Re: [Make-wifi-fast] sophgo risc-v wifi6 chip

Re: [Make-wifi-fast] sophgo risc-v wifi6 chip

[Rich Brown]

[-- Attachment #1: Type: text/plain, Size: 850 bytes --]



> On Jul 28, 2024, at 12:00 PM, make-wifi-fast-request@lists.bufferbloat.net wrote:
> 
> Here, the solutions need to be in full transistors
> (and very low power and very secure.) ... Wi-Fi doesn't need to be
> managed by a complex workstation and 30+ years of software bloat.

If I understand the argument, it's that we should have a pared-down Wi-Fi chip that does exactly one thing, and does it well. It should implement all the best practices (TXQs, ATF, AQL, etc.) to make it (more) secure, deliver the best performance (speed & latency). 

The fact that the algorithms are permanently welded into the silicon doesn't really matter: in practice, "no one ever updates their Wi-Fi drivers" in the bazillions of off-the-shelf routers on the market. They might as well have a good implementation.

Do I have that right? Thanks.

Rich

[-- Attachment #2: Type: text/html, Size: 3652 bytes --]

Re: [Make-wifi-fast] sophgo risc-v wifi6 chip

[Bob McMahon]

[-- Attachment #1.1: Type: text/plain, Size: 3190 bytes --]

yes, and it could be made in a pluggable form similar to the SFP. The
waveguides and the antenna are "one and done" for a building, can be put
behind walls, 40-80 year lifetime, etc. The part that is changing and
gets deprecated every few years by large amounts of invested NRE is the
Wi-Fi chips themselves. The thing that fails are lasers or optics (that's
part of the reason optics are pluggable in data centers.) So incorporating
Wi-Fi analog & digital and the optics (and optics DSP, if needed) on the
same pluggable, low power, part is a very good form factor. The upgrade
cycle is about every 2 years, for those that want the newer 802.11 features
as that standard continues to evolve.

Also, if pacing/ECN actually works at scale then the Wi-Fi chip is further
simplified, doing L4S marking. The packet stuff can be done in a Fi-Wi
concentrator that is 802.3 frame aware or at the host CCAs and the Wi-Fi
chip need worry about packets (which are an artifact.)  Just be a very good
pacer that marks in a way that Wi-Fi receives 802.11 transmits for a local
optimum, e.g. correlated with the MAC access (EDCA) and the aggregation and
the MCS and possibly bulk vs interactive.

Bob

On Mon, Jul 29, 2024 at 5:03 AM Rich Brown via Make-wifi-fast <
make-wifi-fast@lists.bufferbloat.net> wrote:

>
>
> On Jul 28, 2024, at 12:00 PM, make-wifi-fast-request@lists.bufferbloat.net
> wrote:
>
> Here, the solutions need to be in full transistors
> (and very low power and very secure.) ... Wi-Fi doesn't need to be
> managed by a complex workstation and 30+ years of software bloat.
>
>
> If I understand the argument, it's that we should have a pared-down Wi-Fi
> chip that does exactly one thing, and does it well. It should implement all
> the best practices (TXQs, ATF, AQL, etc.) to make it (more) secure, deliver
> the best performance (speed & latency).
>
> The fact that the algorithms are permanently welded into the silicon
> doesn't really matter: in practice, "no one ever updates their Wi-Fi
> drivers" in the bazillions of off-the-shelf routers on the market. They
> might as well have a good implementation.
>
> Do I have that right? Thanks.
>
> Rich
> _______________________________________________
> Make-wifi-fast mailing list
> Make-wifi-fast@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/make-wifi-fast

-- 
This electronic communication and the information and any files transmitted 
with it, or attached to it, are confidential and are intended solely for 
the use of the individual or entity to whom it is addressed and may contain 
information that is confidential, legally privileged, protected by privacy 
laws, or otherwise restricted from disclosure to anyone else. If you are 
not the intended recipient or the person responsible for delivering the 
e-mail to the intended recipient, you are hereby notified that any use, 
copying, distributing, dissemination, forwarding, printing, or copying of 
this e-mail is strictly prohibited. If you received this e-mail in error, 
please return the e-mail to the sender, delete it from your computer, and 
destroy any printed copy of it.

[-- Attachment #1.2: Type: text/html, Size: 5565 bytes --]

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4206 bytes --]

Re: [Make-wifi-fast] sophgo risc-v wifi6 chip

[Bob McMahon]

[-- Attachment #1.1: Type: text/plain, Size: 2554 bytes --]

I wrote Cisco ios software early in my career (predating Wi-Fi and ethernet
switching.) We purchased a sun workstation for $100K that supported
compiles for 8 of us. Last week I built a 7 unit Rpi5 system running full
linux, pacing CCAs, ip routing, bridging, etc., connected via a managed
vlan switch for less than $1000. It's truly amazing that what was once a
full workstation is now so inexpensive.

I think a challenge we face is that these cheap SoCs are blinding us. They
are good for prototyping but not for end users homes which is at
around 2.2-2.3B worldwide. Then maybe ten Wi-Fi radios per home (using the
term radio loosely.)  Here, the solutions need to be in full transistors
(and very low power and very secure.) The advantages of getting rid of
these complex SoCs are so great it's worth doing. Wi-Fi doesn't need to be
managed by a complex workstation and 30+ years of software bloat. It just
needs to move packets well between a waveguide and free space. (The
preferred waveguide is fiber but others will work too.) It's non-trivial
the amount of NRE to do this and our industry to date has not even really
tried. Not sure why but it seems a mistake to me.

Bob


On Fri, Jul 26, 2024 at 3:57 PM Dave Taht via Make-wifi-fast <
make-wifi-fast@lists.bufferbloat.net> wrote:

>
>
> https://linuxgizmos.com/pine64-unveils-oz64-an-upcoming-sbc-featuring-sophgo-sg2000-soc-and-wi-fi-6/
>
> --
> Artists/Musician Campout Aug 9-11
> https://www.eventbrite.com/e/healing-arts-event-tickets-928910826287
> Dave Täht CSO, LibreQos
> _______________________________________________
> Make-wifi-fast mailing list
> Make-wifi-fast@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/make-wifi-fast

-- 
This electronic communication and the information and any files transmitted 
with it, or attached to it, are confidential and are intended solely for 
the use of the individual or entity to whom it is addressed and may contain 
information that is confidential, legally privileged, protected by privacy 
laws, or otherwise restricted from disclosure to anyone else. If you are 
not the intended recipient or the person responsible for delivering the 
e-mail to the intended recipient, you are hereby notified that any use, 
copying, distributing, dissemination, forwarding, printing, or copying of 
this e-mail is strictly prohibited. If you received this e-mail in error, 
please return the e-mail to the sender, delete it from your computer, and 
destroy any printed copy of it.

[-- Attachment #1.2: Type: text/html, Size: 3523 bytes --]

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 4206 bytes --]

[Make-wifi-fast] sophgo risc-v wifi6 chip

[Dave Taht]

[-- Attachment #1: Type: text/plain, Size: 240 bytes --]

https://linuxgizmos.com/pine64-unveils-oz64-an-upcoming-sbc-featuring-sophgo-sg2000-soc-and-wi-fi-6/

-- 
Artists/Musician Campout Aug 9-11
https://www.eventbrite.com/e/healing-arts-event-tickets-928910826287
Dave Täht CSO, LibreQos

[-- Attachment #2: Type: text/html, Size: 699 bytes --]