AMD Working On WiFi RFI Interference Mitigation For Linux

[phoronix]

Phoronix: AMD Working On WiFi RFI Interference Mitigation For Linux

As a step toward further improving AMD laptop support under Linux, AMD engineers have been working on WIFi radio frequency interference (RFI) mitigation support for Linux with their latest laptops...

https://www.phoronix.com/news/AMD-WiFi-RFI-Mitigation

[tildearrow]

Typos: mitikgate, wiFi, WIFi

[M@GOid]

Interesting, didn't knew about that. I wonder if those mitigations can cause any significant performance loss on the WiFi 6 devices.

In the end WiFi never reaches the advertise numbers anyway. This is why I "cable all the things!!!1!" when bandwidth is important.

[Mr.Elendig]

Yay software solution to an engineering problem :/

[aviallon]

Yay software solution to an engineering problem :/

You may not have a choice here, actually.
Regulations are the limiting factor.
You can't simply use any frequency you like for any purpose you want.

[Mr.Elendig]

You may not have a choice here, actually.
Regulations are the limiting factor.
You can't simply use any frequency you like for any purpose you want.

But you can filter and shield.

[ryao]

Interesting, didn't knew about that. I wonder if those mitigations can cause any significant performance loss on the WiFi 6 devices.

In the end WiFi never reaches the advertise numbers anyway. This is why I "cable all the things!!!1!" when bandwidth is important.

Historically, mitigations for Bluetooth co-existence could cause wifi to become unusable in crowded environments. I experienced that first hand on Intel wifi chipsets like the 6300AGN. Turning it off via a kernel module parameter was one of the things I had to do to get usable wifi on a thinkpad t520.

AMD does not make wifi chipsets, so I am surprised to hear that they are working on this.

[Jakobson]

AMD does not make wifi chipsets, so I am surprised to hear that they are working on this.

AMD's PM has that knowledge of the clock speeds of memory. AMD also wants sell processors for laptops. By implementing the feature, it helps selling more chipsets.

[oiaohm]

AMD does not make wifi chipsets, so I am surprised to hear that they are working on this.

From the article.

In particular, working around interference from the GDDR/DDR memory clocks with the local radio module frequency bands used by WiFi 6 / 6e / 7 can cause problems.

Ryao the catch here is your memory controllers clocks can make radio noise on the the same frequencies the wifi module is attempting to use. So yes AMD does not make wifi chipsets but that not the question. The question is does AMD make anything that broadcasts in wifi frequency bands? The answer is yes amd does make items that broadcast in wifi frequency bands and it the MMUs in their CPU and GPUs. Intel did the same thing with their RFI a while back with their MMU. Allowing the wifi module and the mmu to know what frequency each other is using allows noise avoidance.

This does link back to fcc regulation and the like for EM noise and where it allowed.

[X_m7]

I wonder if there's any chance that they'll make Intel Wi-Fi adapters work with this too? My AMD laptop (Ryzen 7 6800H) unfortunately comes with a MediaTek MT7921, while it works fine on a Wi-Fi 5 network with Wi-Fi 6 it sometimes ends up with really awful download speeds even with line of sight to the router, checking the output of `iw wlp6s0 link` when that happens shows that the transmit bitrate is at 1200 Mbps while the receive bitrate is at like 20 Mbps or whatever stupidly low number, and its driver doesn't have any options to disable Wi-Fi 6 and just drop to Wi-Fi 5 from what I can tell. It's so bad that even switching to 2.4GHz only provides a better connection, and at worst it even has trouble just connecting to the network.

So, I ended up just swapping it out with an Intel AX200, and since then I haven't had any issues (speeds are as expected, and it connects quickly and reliably), so I guess maybe it's not something that'd affect me anyway ¯\_(ツ)_/¯