6 GHz Wi-Fi Gets Closer

[Grisu]

And in EU it might be neccessary to use GPS to ask for local usable free frequencies as frequencies from licensed users must no be used in your region by the router.

 

[RMerlin]

Canada tends to mostly follow the USA and the FCC, with a few rare exceptions.

 

[Deleted member 22229]

Canada tends to mostly follow the USA and the FCC, with a few rare exceptions.

That's unfortunate the FCC basically sucks.

 

[RMerlin]

That's unfortunate the FCC basically sucks.

I was only referring to radio compliance, not to the rest.

 

[Deleted member 22229]

I was only referring to radio compliance, not to the rest.

Ok. So whats the rest. FCC only really deals with radio compliance.

 

[Razor512]

The FCC is largely on a mission to make RF communications as unreliable as possible via overly low transmit power limits, limited spectrum, and a host of other arbitrary restrictions.

 

[Makaveli]

Nor will there be any firmware upgrades for the expensive Wi-Fi 6 routers Wi-Fi router makers have pushed to the market for the past two years. 6E's higher frequencies require new hardware.

Not a big deal for me in my setup. I currently only have 1 Wifi 6 capable device. And Wired connections take first priority on my network over wireless.

By the time there is enough devices on 6E it will be about time for a hardware upgrade anyways.

 

[goldbondx]

Not a big deal for me in my setup. I currently only have 1 Wifi 6 capable device. And Wired connections take first priority on my network over wireless.

By the time there is enough devices on 6E it will be about time for a hardware upgrade anyways.

I purchase products with a 5 year lifecycle. I just got rid of ORBI after 5 years. I also have an IoT enabled house where I am hoping additional components are coming out that deal with some of the pitfalls of 2.5. I am not as thrilled about seeing zero 6 enabled accessories/products because everything will suddenly be 6E.

 

[Makaveli]

I purchase products with a 5 year lifecycle. I just got rid of ORBI after 5 years. I also have an IoT enabled house where I am hoping additional components are coming out that deal with some of the pitfalls of 2.5. I am not as thrilled about seeing zero 6 enabled accessories/products because everything will suddenly be 6E.

I also tend to keep my routers for 5 years. You will be fine and I don't think you will be seeing 6E products as quickly as you think. Wait and see what happens.

 

[thiggins]

I purchase products with a 5 year lifecycle. I just got rid of ORBI after 5 years. I also have an IoT enabled house where I am hoping additional components are coming out that deal with some of the pitfalls of 2.5. I am not as thrilled about seeing zero 6 enabled accessories/products because everything will suddenly be 6E.

There will still be plenty of WiFi 6 and 6 products because 6E chipsets will start out expensive. After all, how many WiFi 6 devices have you seen besides smartphones and notebooks? Certainly nothing in IoT.

Expect high-end 5G smartphones to be first in line.

 

[Razor512]

For some reason, tons of motherboard makers are including the Intel AX200, it is annoying since they will often us that rather than the cheaper realtek 2.5GbE chipset and give users an additional Ethernet port.

To make matters worse,you often can't get the 2.5GbE unless you also get the AX200, thus you end up with a $300 motherboard that has the same VRMs as the $220 board.

 

[RMerlin]

For some reason, tons of motherboard makers are including the Intel AX200, it is annoying since they will often us that rather than the cheaper realtek 2.5GbE chipset and give users an additional Ethernet port.

To make matters worse,you often can't get the 2.5GbE unless you also get the AX200, thus you end up with a $300 motherboard that has the same VRMs as the $220 board.

2.5 Gbps Ethernet PCI-E cards are getting quite cheap, so might be a better investment to add your own 2.5 Gbps card then.

There are also a number of USB 2.5 Gbps interfaces appearing at sub-40$ prices. The Pluggable one can be had for 49$ CAD right now (probably about 39$ USD I assume).

The only thing holding me back from upgrading to 2.5G is the cost of switches. I'd need two switches at 250$ each, plus a PoE injector to keep my VoIP phone fed if I replace my current PoE switch with a non-PoE one.


What bothers me more is the fact that there are so few AX client chips out there. As pointed out, it seems like the AX200 is dominating the market. This might be one of the reasons why we still haven't seen a single USB Wifi 6 adapter. I remember when 802.11AC came out, it didn't take that long for USB adapters to come to market.

 

[Razor512]

One issue with add-in cards is how the chipsets handle the PCI express lanes.

Due to the mainstream platforms from AMD and Intel, the direct CPU lanes are needed for your GPU, as adding so much as an X1 card to your second slot, will immediately drop your GPU to 8X, which bottlenecks most cards, especially in less than ideal situations, e.g., a GPU had 4GB VRAM, but to maintain higher texture quality settings, it needs to allocate around 5-6GB. Since modern drivers are aware of different memory pools, they can allocate system memory for texture storage, and swap as needed over the PCI-E bus. For some games, if the bus is at 16X, there will be minimal impact on performance, but at 8X you can end up with hitching and other issues. Outside of gaming, some GPU compute tasks can be very demanding of the bus.

On the other hand, the chipsets will often offer an additional 16 lanes that will share 4 lanes worth of bandwidth from the chipset to the CPU. That bus is often best to use, especially on the X570 platform where it is a PCI express 4.0 X4 connection, but due to space limitations, and often the 2nd full length slot being unusable. The first X1 slot is also often unusable because most video cards will use a little over 2 slots worth of space.

Depending on the ATX board, that will leave you with 1-2 slots at most that can be used for expansion. where you may end up using 1 slot for a better sound card, and another slot for more SATA ports, since many boards are only including 6-8 SATA ports which is not enough if you want a balance of local storage amount and some redundancy. Having the Ethernet controller added to the motherboard directly, and using some of the extra chipset lanes, frees up the 2 physical slots for more common uses.

Ideally it would be good to see a board like this get rid of the AX200, and then drop the price down to like $230, and keep the 2.5GbE https://www.newegg.com/asus-rog-strix-x570-e-gaming/p/N82E16813119111?&quicklink=true

Though really they should be just using 10GbE as the standard. Add it on all new motherboards, and it will create a market where router makers could push more volume, and thus largely keep prices in the same range as 1GbE routers and switches, but offer 10GbE instead, instead of their current method of price gouging because they feel that there aren't enough customers, but the only reason there are not enough customers is because they are price gouging.

Beyond that, odds are likely that Intel will not unlock the 6GHz band on the AX 200, even if the hardware were capable of it.

 

[Makaveli]

One issue with add-in cards is how the chipsets handle the PCI express lanes.

Due to the mainstream platforms from AMD and Intel, the direct CPU lanes are needed for your GPU, as adding so much as an X1 card to your second slot, will immediately drop your GPU to 8X, which bottlenecks most cards, especially in less than ideal situations, e.g., a GPU had 4GB VRAM, but to maintain higher texture quality settings, it needs to allocate around 5-6GB. Since modern drivers are aware of different memory pools, they can allocate system memory for texture storage, and swap as needed over the PCI-E bus. For some games, if the bus is at 16X, there will be minimal impact on performance, but at 8X you can end up with hitching and other issues. Outside of gaming, some GPU compute tasks can be very demanding of the bus.

On the other hand, the chipsets will often offer an additional 16 lanes that will share 4 lanes worth of bandwidth from the chipset to the CPU. That bus is often best to use, especially on the X570 platform where it is a PCI express 4.0 X4 connection, but due to space limitations, and often the 2nd full length slot being unusable. The first X1 slot is also often unusable because most video cards will use a little over 2 slots worth of space.

Depending on the ATX board, that will leave you with 1-2 slots at most that can be used for expansion. where you may end up using 1 slot for a better sound card, and another slot for more SATA ports, since many boards are only including 6-8 SATA ports which is not enough if you want a balance of local storage amount and some redundancy. Having the Ethernet controller added to the motherboard directly, and using some of the extra chipset lanes, frees up the 2 physical slots for more common uses.

Ideally it would be good to see a board like this get rid of the AX200, and then drop the price down to like $230, and keep the 2.5GbE https://www.newegg.com/asus-rog-strix-x570-e-gaming/p/N82E16813119111?&quicklink=true

Though really they should be just using 10GbE as the standard. Add it on all new motherboards, and it will create a market where router makers could push more volume, and thus largely keep prices in the same range as 1GbE routers and switches, but offer 10GbE instead, instead of their current method of price gouging because they feel that there aren't enough customers, but the only reason there are not enough customers is because they are price gouging.

Beyond that, odds are likely that Intel will not unlock the 6GHz band on the AX 200, even if the hardware were capable of it.

You will be hard pressed to find a performance difference from PCIe 3.0 x8 vs x16 on most modern video cards. Maybe at the highend like a 2080ti.

And I'm on a x570 build and don't have any issues with the 2nd full length slot not being usable. This will depend on motherboard and its layout.

Adding 10gbe to motherboard would increase the prices quite abit. I think going 2.5 or to 5 is enough. if you go 10GBE, you also need a 10GBE router/switch which right now will push you into enterprise gear. So $800 for a switch.

 

[mrc]

6E can have it advantages without client support. E.g. as backhaul for mesh-solutions, where the communication to clients still mainly will be on 5 GHz.

 

[Vexira]

In regards to firmware updates, I'm conflicted, one one had I can see the point in just pushing new hardware, but seeing as how android are pushing project treble and expanding the life time of kernel version support, starting with devices that have android 8 factory installed, I don't see why it can't be done, if the hardware supports 6ghz.

But on the other hand if a hardware upgrade takes place and most likely a kernel version update I do hope the build in FQ_codel and cake support into QoS provided the future kernel version supports both.

 

[Frankflash]

an AQM system would be better than just Qos it needs to work with wifi fiber etc

 

[Vexira]

I think you mean SQM which FQ-Codel and cake are forms of.

 

[Frankflash]

I meant to say SQM

 

[RMerlin]

In regards to firmware updates, I'm conflicted, one one had I can see the point in just pushing new hardware, but seeing as how android are pushing project treble and expanding the life time of kernel version support, starting with devices that have android 8 factory installed, I don't see why it can't be done, if the hardware supports 6ghz.

But on the other hand if a hardware upgrade takes place and most likely a kernel version update I do hope the build in FQ_codel and cake support into QoS provided the future kernel version supports both.

It's more than that. Device manufacturers will have to get their devices re-certified before they are allowed to use a new radio band. This is an expensive process, so it's very unlikely to be done on already launched products, unless these are very recent.