USB Wifi 7 adapter - USB-BE92 Nano

[RMerlin]

I see the product is now available for sale, so I guess I can talk about it.

Tri-band WiFi 7 USB adapter with speeds of up to 6500 Mbps

Dual-band WiFi 7 router with speeds of up to 6800 Mbps, support 4096-QAM, Multi-link operation (MLO), 10G port, subscription-free network security and comprehen

www.asus.com

New USB dongle that adds Wifi 7 support. It's based on an RTL8922AU, which unfortunately is limited to 160 MHz. MLO support is expected to be added at a later time.

The current driver still seems a bit iffy, Windows reports its link PHY as AX, while the router does specify the PHY is really BE.

 

[John Fitzgerald]

I see the product is now available for sale, so I guess I can talk about it.

Tri-band WiFi 7 USB adapter with speeds of up to 6500 Mbps

Dual-band WiFi 7 router with speeds of up to 6800 Mbps, support 4096-QAM, Multi-link operation (MLO), 10G port, subscription-free network security and comprehen

www.asus.com

New USB dongle that adds Wifi 7 support. It's based on an RTL8922AU, which unfortunately is limited to 160 MHz. MLO support is expected to be added at a later time.

The current driver still seems a bit iffy, Windows reports its link PHY as AX, while the router does specify the PHY is really BE.

You say the Driver is iffy, are you using it currently? Have one in hand?

 

[wrink]

Would have been great to have Mac driver/compatibility - since 2024 Macs are lacking wifi 7 for the most part…

 

[snakebite3]

Does the adapter support wifi 7 on windows 10? I can't find a straight on wifi 7 support on windows 10. It say compatible windows 10 is the wifi6 only?

 

[sfx2000]

The current driver still seems a bit iffy, Windows reports its link PHY as AX, while the router does specify the PHY is really BE.

Windows - grr...

That being said - if the stars align with the Windows Hardware Quality Lab test suite...

 

[RMerlin]

You say the Driver is iffy, are you using it currently? Have one in hand?

I have one, but I don't actively use it, I got it for whenever I need to test Wifi 7 specific things, such as the Wireless Log support for Wifi 7.

Would have been great to have Mac driver/compatibility - since 2024 Macs are lacking wifi 7 for the most part…

I know Linux supports that Realtek chip. Whether it will ever be supported by Apple is up to Apple. I don't even know if they support third party Wifi adapters at all, or if MacOS supports Wifi 7.

Does the adapter support wifi 7 on windows 10? I can't find a straight on wifi 7 support on windows 10. It say compatible windows 10 is the wifi6 only?

Windows 10 itself does not support Wifi 7. You need Windows 11 (ideally 24H2 for full support).

Windows 10 is going EOL in 11 months anyway.

 

[RMerlin]

That being said - if the stars align with the Windows Hardware Quality Lab test suite...

WHQL validation doesn't mean anything these days. How many broken drivers have been shipped by Intel or Realtek these past few years that were WHQL-validated?

 

[John Fitzgerald]

WHQL validation doesn't mean anything these days. How many broken drivers have been shipped by Intel or Realtek these past few years that were WHQL-validated?

I realize ASUS has it's own Drivers and they aren't mature yet, but have you tried a generic, which is listed as Windows 10 and 11 compatible?

Realtek RTL8922AU | Device Drivers

Realtek RTL8922AU is a WiFi 7 802.11be based chipset.

oemdrivers.com

This is selling at BestBuy (US) for $70.00

 

[RMerlin]

I realize ASUS has it's own Drivers and they aren't mature yet, but have you tried a generic, which is listed as Windows 10 and 11 compatible?

Exact same version.

Realtek develops the driver, Asus merely reposts them.

 

[sfx2000]

WHQL validation doesn't mean anything these days. How many broken drivers have been shipped by Intel or Realtek these past few years that were WHQL-validated?

LOL - at least they are certified as broken

WHQL has always been a problem as the vendors self-certify...

 

[sfx2000]

Adapters for WiFi7/BE will be a challenge I think...

Esp for USB devices, or m2/mPCI devices using USB - it's a complex situation there...

Outside of Intel - the cards I've been looking at - there's an entire RTOS running on the chipsets, and there, the best solutions are both bands running concurrently...

On the router side - it'll be similar at some point...

I've been there already with 3GPP 5G and small cell chipsets - one ends up with a dual stack - linux running on top of an RTOS of one kind or another - and the RTOS is running the chipset...

 

[TheLostSwede]

Outside of Intel - the cards I've been looking at - there's an entire RTOS running on the chipsets, and there, the best solutions are both bands running concurrently...

On the router side - it'll be similar at some point...

Broadcom has been doing that for years already.

 

[sfx2000]

Broadcom has been doing that for years already.

Yes/No - the NIC's have been running on private RTOS, mostly because they have to - Broadcom generally uses ThreadX there because licensing there allows things to be private.

My point was that the Router SoC itself is getting close to private support with the underlying RTOS...

One can still run linux for userland - think of this as something above the bootloader, and something under the user view...

 

[sfx2000]

I know Linux supports that Realtek chip. Whether it will ever be supported by Apple is up to Apple. I don't even know if they support third party Wifi adapters at all, or if MacOS supports Wifi 7.

Apple is going to do what they do when it seems right for them...

Third-party NIC drivers have to go thru all the loops there with Apple for kernel extensions...

Apple is always a good benchmark for what actually works - but not a good choice for folks that want to explore the leading edge... They have to write their own drivers for any WiFi chip they support - so it's actually a big deal when they jump into the pool..

 

[wrink]

Apple is going to do what they do when it seems right for them...

Third-party NIC drivers have to go thru all the loops there with Apple for kernel extensions...

Apple is always a good benchmark for what actually works - but not a good choice for folks that want to explore the leading edge... They have to write their own drivers for any WiFi chip they support - so it's actually a big deal when they jump into the pool..

I have standardized on Apple for my personal and even work devices, because of the same/similar reasons I've standardized on Merlin for years: I need my devices/network to work reliably and predictably, more than I need the devices to show off a spec-sheet. I likely wouldn't have stuck with Asus if it weren't for Merlin (yes, I have donated over the years )

Yet, I'm also an explorer at heart and in my profession as a researcher - so I occasionally have to explore, as I'm doing now with a Wifi7 Asus router, the BT10 that I ordered today. What convinced me was the fact that I hope to be able to use the 6 GHz band that Apple has already adopted with Wifi6E. Incremental, but stable progress, I would hope.

 

[TheLostSwede]

Yes/No - the NIC's have been running on private RTOS, mostly because they have to - Broadcom generally uses ThreadX there because licensing there allows things to be private.

My point was that the Router SoC itself is getting close to private support with the underlying RTOS...

One can still run linux for userland - think of this as something above the bootloader, and something under the user view...

I think you misunderstood. The Broadcom WiFi modules in their routers have a Cortex-A7 SoC with its own OS, that is not accessible,

 

[sfx2000]

I think you misunderstood. The Broadcom WiFi modules in their routers have a Cortex-A7 SoC with its own OS, that is not accessible,

Broadcom WiFi is a complex topic - as there are at least two current architectures in play - one for AP's (WL and brcmsmac) and client silicon for things like mobiles and whatnot (SDIO and USB)...

The AC1900 class WiFi chips, according to sources, have an ARM Cortex-A9 embedded - it runs a mini-RTOS, and as you mention, not accessible from the linux userland...

The SDIO clients are interesting, as they are FullMAC, and many are combo chips - e.g. WiFi and BT...

Those devices have two arm cores for the WiFi4/5 class devices - one being are Cortex-A7 and the other being a Cortex-M3 - The A7 handling WiFi duty, and the M3 dedicated for BT as a hard RT solution...

This is likely all moot at this time, other than to compare notes from 10-15 years ago...

Looking forward - the amount of processing power needed due to WiFi7 complexity - we'll likely see Cortex-A72 class cores at a minimum - saying A72 class, as RISC-V is getting a lot of attention here..