro Gen 2 still going strong here. Best wifi mesh system I have ever used.

 

[Razor512]

There is one thing I noticed, the Intel AX200 doesn't seem to support 80+80MHz in order to get 160MHz without using DFS channels. With the tech simply force everyone into using DFS?
Is there any way for the people developing these WiFi standards to get the FCC to remove all restrictions on DFS channels, with the overly low transmit power limits on WiFi, those restrictions should be rendered 100% unnecessary.

 

[sfx2000]

There is one thing I noticed, the Intel AX200 doesn't seem to support 80+80MHz in order to get 160MHz without using DFS channels. With the tech simply force everyone into using DFS?
Is there any way for the people developing these WiFi standards to get the FCC to remove all restrictions on DFS channels, with the overly low transmit power limits on WiFi, those restrictions should be rendered 100% unnecessary.

IIRC - 80+80 needs two radios, whereas 160 can be covered by a single radio.

Regarding DFS, the interference risk outweighs the benefits...

 

[Razor512]

Mainly wondering because routers like the Netgear R7800 will do 80+80, but it doesn't seem like any clients support that.

 

[CrystalLattice]

Additional info on Wifi 6, 5G, and CBRS wireless technologies:

https://www.capacitymedia.com/artic...brs-one-wireless-technology-to-rule-them-all-

As usual, Tim Higgins posts on WIFI 6 are the ONLY reliable sources of dependable information on the subject. Superb work done, just as he continues to develop a major technical testing protocol, and to walk the minefields of various companies. Thank you.

 

[CrystalLattice]

The Intel AX200 aftermarket card, as well as the AX200 integrated on motherboard, are known to be possibly defective and un-workable with Windows 10. look it up if you have customer issue.

 

[L&LD]

I have used many Intel AX200 cards with no issues. They don't have the throughput (on certain client devices) that AX201 cards do, but they are still faster than whatever it was I replaced them with. Each and every customer very happy with the increase in the throughput and responsiveness of their laptops.

On my own 2-year-old laptop, I saw the same increase in performance. Only when I compared to a CNVi AX201 based laptop did I see obvious improvements over the AX200 adaptor.

Nothing to believe here except what is in front of my eyes.

We can both be sharing links to support any 'side' we want indefinitely. Let's just say that some know how to properly switch out cards, uninstall the old drivers and then install the proper drivers and ideally, use a new SSID too to get the full performance they offer.

If any are truly defective? Return/RMA at that point.

 

[CrystalLattice]

I have used many Intel AX200 cards with no issues. They don't have the throughput (on certain client devices) that AX201 cards do, but they are still faster than whatever it was I replaced them with. Each and every customer very happy with the increase in the throughput and responsiveness of their laptops.

On my own 2-year-old laptop, I saw the same increase in performance. Only when I compared to a CNVi AX201 based laptop did I see obvious improvements over the AX200 adaptor.

Nothing to believe here except what is in front of my eyes.

We can both be sharing links to support any 'side' we want indefinitely. Let's just say that some know how to properly switch out cards, uninstall the old drivers and then install the proper drivers and ideally, use a new SSID too to get the full performance they offer.

If any are truly defective? Return/RMA at that point.

somehow i already knew you'd say that. go look on the intel company customer forums.

 

[sfx2000]

I wouldn't worry too much about 160 or 80+80 - QCA based AP's will advertise capability here on the beacon frames...

That being said - clients will generally not do that, as they also have to be either 80+80 or 160.

80+80 on the client side needs two radio chains, one for each channel - and 160MHz on client side is a real challenge.

Anyways - just saying - here's a screen shot...

This is an AP configured for 80+80...

 

[avtella]

Usually people without issues don’t comment or bother to look up issue threads, hence why “forums” can be skewed.. and will not necessarily represent reality in regards issues or failure rates. I for example have 3 laptops upgraded to the AX200 and know others with it as well whose laptops came with it, they have no issues. I’m not discounting the fact that there are those with issues but randomly claiming failure 5-15% rates by scouring a few forums is a bit extreme.

For those with issues it could also be the router having a chipset compatibility issue, laptop issue/antenna layout issue on certain laptops or actual driver issues, other causes... Windows itself has had issues with various adapters at times, most recently with Broadcom and some older Intel devices in 1903.

 

[thiggins]

The Intel AX200 aftermarket card, as well as the AX200 integrated on motherboard, are known to be possibly defective and un-workable with Windows 10. look it up if you have customer issue.

I've no functional problem using the AX200 on multiple systems and Win10. Performance could be better, but the card works.

Linux is a different story. Uplink is basically broken. See the RAX40 section of my WiFi6 review.

If you have specific facts to present, state them or link to them.

 

[Dennis Bland]

Thorough and detailed investigation as always Tim. It will be interesting to test ax performance in the presence of multiple ax clients - which would quickly degrade the performance of existing contention-based clients. My understanding is ax doesn't give a huge speed increase over ac, but the packet scheduling provides much more robust performance when the channel is shared with other clients.

But as you pointed out, there is still some work to do to make single-client ax operation more stable. One step at a time.

 

[thiggins]

I have been working for weeks on OFDMA multi-STA tests. So far, I have yet to see any consistent, significant improvement.

My tests use four STAs, which is what I'm told will provide best chance of showing significant total throughput gain.

Some routers are configured that they won't try to use OFDMA (HE-MU frames) until four AX devices are present. Others will do it with two AX STAs.

What do you mean by "more robust performance"?

 

[RMerlin]

Some routers are configured that they won't try to use OFDMA (HE-MU frames) until four AX devices are present. Others will do it with two AX STAs.

The whole thing is starting to smell so much like marketing non-sense to me... "You MAY find SOME benefit IF you have this many clients, they all use OFDMA, the moon is aligned with Mars, and the outside temperature is between 18C and 22C. Expected benefit is around 1-2%. Now go buy new Wifi 6 devices!"

It's starting to look even less worthwhile than MU-MIMO (which, again, looked great on paper, but did next to nothing in real life scenarios).

 

[coxhaus]

Just wait for the enterprise guys to figure it out. If it shows up in their systems then it will be worthwhile.