jaszyThis is a good thing. Mediatek is usually undercutting on premium hardware cost. Happened with last gen AC wave 2 stuff.

4x4 AC Wave 2 mediatek chips were being used for 3x3 "spec" SKU's since it was more logical than buying 3x3 BCM/QCA hardware at a higher cost from suppliers. Only Downside? Slower MIPS main CPU. Couldn't handle many clients without bogging down, but range performance was certainly a level ahead of 3x3 BCM/QCA, even with a disabled/non attached antenna.

Low cost client(s) would also help IoT significantly as the market SUCKS right now. 2.4G AX is underutilized relative to its benefits. 5G doesn't really matter ATM.

Edit: If they're using A53 for all cores on higher end products, this already outclasses BCM solutions from 2019 (Radio SoCs are pure A7, even the higher end 4x4 chips). I assuming some of these solutions combine main CPU + radio CPU cores as a full A53 SoC config? Basically similar to BCM67XX (A7) but much much better.

Looking forward to white papers.

yotano211Mediatek and Realtek wifi modules suck, I'll never use them in any laptop that I use in the future. I had one in a laptop that I used for a few weeks last month. The wifi module would not pick up wifi signals from the 2nd floor hotel room.
I changed the wifi module to a intel module, and sure enough it picked up a wifi signal.

TheLostSwedeSee, that's not really fair, as notebooks often get the really crappy 1x1 modules from MTK and Realtek, which as you point out, are pretty rubbish. But so is Intel's 1x1 modules, but they're rarely seen in products.

yotano211Most of the laptops I get are higher end models so I only get intel or killer wifi cards. The wifi card on a laptop that I tried out was on the Asus g15 advantage edition. Great laptop overall, sucky wifi module.

TheLostSwedeEven some expensive laptops come with 1x1 modules though, beats me as to why. Doesn't seem to be the case with that specific laptop though.
Considering something like 50 percent of lower cost routers are MTK based, I would say they're not that bad, since those products simply wouldn't sell if it was the case.
100 percent of Wi-Fi 6E USB dongles to date use Realtek.

TheLostSwedeUndercutting, yes, but sadly they also have some limitations and issues.
From a previous job, I can tell you that they're quite slow when it comes to fixing driver bugs in their router hardware.
We informed them about an issue and it took them two or three months to respond and another couple of months to fix the issue.
It wasn't anything serious, it just stopped Wi-Fi from working if you paired one of their Wi-Fi chips with one of their SoCs...

MIPS CPUs are actually better than ARM CPUs in router, so even though the CPU is "slower" it doesn't mean it's worse. Sadly MIPS haven't kept up, since they've been shifted around and sold off a couple of times, which meant the architecture has fallen behind quite a bit and no-one seems to use their more recent cores for anything. Edit: Actually, it seems Realtek is using the MIPS interAptiv cores in some of their new 10/2.5Gbps switch SoCs for up to 12 10Gbps ports or 24 2.5Gbps, plis six 10Gbps ports.
If you compare a MIPS SoC to an ARM SoC, you'll notice that the ARM SoC uses multiple offload co-processor to handle things like hardware NAT, routing, etc. whereas MIPS SoC don't, outside of a crypto engine. This does at least suggest to me, that MIPS SoCs are more capable as router chips than ARM SoCs.

Are there even any single band 2.4GHz 802.11ax Wi-Fi chips?

Broadcom has a range of chips, some are "B53" (Cortex-A53), but both QCA and Broadcom decided to go back and do some quad and even triple core Cortex-A7 SoCs for cheaper routers for some reason, possibly because MTK had the MT7623, but it wasn't a particularly successful router SoC for MTK.
Also QCA was running on Cortex-A15/Krait for a very long time and only the IPQ807x series got a Cortex-A53 upgrade.

jaszyWell yeah, the MIPS MTK stuff is/was treated as second rate relatively which ends up more relatively buggy from what I've seen. I think a shift to faster ARM products would streamline stuff, but I guess that depends on MTK themselves.

jaszyDoubt theres a single band 2.4G AX chip, but I would prefer the current dual band stuff to at least show up in clients. A lot of brands want pricing to relevant performance.. which still favors older N/AC hardware..

jaszyA lot of the current BCM AX 4x4 routers have subpar legacy performance relative to a Legacy BCM AC 4x4 stuff in terms or range in subjective personal testing between similar products of the same "brand". Which makes the argument kinda odd.. My $550 USD AXE router is worse than my $150 AC W2 router in regards to performance in my environment due to average client.. lol

jaszyI personally view AX as "AC Wave 3" in my head (5G). Better SNR (7-9Db) and marginal QAM rate improvements. 160mhz isn't relevant on 5G (AC W2 can do it).. at least in the US Region. DFS requirements blow. 2.4G AX is another beast though.. Significant performance to range improvements, especially as a 2x2 client, but Im pretty sure you know that :)

jaszyAs for BCM, stuff like the 67XX triple and quad core A7 SoC hardware is basically co processor + main cores in tandem. I never understood why marketing positioned it this way... The higher end B53 router stuff ( B53 4908) is basically "6 core" if you look at it from the perspective of the low end stuff. IE: 4 core A53 + 2 A7 cores on radio themselves... 67XX is 2+2 or 3+1 from how I looked at it.

The IPQ8065/8054 A7 also was quite popular during the W2 transition period, but it never caught on well outside of the "higher end" stuff like R7800/RT2600. At least on the router end.

TheLostSwedeThe hardware is in general decent, but their software support is lacking and they're hard/impossible to work with unless you place big orders.
All their ARM chips to date for routers have been disappointing or not even delivered in terms of all of the features they implemented., which really was/is a shame.

Yeah, but just look at "smart" TVs, a lot of them have used crappy network implementations, both for wired and wireless. For some reason it seems like the makers don't even spend five minutes to do a speedtest, they just check they get a signal and calls it a day. It's really rather insane. The fact that you get 802.11ac Wi-Fi in a TV, but only 10/100Mbps is also telling of how little these companies care, as they rather save that $1, even though they charge you $2,999 for that brand new 65-inch OLED TV.

When it comes to range, well, that depends on a lot more than the Wi-Fi chipset, such as the choice of PA/LNA, antennas and even placement of antennas on the device. As such you can compare two different routers unless they use exactly the same placement of the antennas and the same PA/LNA parts.

Not moved to ax yet, as the hardware I want doesn't really exist yet and a lot of what's out there is insanely expensive for no practical gains for what I use Wi-Fi for.

Well, their even older stuff is single or dual core Cortex-A9, so obviously the performance is going to be worse at times on the Cortex-A7 based stuff, but not always.
You know what, I had missed the fact that they used a Cortex-A7 in their Wi-Fi parts, although I was aware they had a built in "offloading" processor, I expected it to be some kind of DSP, like what QCA does in their processors.

The IPQ8065 is Krait based, which is similar to a Cortex-A15, not an A7. Yeah, I have an R7800, best router I've ever had, especially with Voxel's firmware on it. Never had to manually reboot it once, which is kind of insane. Got a couple of "cheap" TP-Link devices that are based on older Atheros MIPS chips and they're not 100% stable, although OpenWRT seems to have improved upon that.

I just wish someone would make a router that actually got regular updates, like the equivalent of a Pixel phone, but as a router. It's as if the router makers haven't figured out that they're the first point of defence against all sorts of crap on the internet and when people are running routers with a five year old firmware, as they don't know better... :(
If I had the money, I would do it myself, but alas.

btarunrIn all likelihood, AMD may rebadge that Filogic 630 with some AMD PRO feature-set driver rivaling Intel vPro for Wi-Fi.

jaszyYeah, most add on usb networking sticks on many TV's outperform the internal chipset.. it's pretty odd. But thats the thing, every form of "IoT" is trying to save money.. I understand it, but the AX push has been lack luster in general.Well

jaszyI'm more or less targeting the industry itself as I do think AX 2.4G has clear benefits due to possible throughput at range (low spectrum). Compared to 2.4G N.. its a big step up. 5G AC vs AX? Well, shouldn't really matter at this point in time.. any 2x2 5G client should be fine enough for most consumers.

jaszyAs for same hardware AX vs AC, I have a GT-AC2900 variant of the AC86U. I've also used the AX variant (AX86U) with the same antenna setup (they're identical externally and internally I think). The older AC variant is the 1.8 ghz B53 4906 dual core which should be the same as the 4908 (quad) with 2 cores disabled, but I could be wrong. Only difference is the Radio "co processors" and speed. AC variants run at 800mhz internally. The AX radios should be 1.5ghz. AX86U is better at AX stuff and closer range, but the AC variant with BCM4366E does better at distance (30~ft).

jaszyA lot of it is firmware, but Legacy AC stuff (including QCA R7800) seems better.. at least in my environment and channel selection. AX clients would highly favor modern hardware though.. Much more throughput.

jaszyYeah BCM has a weirder setup.. but modern marketing is completely bonkers as both 3rd party brand (Asus/Netgear) and BCM like to tout the SoCs as triple and Quad core..A dual band 4906 setup with co processors is technically better being 1.8ghz B53... 2+2 setup (main+co processor) instead of 3 + 1 or 4 + 0. I'm surprised a company like ASUS doesn't take advantage of that in higher end products if a "lower end" product is automatically inflated from the start... Also makes more recent products like the 3x3 AX68U seem worse than it really is being marketed as "dual core". I know the engineers don't really get involved with marketing, but I'm triggered by that...

jaszyAh, I read krait 300 was A7 but there seems to be mixed information there. My mistake I guess.

jaszyAs for updates/support/warranty I like ASUS. The GT-AC2900 (2019) was still getting updates well into June 2021 with quite a bit of fixes and this is based off a AC86U (2017) firmware with slight changes.. and even if that isn't an option, theres always merlin, but he doesn't really support GT stuff. I guess it doesn't matter in my case.

TheLostSwedeWell, 6E seems to be pushed a lot harder than 6, even though it is still not possible to use it legally in a lot of countries.

Well, it would make sense for some devices, but I have a feeling it'll draw more power than 802.11n, which might also be a reason why it's not used in a lot of devices. Then again, it also depends on what you call IoT.

Still going to be different PA/LNAs in those, but I guess it's a reasonably fair comparison in as much as possible. Some of it has to do with drivers as well though. I presume you read the reviews over on Smallnetbuilder? He got very different results out of some fairly similar Asus hardware as well.

Well, the R7800 is one of those once in a generation type of hardware that delivers more than expected. Yes, it seems like if you disable support for older standards, you get much better range and throughput on 802.11ax devices, but I doubt anyone has only 802.11ax devices on their network.

Marketing, marketing, marketing...

Well, the Krait stuff is a bit complicated, but my understanding is that it's closest to a Cortex-A15, even though it's not the same. Not really a mistake, as Qualcomm like to keep secrets, like all of the router chip makers, which is really annoying, as their "trade secrets" aren't really all that secret.

That's only because Asus got slapped by the FCC. Look up some of their more obscure models, especially ones based on MTK hardware and you'll see updates that equal or are even worse than TP-Link, i.e. no more than two or three firmware updates for the life of the device. Obviously none of those are sold in the US market. Merlin does excellent work as well, but only supports Broadcom based routers, just like Voxel only does a few QCA based models from Netgear.

jaszySee.. I don't really see a valid 6E push. I feel it's premature and a lot of people don't understand what they're getting into. I do have a tri band 6E router, but I don't see the practical benefits outside of low 6ghz wireless backhaul (future products) or dedicated "gaming" setups on PC via intel AX210. (Uninhibited lower spectrum 6ghz/160mhz at full power output). 5G 160mhz is stupid considering DFS in US region.

But then you start asking if it's more wiser to just wire/rewire your house when you factor current router cost for tri band 4x4 hardware (Early adopter pricing obviously sucks). I for one cant recommend it to an average consumer.. Possibly even 2-3 years down the line. Clients are too important and the lack of reg 5G/2.4G AX is apparent.

Yes I've been on SNB. And my general subjective testing seems to favor the older AC 4x4 stuff regardless of multiple AX routers, but I haven't exactly messed with the higher end 12 stream QCA stuff yet to be fair.. Might be a BCM issue? I don't want to buy radios under 3x3 config as the 2x2 stuff I tested (AX3000-Intel/BCM) is mostly trash propped up by marketing numbers. But I guess thats obvious being a 2x2 router product.. :)

Well I've seen it with QCA ASUS models too (They have an "R7800" hardware model, just never got propped up). ASUS is quite invested into BCM, like their monitor division is to AUO.