๐
For the official Wi-Fi performance numbers for reviews on this website, I generally need a computer-based client. So, for the latest Wi-Fi standard, I need a Wi-Fi 7 computer. Without one, there’s no real-world testing.
Sure, some phones support the new wireless standard, and I have a few, including a Pixel 8 Pro currently on the way. But phones have limited bandwidth needs, and you can only perform tests on them via apps, which are generally inaccurate since the Internet and Wi-Fi are two different things.
For that reason, since Intel announced its Wi-Fi 7 chips, I’ve been bugging hardware vendors about when I can get a motherboard or a laptop with built-in Wi-Fi 7 or if there’s an add-on Wi-Fi 7 adapter.
Finally, that happened, and now I just got myself a real Wi-Fi 7-enabled computer. Or did I?
Building a Wi-Fi 7 computer can be a daunting task
Of all the vendors I got in contact with, Asus was the one that managed to make available the first Wi-Fi 7-enabled motherboard, the ROG Strix Z790-E II, and the company graciously offered to send me a loaner!
The board arrived on the last Friday of October. Thank you, Asus!
This is a great board. Besides built-in Wi-Fi 7, it also has USB 3.2 Gen 2×2, PCIe 5, four M.2 NVMe slots, each with its own heat sink and built-in Q-latch lockโyou won’t need to fumble with a little screw –and many other good stuff. It even has an onboard power button which is great for an open testbed.
I recommend it if you have $500 to spare.
Other parts and their costs
But to build a computer, I’ll still need a load of other parts. The board is just the beginning.
Specifically, I bought the following necessities:
- Intel Core i5-12600K CPU: $250
- 16GB Samusng DDR5 RAM: $70
- RAID Max Power Supply: $50
- 2TB Crucial T700 NVMe PCIe 5 SSD: $270 (the non-heatsink versionโthe board already has built-in heatsinks for all of its M.2 slots.)
Assembling the CPU, RAM, and SSD on the motherboard board is relatively easy, and the process took me less than 5 minutes.
So far, I’ve spent over $700 on hardware (tax included), plus $200 for a Windows 11 Pro license.
And then, I needed a chassis to mount the board and the power supply. For a test client machine, I decided to opt for a portable open case and picked this one:
- HAIHUANG DIY Test Bench ATX Computer Case: $66
It’s an excellent case, but, in hindsight, it was a bad first-time decision.
A case-of-the-Monday case
What you won’t notice from the pictures is that this computer case arrived completely unassembled.
It includes nine metal bars of different lengths and just shy of a million other little metal and plastic pieces to hold them together. Many of them need to be installed in the correct orderโelse, you’d need to dismantle the whole thing to put a particular piece in its right spot.
The package was like a mean Lego set, and the instruction, which I had to scan a QR quote to view on my phone screen, literally says in part, “Please use your brain”. That’s after I already used my brain to figure out how to display the English version (it was originally in Chinese.)
After hurting my nervous system, my hands, my eyes, and mostly my ego pretty badly, I did manage to put the computer together properlyโthe case turned out to be perfect for my needs. It only took me almost four hours.
It was a learning experience. If I ever get a second unit, which I might, it’d likely only take me about 30 minutes or even shorter.
The thing is, it was four hours I didn’t have. Those were the last days of October. We had three little kiddos excited for Halloween and lots of decorations to put up, etc. This kind of untimely project could tear a family apart!
But, hey, I did it!
We had a fun Halloween and still are together.
A Wi-Fi 7 Windows computer that makes no difference
It was quite exciting to get the new computer running. I installed Windows 11 Pro version 23H2โthe hardware fully supports it, so I didn’t need to use any tricks like on unsupported hardwareโactivated Windows and updated all the drivers to the latest.
Then came the moment of truth.
I connected the machine to the Asus RT-BE96U I was testing for an upcoming review and a few other Wi-Fi 7 routers I had, and it worked right away.
Alas! That was when I recognized a few sort-of-unexpected things.
- First, the motherboard has an Intel BE202 embedded. This chip only supports the 160MHz channel widthโper the standard, it can negotiate at around 2.9Gbps at best.
- Second, because Wi-Fi 7 is not ready, the latest Intel driver only allows it to negotiate at 2.4Gbpsโ2402Mbps, to be preciseโthe same as the Intel AX210 Wi-Fi 6E client.
- Finally, the MLO feature is not supported on the client until the next release of Windowsโnamely, Windows 11 24H2 or Windows 12, depending on how Microsoft makes up its mind.
If you’re new to Wi-Fi 7, check out this primer post on the Wi-Fi standard, or open the cabinet below for a few highlights.
Wi-Fi 7 highlights
1. The all-new 320MHz channel width
The first thing to note about Wi-Fi 7 is the new and much wider channel width, up to 320MHz, or double that of Wi-Fi 6/6E.
This new channel width is generally available on the 6GHz band, with up to three 320MHz channels. However, Wi-Fi 7 can also combine portions of the 6GHz and 5GHz bands to create this new bandwidthโmore in the Multi-Link Operation section below.
Details of Wi-Fi channels can be found here, but the new channel width generally means Wi-Fi 7 can double the base speed, from 1.2Gbps per stream (160MHz) to 2.4Gbps per stream (320MHz).
So, in theory, just from the width alone, a 4×4 broadcaster 6GHz Wi-Fi 7 can have up to 9.6 Gbps of bandwidthโor 10Gbps when rounded up. But there’s more to Wi-Fi 7’s bandwidth below.
Wi-Fi 7 also supports double the partial streams, up to 16. As a result, technically, a 16-stream (16×16) Wi-Fi 7 6GHz band can deliver up to over 40Gbps of bandwidth, especially when considering the new QAM support below.
Like Wi-Fi 6 and 6E, initially, Wi-Fi 7 will be available as dual-stream (2×2) and quad-stream (4×4) broadcasters and dual-stream clients. In the future, the standard might have 8×8 broadcasters and single-stream or quad-stream clients.
Again, you need a compatible client to use the new 320MHz channel width. Existing clients will connect using 160MHz at best. In reality, the 160MHz will likely be the realistic sweet-spot bandwidth of Wi-Fi 7, just like the 80MHz in the case of Wi-Fi 6.
2. The 4K-QAM
QAM, short for quadrature amplitude modulation, manipulates the radio wave to pack more information in the Hertz.
Wi-Fi 6 supports 1024-QAM, which itself is already impressive. However, Wi-Fi 7 will have four times that, or 4096-QAM. Greater QAM means better performance for the same channel width.
As a result, Wi-Fi 7 will be much faster and more efficient than previous standards when working with supported clients.
Wi-F 7 vs. Wi-Fi 6/6E: The realistic real-world speeds
With the support for the wider channel width and higher QAM, Wi-Fi 7 is set to be much faster than previous standards on paper.
You might have read somewhere that Wi-Fi 7 is “up to 4.8 times faster than Wi-Fi 6,” and hardware vendors will continue to combine the theoretical bandwidth of a broadcaster’s all bands into a single colossal numberโsuch as BE19000, BE22000, or BE33000โwhich is excellent for advertising.
Like always, these numbers don’t mean much, and things are not that simple. In reality, a Wi-Fi connection generally happens on a single band at a timeโthat’s always true for Wi-Fi 6E and older clientsโand is also limited by the client’s specs.
The table below summarizes what you can expect from Wi-Fi 7’s real-world organic performance compared to Wi-Fi 6E when working on the 6GHz.
| Wi-Fi 6E | Wi-Fi 7 | |
| Max Channel Bandwidth (theoretical/top-tier equipment) | 160MHz | 320MHz |
| Channel Bandwidth (widely implemented) | 80MHz | 160MHz |
| Number of Available Channels | 7x 160MHz, or 14x 80MHz channels | 3x 320MHz, or 7x 160MHz channels, or 14x 80MHz channels |
| Highest Modulation | 1024-QAM | 4096-QAM |
| Max Number of Spatial Streams (theoretical on paper / commercially implemented) | 8 / 4 | 16 / 8 (estimate) |
| Max Bandwidth Per Stream (theoretical) | 1.2Gbps (at 160MHz) 600Mbps (at 80MHz) | โ 2.9Gbps (at 320MHz) โ 1.45Gbps (at 160MHz) |
| Max Band Bandwidth (theoretical on paper) | 9.6Gbps (8×8) | 46.1Gbps (16×16) |
| Commercial Max Band Bandwidth Per Band (commercially implemented) | 4.8Gbps (4×4) | 23Gbps (8×8), or 11.5Gbps (4×4) |
| Available Max Real-word Negotiated Speeds(*) | 2.4Gbps (via a 2×2 160MHz client) 1.2Gbps (via a 2×2 80MHz client) | โ 11.5Gbps (via a 4×4 320MHz client) โ 5.8Gbps (via a 2×2 320MHz client or a 4×4 160MHz client) โ 2.9Gbps (via a single stream 320MHz client or a 2×2 160MHz client) โ 1.45Gbps (via a single stream 160MHz client or a 2×2 80MHz client) |
| Available Clients (example) | 2×2 (Intel AX210) | 2×2 (Intel BE200 / Qualcomm NCM865) |
(*) The actual negotiated speed depends on the client, Wi-Fi 7 specs, and environment. Real-world sustained rates are generally much lower than negotiated speedsโcapping at about two-thirds at best. Wi-Fi 6/6E has had only 2×2 clients. Wi-Fi 7 will also use 2×2 clients primarily, but it might have 4×4 and even single-stream (1×1) clients.
Like Wi-Fi 6 and 6E, Wi-Fi 7 has been available only in 2×2 specs on the client side. That, plus the sweet-spot 160MHz channel width, means, generally, it’s safe to conservatively expect real-world rates of the mainstream Wi-Fi 7 (160MHz) to be about 20% faster than top-tier Wi-Fi 6E (160MHz) counterparts.
However, the new standard does have more bandwidth on the broadcasting side. So, it can handle more 2×2 clients simultaneously with high-speed real-world rates. And that’s always a good thing.
3. Multi-Link Operation
Multi-Link Operation, or MLO, is the most exciting and promising feature of Wi-Fi 7 that changes the norm of Wi-Fi: Up to Wi-Fi 6E, a Wi-Fi connection between two direct devices occurs in a single band at a time. MLO changes that.
It’s worth noting that MLO is a feature and not the base of the standard, meaning it can be supported by a particular device or not.
In a nutshell, MLO is Wi-Fi band aggregation. Like Link Aggregation (or bonding) in wired networking, it allows combining two or more Wi-Fi bands into a single Wi-Fi link (single SSID). That said, you can have MLO as long as the broadcaster has more than one band, which is the case with all Wi-Fi 7 hardware.
Still, generally, there are two MLO operation modes:
- STR-MLMR MLO (Simultaneous Transmit and Receive Multi-Link Multi-Radio): It’s multi-link aggregation using all available bands (2.4GHz, 5GHz, and 6GHz) to deliver higher throughput, lower latency, and better reliability. (For dual-band hardware, such as the Asus RT-BE88U, this mode combines the 2.4GHz and 5GHz band.)
- E-MLSR MLO (Enhanced Multi-Link Single Radio): It’s multi-link using dynamic band switching between 5GHz and 6GHzโthis mode is only available to broadcasters with these two bandsโto deliver load balancing and lower latency.
No matter which mode is used, the gist is that the bonded link delivers “better” connection quality and “more” bandwidth.
It’s important to note, though, that at the end of the day, MLO increases the bandwidth, allowing different applications on a client to use the two bands simultaneously. The point here is that no application on the client can have a connection speed faster than the fastest band involved. A speedtest application, for example, still uses one of the bands at a time. This connection speed is still limited by the hardware specs on both ends of the link, whichever is lower.
So, the MLO feature affords supported clients the best probability of connecting successfully at the highest possible speed using the fastest band at any given time, which changes depending on the distance between the client and the broadcaster.
Considering the vast amount of pre-Wi-Fi 7 clients on the market, keep the following in mind about MLO in consumer-grade hardware:
- By nature, link bonding will be more complicated than single-band connectivityโthere are just too many variables.
- MLO only works with supported Wi-Fi 7 clients. Some Wi-Fi 7 clients might not support it. (A Windows computer must run Windows 11 24H2 or later to support MLO.) Considering the different performance grades and hardware variants, the result of MLO will vary case by case.
- Wi-Fi 6 and 6E and older clients will still use a single band at a time when connecting to a MLO SSID. And they might pick whichever of those is available in the bonded link. You might get frustrated when they use the slow band instead of a faster one, like in the case of Smart Connect. That happens.
- An MLO SSID requires the WPA2/WPA3 or WPA3 encryption method and won’t allow legacy clients to connect. This can be a big headache for those assuming the SSID will just work with all clients. In other words, turning MLO on can cause a big compatibility issue with the hardware’s primary SSID(s).
- The reach of the bonded wireless link is as far as the range of the shorter band.
MLO in real-world usage: Great for wireless mesh, but client compatibility can be a big issue
In so-far real-world experience, MLO has proven to be a game-changer in a wireless mesh network by fortifying the Wi-Fi link between broadcastersโthe backhaulโboth in terms of speed and reliability.
Wi-Fi 7 mesh systems, via my testing method, have shown sustained wireless backhauling links over 5Gbps at 40 feet away.
In terms of range, the bonded link has the reach of the shortest band involved. Since the 6GHz band has just about 75% of the range of the 5GHz when the same broadcasting power is applied, MLO can only be truly meaningful with the help of Wi-Fi 7’s fifth and optional feature, Automated Frequency Coordination, mentioned below.
On the other hand, for devices (clients), the effect of MLO has proven to trade the (lack of) backward compatibility for a relatively subdued impact on performance.
Specifically, with single broadcasters or mesh systems with wired backhauling, the feature plays an insignificant role and generally does not noticeably improve the real-world rates of individual clientsโcurrently available at 2×2 specs, such as the Intel BE200 or Qualcomm NCM865 chipsโdespite the higher negotiated speed of the bonded link.
In a way, MOL is the alternative to the finicky “Smart Connect“, where a single SSID is used for all of the broadcaster’s bands. In fact, you can think of MLO as the enhanced version of Smart Connect with higher bandwidth and security requirements.
Some hardware vendors, such as Linksys or Asus, require Smart Connect for their broadcaster’s primary SSID before MLO can be turned on. In this case, you have to choose between the following in terms of SSIDs:
- Having a primary SSID (via Smart Connect), which is MLO-compliant in terms of security, and an optional 2nd virtual MLO-enabled SSID. This SSID likely won’t work with many legacy clients, and users will need to use the hardware’s virtual or Guest SSIDs (if available)โAsus has plenty of them via its SDN featureโwith lower security requirements to support legacy clients. Or
- Turning off Smart Connect to manage the band individually and without MLO.
Other vendors, such as TP-Link, always use MLO as a secondary virtual SSID, which is the way they handle Guest or IoT SSIDs. This approach saves the users from the MLO trapโthey won’t be prompted to turn it on by defaultโbut the compatibility with the MLO-enabled SSD remains.
That said, MLO is best used when you have mostly Wi-Fi 6E and newer clients, which won’t be the case until years from now. In the meantime, this feature should be turned off when you use a single Wi-Fi 7 broadcaster or a mesh with wired backhauling unless you have the option to create additional SSIDs with lower security requirements for existing clients.
4. Flexible Channel Utilization (FCU) and Multi-RU
Flexible Channel Utilization (FCU) (a.k.a. Preamble Puncturing) and Multi-RU are two other items that help increase Wi-Fi 7’s efficiency. With FCU, Wi-Fi 7 handles interference more gracefully by slicing off the portion of a channel with interference, 20MHz at a time, and keeping the clean part usable.
In contrast, in Wi-Fi 6/6E, when there’s interference, an entire channel can be taken out of commission. FCU is the behind-the-scenes technology that increases Wi-Fi’s efficiency, similar to the case of MU-MIMO and OFDMA.
Similarly, with Wi-Fi 6/6E, each device can only send or receive frames on an assigned resource unit (RU), which signi๏ฌcantly limits the ๏ฌexibility of the spectrum resource scheduling. Wi-Fi 7 allows multiple RUs to be given to a single device and can combine RUs for increased transmission efficiency.
5. Automated Frequency Coordination
Automated Frequency Coordination (AFC) is an optional feature and deals with the 6GHz band, so it’s not Wi-Fi 7-exclusiveโthe band was first used with Wi-Fi 6E. It’s not required for a Wi-Fi 7 broadcaster’s general function. In fact, it wasn’t even mentioned in the initial certification by the Wi-Fi Alliance.
Due to local regulations, the 6GHz band’s availability, hence, the implementation of the AFC feature, differs around the world. For this reason, some Wi-Fi 7 broadcasters, such as the Asus RT-BE88U or the TP-Link Archer BE230, forgo this band to remain dual-band.
Still, Wi-Fi 7 makes AFC more relevant than ever. That’s because the 6GHz band has the highest bandwidth (fastest) yet the shortest range compared to the 5GHz and 2.4GHz bands when using the maximum allowed broadcasting power. Originally, AFC was intended only for outdoor applications, but when implemented, it’s significant for all applications.
Here’s how AFC would work when/if available:
The feature enables a 6GHz broadcaster to check with a registered database in real-time to confirm that its operation will not negatively impact other existing registered members. Once that’s established, the broadcaster creates a dynamically exclusive environment in which its 6GHz band can operate without the constraint of regulations.
Specifically, the support for AFC means each Wi-Fi 7 broadcaster can use more broadcasting power and better flexible antenna designs. How much more? That depends.
However, it’s estimated that AFC can increase the broadcasting power to 36 dBm (from the current 30 dBm limit) or 4 watts (from 1 wat). The goal of AFC is to make the range of the 6GHz band comparable to that of the 5GHz bandโabout 25% more.
When that happens, the MLO feature above will be truly powerful. But even then, Wi-Fi 7’s range will remain the same as that of Wi-Fi 6, which is available only on the 5GHz band. Its improvement is that its 6GHz band now has a more extended reach than in Wi-Fi 6E. In other words, AFC allows the 6GHz band to have at least the same range as the 5GHz. And that’s significant.
This feature requires certification, and its availability is expected to vary from one region to another. Hardware released before that is said to be capable of handling AFC, which, when applicable, can be turned on via firmware updates.
A crude AFC analogy
Automated Frequency Coordination (AFC) is like checking with the local authorities for permission to close off sections of city streets for a drag race block party.
When approved, the usual traffic and parking laws no longer apply to the area, and the organizers can determine how fast traffic can flow, etc.
So after spending about $1000 on partsโit would have been over $1500 if I had to buy the motherboardโand hours of hard labor, I ended up with a real Wi-Fi 7 computer that’s about as good as any Wi-Fi 6E machine, through no fault of anyone.
But the effort wasn’t completely futile. At least we all now know two things:
- Many Wi-Fi 7-enabled motherboards will use the Intel BE202(*), which has only half the channel width (hence bandwidth) as the Intel BE200. And
- No matter which chip you get, from the client’s perspective, you won’t get a real Wi-Fi 7 experience until the standard is certified, which is sometime next year.
(*) When Wi-Fi 7 is finalized, the Intel BE202 is still a solid upgrade considering it supports other features of Wi-Fi 7, including MLO and higher QAM.
In a way, I, like usual, took one for the team. Speaking of which, I’m in the process of getting a different motherboard that features the Intel BE200, which is expected to be even more expensive than the ROG Strix Z790-E II. I need one to complete the testing. Until then, all these Wi-Fi 7 routers won’t review themselves.
Who knows, Asus might come to my rescue again.
Update: I later replaced the motherboard with an ASUS ROG Maximus Z790 Formula and finally had a desktop computer with a built-in Intel BE200 adapter. And Wi-Fi-wise, the new machine performed the same as when you use an add-on adapter card.
The takeaway
As lamented in every previous Wi-Fi 7 review, the standard is not ready. Despite the many broadcasters, upcoming and already released, and now the availability of clients, everything is still in draft.
At this stage, the new standard only works to a certain extentโit’s basically Wi-Fi 6E with a bit more. Its real potential will not be realized until well into 2024.
But there’s no rush, as it never is with Wi-Fi. Networking is about getting connected at the speed you need in real-time and never about which standard is being used, especially when what you’re using, namely Wi-Fi 6 or 6E, is already more than enough.
Have you looked at the R770 from RUCKUS for the AP side? I’ve seen some LI posts of them trickling out. It would change the client limitations but could provide an interesting data point.
I’ve already gotten myself a few clients, Rick. You can, too.
Hi Dong,
Thanks for doing this. I was about to pull the trigger on one of the Asus ROG Wifi 7-capable mobos to use with the RT-BE96U as well, until I read your article. I guess I’ll be mostly sticking with my current mix of Wifi 6/e clients until specifications of the new wireless standard are fleshed out further. Maybe get a Wifi 7 smartphone just to get a small taste of the new standard.
The board is great, Richard, but yes, it’ll only get cheaper as time goes by and don’t get it if Wi-Fi 7 is the sole reason. Phones work well, I’ve gotten a few, but it generally makes no difference other than seeing that little 7 icon near the Wi-Fi bars. ๐
So WiFi 7 with a huge asterisk ? I wonder what, if anything the current BE200 and BE202 chips are missing from the final version of the standard ? Also, do you know if any WiFi 7 features (MLO) are enabled yet in Linux? Also, even if the clients supported MLO, does the router firmware support it yet?
The current hardware is supposedly capable of delivering all features of Wi-Fi 7 but the details need to be agreed before they can be implemented via firmware updates. For now, Wi-Fi 7 devices all work as the common base level without any bells and whistles.
I’d give it a couple of years before you can get Wi-Fi 7 in all of its glory.
That was a great read, thanks a ton! I reckon we’ll have to wait till 2025 for Wifi7 to be ready. But hey, Wifi 6E is already pretty good, with phones and tablets jumping on board and routers/ap options becoming top-notch.
It’s crazy to think that it’s only this year that the iPhone 15 finally got on the wifi 6E bandwagon, after sticking with wifi 6 for a good 4 years. Can you imagine when Wifi 7 finally hits our phones? maybe a few top end android phones might…
All high-end Androids already have it, Fred. Apple is generally behind in tech, especially Wi-Fi. Folks within the ecosystem generally are not aware of that.
Thanks for this , always a sobering read to debunk the vendor hype.
๐