Hacker News Reader

Comment by strongpigeon | Hacker News Reader

I’m curious about the 2.5 Gbps interface, which isn’t even half the theoretical max of a single connection (~5.6 Gbps) and far from the theoretical AP Max (~36 Gbps). I guess they don’t have any PoE 10Gbps switches.

Edit: Turns out they do have one. [0]

[0] https://store.ui.com/us/en/pro/category/all-switching/produc...

View on HN

You never get close to the theoretical maximum throughput of modern WiFi radios.

Check out the review of one of the fastest 6E radios: https://www.smallnetbuilder.com/wireless/wireless-reviews/li... The highest throughput they can get is still short of 2.5G even when combining all three bands, and that's in controlled laboratory conditions. In the real world you'll get much less.

10G wired ethernet is significantly more expensive and power hungry. The extra heat and expense aren't worth it for something you'll never use in the real world.

While I generally agree that in the real world you never get close to the theoretical max bandwidth, it is worth pointing out that 7 is significantly faster than 6E, so in laboratory condition you will definitely be limited by the 2.5 GbE bottleneck. With a single MIMO stream WiFi 7 provides 2882 Mbit/s (320 MHz or 160+160 MHz channel, 4096-QAM, 5/6 coding) vs WiFi 6 which provides only 1201 Mbit/s (160 MHz channel, 1024-QAM, 5/6 coding). So a single MIMO stream can exceed the 2.5 GbE capacity, and multiple MIMO streams would exceed even further.

Exactly. Really wish the world pushes towards 5Gbps Ethernet instead.

I run a home lab and I rarely see any advantage between 2.5 and 10g. Not to say that there is none, seeing a large file transfer at 1gbps or more is fun, but the time savings vs the cost just isn't quite there for me yet.

My entire file library is only about 10 TB, and most of that is from ripped DVDs & blurays. My opportunities to transmit or receive more than a gig of data in a single transfer is extremely rare, so it makes little difference to me (and my assumption is by proxy, most users) that it could go ~4 times faster.

For actual business work, though, any lab I set up from the future on will be min-spec to 2.5gig and expected to run 10gig on at least any IT machine/vlan/subnet, whatever.

But for the home consumer, 2.5 is probably 10-50x their home internet connection and more than sufficient for normal daily use.

Another thing to think about is that theoretical max is the theoretical max of that collection of channels (thinking MIMO). So while a single AP might only have 2.5Gbit uplink, you might have multiple APs trying to talk somewhat close. The faster a radio can blast out it's packet and get quiet the faster another client can use that channel.

Faster wifi means more people get more data in the same space, even if the individual uplinks are less than the max throughout of the connection. The air is just a single big shared wire.

And then as mentioned this max speed is pretty theoretical of the raw PHY, it's not necessarily the true raw count of bytes you care about being transmitted in your probably less than perfect space.

In any normal deployment, you're probably not going to be pushing even a 1.5gbps through these during heavy use. The real benefit of 7E is reduced latency, not bandwidth.

If you need high throughput, they and other vendors sell that.

There are tons of 10gbps PoE switches on the market.

Well, given how disappointing the rest of their AP lineup performs in real world scenarios (from my anecdotal experience, ofc), it won't need more than 2.5 GbE. Probably barely exceeds 1 Gbps.

I'm pulling ~600 Mbps with their WiFi 6 offering. Feels like with enough clients on WiFi 7 it shouldn't be that hard to saturate 2.5 GbE.

I'm at 1200 Mbps with my tp-link AX55 wifi 6... $130 on amazon...

Given that that device only has 1gig ethernet, I highly doubt you are seeing 1.2gbps data transfer speeds

XG 6 PoE, 4 ports for $600

https://store.ui.com/us/en/pro/category/all-switching/produc...

Ah thanks for that. Somehow missed that one!

You should be able to get 70% of the 5.6Gbps in an ideal situation. 320Mhz Channel with 6Ghz standing in direct line of sight with no obstacle. Roughly 3.92Gbps.

So while I agree with a lot of other comment you will highly unlikely push the 2.5Gbps Ethernet in real world scenario. I would still have love to see it come with 5Gbps Ethernet. Which to be fair still hasn't gotten enough attention.

The higher wireless link speed is probably more useful for reducing the amount of radio traffic in time (a set amount of data transmitted will happen in a shorter time)

Managing (somewhat reluctantly) wifi in a multi tenant office building was frustrating because of all of the competing traffic, I’m just guessing that higher speeds means move overall traffic across the spectrum in noncoordinated environments.

The AC Pro was Wifi 5 with a theoretical max bandwidth of 1300mbps on a 1gbps connector.

I'm assuming that this speed is its ability to transfer between 2 wifi connected devices in optimal conditions and not transferring files from a NAS to a wifi receiver.

> and far from the theoretical AP Max (~36 Gbps)

How'd you come to 36 Gbps?

2.4: 688 Mbps 5: 2882 Mbps 6: 5765 Mbps

That only comes to 9.3 Gbps, and if you ever get half that in goodput you're having a hell of a good day. I wouldn't expect higher speed interfaces on such a low end (albeit modern) AP.

Got it from there: https://www.intel.com/content/www/us/en/products/docs/wirele...

2.5 GbE can operate over CAT 5e.

Reply on news.ycombinator.com