Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

synthead · 2026-06-18 00:40:24

I have a Intel Corporation Wireless 8260 (rev 3a) in my laptop, and at one point in time I was able to achieve 866.7 MBit/s tx/rx with it. I don't know what changed, but now, I am only getting 433.3 MBit/s maximum receive, and I know that this is happening because I'm only achieving VHT-NSS 1 with my connection:

$ iw dev wlp4s0 link
Connected to <censored> (on wlp4s0)
	SSID: <censored>
	freq: 5745.0
	RX: 108897 bytes (393 packets)
	TX: 41220 bytes (304 packets)
	signal: -51 dBm
	rx bitrate: 390.0 MBit/s VHT-MCS 8 80MHz short GI VHT-NSS 1
	tx bitrate: 866.7 MBit/s VHT-MCS 9 80MHz short GI VHT-NSS 2
	bss flags: short-slot-time
	dtim period: 3
	beacon int: 100

My AP is fully capable of doing 2x2 MIMO, and it even thinks that I'm on a 2x2 connection. However, my 8260 only connects with a 1x2 connection, and cannot achieve 866.7 MBit/s. Like mentioned, this has worked in the past. I don't know what changed, or why it's only doing one channel.

It's not a power saving setting:

$ iw dev wlp4s0 get power_save
Power save: off

dmesg seems happy:

[46932.471735] wlp4s0: authenticate with <censored> (local address=<censored>)
[46932.473684] wlp4s0: send auth to <censored> (try 1/3)
[46932.518943] wlp4s0: authenticate with <censored> (local address=<censored>)
[46932.518960] wlp4s0: send auth to <censored> (try 1/3)
[46932.535745] wlp4s0: authenticated
[46932.537548] wlp4s0: associate with <censored> (try 1/3)
[46932.552901] wlp4s0: RX AssocResp from <censored> (capab=0x1111 status=0 aid=1)
[46932.556201] wlp4s0: associated
[46932.587317] wlp4s0: Limiting TX power to 30 (30 - 0) dBm as advertised by <censored>

I don't really know what to tweak or check. What would you do?

Last edited by synthead (2026-06-18 00:40:55)

seth · 2026-06-18 06:58:45

at one point in time I was able to achieve 866.7 MBit/s tx/rx with it.

Can you narrow the time frame? You could correlate that w/ the pacman log, notable candidates would be the kernel, firmware, wpa_supplicant/iwd and every power management tool (nb. the TX still lists VHT-NSS 2 so it might be an issue w/ the passive activation)

It's not a power saving setting:

The intel modules (iwlwifi and iwlmvm) have power saving parameters, https://wiki.archlinux.org/title/Power_ … interfaces

Sanity check: absolutely nothing about the AP or environment (more STAs) changed? No firmware updates, remodeling your home, moving to a different table, a brand new magnetic whiteboard acting as perfect EM shield or stuff like that?

synthead · 2026-06-19 06:55:32

Thank you for your reply!

seth wrote:

at one point in time I was able to achieve 866.7 MBit/s tx/rx with it.
Can you narrow the time frame? You could correlate that w/ the pacman log, notable candidates would be the kernel, firmware, wpa_supplicant/iwd and every power management tool (nb. the TX still lists VHT-NSS 2 so it might be an issue w/ the passive activation)

This is not going to be a great answer: I've been living with this for about a year and a half and decided to finally look into it. I have considered booting a live Arch env to see if a "new" system works differently that my own, and possibly downloading old ISOs to see if there is a regression anywhere.

seth wrote:

It's not a power saving setting:
The intel modules (iwlwifi and iwlmvm) have power saving parameters, https://wiki.archlinux.org/title/Power_ … interfaces

I know, but I haven't tweaked anything, don't have any custom settings in any modprobe.d file, etc. Is there anything you'd like me to inspect?

seth wrote:

Sanity check: absolutely nothing about the AP or environment (more STAs) changed? No firmware updates, remodeling your home, moving to a different table, a brand new magnetic whiteboard acting as perfect EM shield or stuff like that?

I have been doing regular updates on my hardware, sure. I am getting these speeds while being about 15 feet away from it, so it really shouldn't be a reception problem.

seth · 2026-06-19 07:16:21

I know, but I haven't tweaked anything, don't have any custom settings in any modprobe.d file, etc. Is there anything you'd like me to inspect?

I'd first and foremost turn that off and also aspm to see whether this changes anything.
"iwlwifi.power_save=0 iwlmvm.power_scheme=1 pcie_aspm=off", https://wiki.archlinux.org/title/Kernel_parameters

15 feet away from it, so it really shouldn't be a reception problem

You think, but the magnetic whiteboard is based on a real story that's lost somewhere on this forum.
Since this is a notebook, check the behavior in direct line of sight and the notebook rotated in 90° steps.

I have been doing regular updates on my hardware, sure

Does that mean you've opened the notebook and got anywhere close to the wifi chip or antenna?

synthead · 2026-06-19 20:43:31

seth wrote:

I know, but I haven't tweaked anything, don't have any custom settings in any modprobe.d file, etc. Is there anything you'd like me to inspect?
I'd first and foremost turn that off and also aspm to see whether this changes anything.
"iwlwifi.power_save=0 iwlmvm.power_scheme=1 pcie_aspm=off", https://wiki.archlinux.org/title/Kernel_parameters

I added this to my Linux entry in grub.cfg, but it didn't make a difference. It still connects with VHT-NSS 1.

seth wrote:

15 feet away from it, so it really shouldn't be a reception problem
You think, but the magnetic whiteboard is based on a real story that's lost somewhere on this forum.
Since this is a notebook, check the behavior in direct line of sight and the notebook rotated in 90° steps.

This has been consistent for more than a year, and I have never seen it reach 866.7 Mb/s. I have the link speed in my i3bar, and am able to observe it in my userland while I use my computer.

seth wrote:

I have been doing regular updates on my hardware, sure
Does that mean you've opened the notebook and got anywhere close to the wifi chip or antenna?

I meant updates on my hardware, as in, I have updated the firmware on my APs. Sorry for the poor phrasing.
I have opened my laptop occasionally, and I did so a few days ago. I inspected the antenna cables and connectors, and everything seems fine and intact.

seth · 2026-06-19 21:35:07

This has been consistent for more than a year

https://archive.archlinux.org/iso/
Test the behavior w/ a current and a much older iso and see whether there's a difference so we can tell whether there's some regression or have to assume other causes.
If there's a difference you could downgrade the kernel, https://archive.archlinux.org/packages/l/linux/ (the 6.14 kernels or so will most likely still boot fine but you may want to install a LTS kernel in case something goes wrong) and also https://archive.archlinux.org/packages/ … are-intel/ (nb. that 1y ago is pretty much exactly at the point where the linux-firmware package got split)

synthead · 2026-06-21 08:00:05

Well, fascinating discovery: in a live env from archlinux-2026.06.01-x86_64.iso, my Rx is only one channel (VHT-NSS 1). This is the same as my "installed" env. However, in a live env from archlinux-2020.12.01-x86_64.iso, I am getting a full VHT-NSS 2, and maintain a connection at 866.7 Mbps Rx and Tx!

Both are using iwlwifi, and no module options have been configured in any env, including any power options. So there definitely was a regression somewhere that led to the default settings to halve the receive speed. I redirected a handful of commands to files and stashed them to take a look in my own "installed" system, so I'll be comparing notes and will keep y'all updated!

synthead · 2026-06-21 08:32:54

Nothing yet, but writing this down for posterity reasons: the archlinux-2020.12.01-x86_64.iso image includes iwlwifi-8000C-36.ucode with version 36.ad812ee0.0, where my updated system has a newer version of the firmware: 36.c8e8e144.0. I copied the firmware from the archlinux-2020.12.01-x86_64.iso system, compressed it with zstd (since today's kernel modules are compressed), and put it where it goes in /lib/firmware. I was able to witness dmesg loading the old firmware, but I am still only able to connect using VHT-NSS 1.

synthead · 2026-06-24 01:35:06

After a lot of trial and error, I found that this regression occurred in the Linux kernel, between Arch package versions 6.15.9.arch1-1 and 6.16.1.arch1-1. If I downgrade only the linux package to 6.15.9.arch1-1, I get VHT-NSS 2 on Tx and Rx. If I downgrade to 6.16.1.arch1-1 (one version above the working version), I can observe exactly the same behavior on my up-to-date system.

Last edited by synthead (2026-06-24 01:35:54)

synthead · 2026-06-24 05:57:12

Posted this bug to the Kernel.org bug tracker: https://bugzilla.kernel.org/show_bug.cgi?id=221685

seth · 2026-06-24 19:52:35

I hope you bisected yourself to that kernel and didn't try them in a linear latter
In case upstream remains silent, are you ready for some more chores?

https://wiki.archlinux.org/title/Bisect … s_with_Git
Gromit still has a bunch of bisection kernels at https://pkgbuild.com//~gromit/linux-bisection-kernels/
You could test some 6.16 rc's or some of the 6.15 kernels (that have 6.15.r1234.g56789 patterns, not the 6.15.rc ones that sit between 6.14 and 6.15)

Arch Linux

#1 2026-06-18 00:40:24

Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#2 2026-06-18 06:58:45

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#3 2026-06-19 06:55:32

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#4 2026-06-19 07:16:21

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#5 2026-06-19 20:43:31

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#6 2026-06-19 21:35:07

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#7 2026-06-21 08:00:05

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#8 2026-06-21 08:32:54

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#9 2026-06-24 01:35:06

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#10 2026-06-24 05:57:12

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

#11 2026-06-24 19:52:35

Re: Intel Corporation Wireless 8260's RX can't do VHT-NSS 2 (866.7 Mbps)

Board footer