AMD Ryzen™ AI 5 340 random moving pixels on resume from suspend

[Fuzzman]

Which Linux distro are you using?

Reproducible on multiple, including Ubuntu 25.04, Fedora 42, Arch (rolling)

Which kernel are you using? 6.14, 6.15, 6.16

BIOS is updated to 3.04

AMD Ryzen™ AI 5 340

Initial installs are flawless on every distro tried. Anytime laptop is resumed from suspend, there will be random small - usually red but not always - pixels that flash as things change on the screen (typing or video or scrolling). If nothing on the screen changes then the pixels just stay as-is. Sometimes it will show up after a reboot without suspend following a period of usage, but generally a fresh reboot cleans it up until I suspend the machine.

I’ve attempted to screenshot this behavior but it does not show up using native screenshot tools. Will try taking a pic with the phone camera to see if that helps show it.

Adding kernel params such as dcdebugmask=XXXXX (multiple options), and I’ve attempted to downgrade mesa to older versions as some forums have suggested.

Nothing works. Was excited to receive this new laptop this week. Now I’m 3 days in and beating my head against a wall.

[Fuzzman]

The circled line of dots is what I’m referring to (had to zoom in on a single set in order to make them visible).

Additional information - as I move the mouse pointer the dots will randomize, growing and shrinking both in number dots and in groups across the screen.

Anyone have any thoughts?

[Eagle]

Hi @Fuzzman Are you running any other software when this happens? Any games, etc?

[Fuzzman]

Nope.

Vanilla install running nothing can reproduce. It does not happen on an external monitor, but the built in screen will have it no matter what I’m running.

BTW, it occurs with and without an external monitor.

[Eagle]

Thanks @Fuzzman I’ll get this in front of the Linux team today.

[Fuzzman]

Additional information.

I’ve tried various desktop environments. Occurs on both wayland based like plasma and gnome and xorg based like bspwm.

It is more pronounced when there are images on the screen. Meaning a wallpaper, website, etc as opposed to a solid color background, but will occur then too. The dots align with color boundaries, such as where the boundary of an image or icon begins, or within an image where there is a color change. Not all the time, but predominantly.

[Fuzzman]

Video after fresh Fedora 42 install. Only Firefox running.

August 25, 2025

[Fuzzman]

Downloaded and installed Windows 11. Ran for almost an hour. Multiple apps and websites tested. Suspend and wake operations multiple times.

Zero issues. Right now it doesn’t appear to be a hardware issue.

Will be reinstalling Fedora 42 and messing with kernel params and such again. Also, I have seen some instances of the USB A expansion and the HDMI expansion cards not recognizing devices on first plug in. Removing both and seeing if there is something there with Linux.

[Fuzzman]

So to summarize findings to date:

1. Problems persist on any Linux distribution installation (Ubuntu, Fedora, Arch, CachyOS, Linux Mint)
2. Problems persist when mesa packages are downgraded to <= version 24 (per some forum messages) or when running latest package (25.2.x)
3. Problems persist when passing kernel arguments (dcdebugmask=0x10, et al)
4. Problems persist with multiple kernels (6.12.x through 6.16)
5. Bios is upgraded to 3.04
6. Problems persist at multiple resolutions and refresh rates
7. Problems persist when adjusting VRAM, even all the way up to 8GB
8. Problems are not specific to an application (i.e. Firefox, games, etc.) but will occur randomly no matter what is/isn’t running
9. Problems occur running both Wayland and Xorg based environments
10. Problems do not persist when running Windows 11 so it does not appear to be hardware related, however I bought this specifically to run Linux and did so based on the product specs page as seen below. (URL is for this specific model and tux logo)

At this point I have to wonder if you ever even tested this model (CPU/GPU) with Linux before selling it and advertising it as Linux capable.

Please advise how to address this, or provide the relevant information to obtain a refund.

[tlvince]

@Fuzzman which display are you using? While debugging a different issue ( Flickering when using Firefox under KDE Wayland on Ryzen AI 300 ), I observed the same with the 2.1k display, but not the 2.8k.

[Fuzzman]

2.1k. I saw your post during my troubleshooting. Mine doesn’t have a wide and tall bar flickering, my problem is random colors - mostly .5mm thick or smaller - popping up all over the place and flickering.

I will add that this occurs on the desktop, running Firefox, running Chromium, running an image viewer. In other words, not specific to an application but a general problem.

[tlvince]

Yeah, I also see random coloured checkerboard artifacts with the 2.1k. I’m fairly certain this is a amdgpu or mesa bug.

[Fuzzman]

Probably.

My frustration is that having cut my teeth on Slackware back in 1999, I’m no noob to trying to get Linux to run on things, but this is the first time I’ve purchased a machine that explicitly advertised itself as supporting Linux, going so far as to list an officially supported set of Linux distributions.

To me that means that the makers had actually put their officially supported Linux distro through its paces. Even when I downgrade packages, kernels, etc., the problem persists, so it is evident that they didn’t test this combo when first released. Or at least didn’t test it longer than it took to install and check that modules loaded as expected and the basic keyboard functions worked.

[Fuzzman]

I’m being offered a replacement 2.1k display. I will take this and try it out, but based on other posts I’m highly suspicious that the issue is mesa + amdgpu + this particular 2.1k display.

If I could pay the difference and get a 2.8k display to avoid being stuck in a loop of retrying the same hardware, I will do so.

I will see what Support says.

[tlvince]

In my experience (and others who’ve received hardware swaps), the replacement 2.1k display will not resolve the issue and the issue is still there on the 2.8k display, but manifests differently. Please keep us updated nonetheless.

[Fuzzman]

ROOT CAUSE IDENTIFIED….(maybe - see bottom update)

As part of troubleshooting everything I could think of, I replaced the SSD with another, older one that I had laying around.

So far I’ve gone 2+ hours without a single issue with the display. For reference, the original SSD that I was seeing problems with was a Kingston SNV3S 1TB.

The one where so far no issues have occurred is a Kingston SNS8154P3 256GB.

Will continue to use throughout the rest of the day and see how things do or don’t progress.

UPDATE - NO GOOD

Nope.

Still seeing artifacts, though less consistently. Will see what the new SSD does, but I’m no longer convinced that this is the culprit.

UPDATE - NEXT MORNING (Aug 27)

Yesterday, I had Amazon same-day deliver me a WD SN7100 2TB nvme since that is one of the models Framework sells with their configurations.

Installed it and then put Arch on it w/Hyperland (wayland). Setup my profile a bit and opened a bunch of tabs in Chromium, couple of alacritty windows, etc.

And I let it sit without suspending all night.

As of this morning, this system is way more usable than it has been since receiving it almost 1 week ago. I have seen exactly two odd artifacts in Chromium. Nothing from the desktop background or other apps (yet).

Whereas before I couldn’t scroll a website without artifacts all over the place, now I am seeing just an odd one here and there and not consistently.

So in summary:

1. There is a correlation between the SSD used and the graphics system (PCIe/GPU isolation issue??)
2. I will be re-installing my original Kingston SSD to validate if this is correlation or causation (after installing all the same software as on the WD SSD)
3. Will also be testing today with a lot of suspend/resume operations with the new WD nvme since that seemed to make things worse on the original Kingston
4. VRAM is set to the default (I had previously tried giving it 8GB since others in forums had said that would help - it didn’t)
5. Will report back as I progress through the day

UPDATE - PROBLEMS PERSIST (Aug 27)

Ok, so that theory is out the window. Unplugging the external monitor and running off of battery absolutely continues to have artifacts show up. With an external monitor plugged in and using the built-in display while powered with the power supply, there is only minor amounts of artifacts happening.

When unplugged from both external power and an external monitor, it is back and more widespread.

I should have gone with the Intel chipset…..

[Fuzzman]

Video of issue using WD SN1700 SSD, unplugged (battery only) and no external monitor - go full screen as it is somewhat blurry. This is a static image on a web page. Video playback is much worse.

August 27, 2025

[Fuzzman]

On AC + external monitor means fewer artifacts on the built-in display.
Battery powered with no external monitor makes mobile use unusable.

[Fuzzman]

Hey Framework Support team, how much time should a customer expect to wait between filing a support ticket and getting a response? Other than an automated response Monday morning, I am still waiting on someone, somewhere, to respond from Support. (not talking about Eagle - they have been great to correspond with but is not the person who handles support)

You advertise this laptop, specifically with this chipset, as being Linux compatible and I, and others, can confirm that it does not work out of the box with Fedora 42.

I do not want to go through support hell of replacing part by part. Every option both in this community forum and a multitude of others has been tried. Kernel parameters that shutdown various AMD power saving options, ramping up power consumption to 12 or more watts also do not fix this.

Fedora 42 (and others) do not work with this chipset when on battery power. The screen becomes unusable with artifacts displaying over every image.

I am formally requesting one of two resolutions:

• Full refund
• Replacement with an Intel chipset (I will pay the difference between the two - not looking for a handout)

[Fuzzman]

Note that for the above statement that I was “offered a replacement 2.1k display”, that was via a chat with Eagle telling me that Support wanted him to get me to open a ticket for a replacement.

I had already opened a ticket Monday morning which we tracked down. Still no response from Framework Support to that ticket.