Intel Preparing Wi-Fi 7 / 802.11be / EHT Support For The Linux Kernel

[phoronix]

Phoronix: Intel Preparing Wi-Fi 7 / 802.11be / EHT Support For The Linux Kernel

Intel's open-source engineers have already begun preparing Linux kernel patches around 802.11be Extremely High Throughput (EHT) support that will be known as WiFi 7...

https://www.phoronix.com/scan.php?page=news_item&px=Intel-WiFi-7-802.11be-Linux

[M@yeulC]

It's a bit weird that Wi-Fi might be able to surpass Ethernet speeds soon in consumer devices.

I hope this will help bring 40GBE (or at least 2.5-10) to the mass market, at affordable prices.

[numacross]

It's a bit weird that Wi-Fi might be able to surpass Ethernet speeds soon in consumer devices.

I hope this will help bring 40GBE (or at least 2.5-10) to the mass market, at affordable prices.

Theoretical speeds that are achievable in a lab are far from what you'll get in reality. I doubt WiFi can surpass Ethernet in speed in the real world without extravagant setups that cost way more. I don't think even WiFi 7 is full duplex yet (as in being full duplex in the same channel, and not using separate channels at the same time with multiple radios).

[NobodyXu]

Theoretical speeds that are achievable in a lab are far from what you'll get in reality. I doubt WiFi can surpass Ethernet in speed in the real world without extravagant setups that cost way more. I don't think even WiFi 7 is full duplex yet (as in being full duplex in the same channel, and not using separate channels at the same time with multiple radios).

I suppose that you can only get close to full speed if you stay near your WiFi router...

If your computer and your WiFi router are one wall away, then the Ethernet will be much faster.

Not to mention most people cannot even fully utilize WiFi 6.

[numacross]

I suppose that you can only get close to full speed if you stay near your WiFi router...

Not only that, but you also need the optimal number of antennas in the client device, which usually is more than in mainstream devices.

[NobodyXu]

Not only that, but you also need the optimal number of antennas in the client device, which usually is more than in mainstream devices.

Yeah, I guess at this point, ethernet will just be faster with lower latency and more reliable.

[kylew77]

Linux is prepping wifi 7 while all the *BSDs are still working on AC rather yet wifi AX (WiFi 6), but since most basic residential plans where I live are at wifi n speeds or wifi g speeds I guess it really doesn't matter in the end.

[DataSmasher]

The low latency is pretty exciting to me, when I ping quad 9 on wifi I get ~18ms on average, and about 20% of that time is communication from my router!

[risho]

what is usually the difference between the advertised speed vs the actual speed? If they advertise 30-40 gigabit does that mean that its actually possible that wifi 7 could be reliably 10 gigabit under reasonable conditions? also is this 30-40 gigabit for the 6ghz spectrum or could we also reasonably expect multi-gigabit from the 5.4 ghz spectrum as well?

I also wonder if this reduced latency could wind up making stuff like steam remote play more reliably under wifi

[Spacefish]

If they advertise 30-40 gigabit does that mean that its actually possible that wifi 7 could be reliably 10 gigabit under reasonable conditions?

Probably not, as most clients are only 2x2, so under optimal conditions you would have 320MHz x2 with 4096QAM which depending on guard interval will give you a theoretical maximum arround 10GBit, but 4096QAM on 2x2 is unlikely as the channel sparation wouldn´t bet that good in the realworld to pull this off.

I think stable 5GBit is more in the range of possibilities when your environment is "clean" (no other devices poluting the spectrum) and you have a good channel between your AP and STA.

These crazy 16x16 setups (16 TX and 16 RX Chains + 16 Antennas) are meant for high density APs in office spaces, stadiums or crowded foodways in a large city.. 98% of clients will be 1x1 or 2x2..
The benefit of the newer standards is to have a lot more clients associated to the same AP and send to them simultaniously by giving them part of the spectrum in the same frame OFDMA with even varying TX Power. So the new standard can utillize the availiable spectrum much better if there are a lot of clients and in this case having more antennas and RF-Chains benefits the AP.
But there won´t be 16x16 clients or 8x8 clients anytime soon (apart from some crazy expensive "gaming" hardware with externals antennas, with no real use).
All that is, if most clients support 802.11ax+ as well in the cell (which is not the reality today, but will be in the future)