Downgrade the UEFI to 3.03. The problem with long boot comes upstream from AMD and their PI code.
Some games are just crash prone. I get GPU disconnected crashes on my 5080 in Windows on my other PC playing Dragon Age Inquisition after about an hour or so. But only in that game. Why? Because some game engines combined with some drivers are just that way
The problems in this thread are something that Framework can’t do much about and they are stuck with the problems created by others.
I’m well aware of the source of the problem, that’s why I attributed it to the firmware update. FW could have done testing and not included bad PI code in their firmware release. Pardon me for expecting upgrades to make things better, not worse.
Cool. It crashes running Firefox, and sitting idle on the desktop.
Framework could have refrained from marketing and selling a half-baked piece of sh*t.
Please don’t reply to me again, they don’t need you carrying water for their sales department.
My GPU drivers are also crashing many times + many other small issues I described here AMD Drivers Frequently Hanging and Crashing - #8 by k_peter_06
I really hope there will be a proper update either from FW or AMD to fix all these issues, as the device is almost unusable. I’ve been chatting and opening cases with FW support weekly in the past 2 months since I got my FW Desktop.
Here are also 2 other recent threads with the same issues:
Did you use the AMD bug reporting tool for your crashes? AMD does monitor these reports. And they recommend this as a good path to get visibility to the graphics driver team.
Pretty sure most of us do, but it’s still a fairly opaque process, and it’s not always clear that it’s the driver’s fault. For hardware with a premium, it’s a bit of an unsatisfying process. I chose the Framework Desktop because it combined all the positive traits I wanted in a machine. Fast, compact, low-power consumption, and despite the soldered component, at least baseline configurability. This machine is meant as my workstation and for my leisure, and I’m pretty sure many can mirror that sentiment. But as I stated in my thread, most GPU-heavy tasks, often, leave the machine in a different range of frustrating bugs.
Are you able to share the bug report with us as well? We can also add some support from our side.
Yes, and I submitted detailed reports many times, but never got an answer… so not sure if anyone is checking them.
How can I share the bug report with you? Is there somewhere a log or dump file saved?
You can send it via a support ticket or Github issue. You can also reference this thread as well in the ticket
As far as my latest experiences has been, using the drivers provided by Framework in their driver bundle and using 7zip to open up their bundled installer to pull out the AMD drivers is still lately the best for stability, may not be the best for the highest available frame rate possible with this chip set, but for stability has to be the framework provided version.
I’m still playing around with the very latest edge that AMD is providing, currently on 25.12.1 and so far with high demanding games like Need for Speed Unbound or Dying Light The Beast seems to be relatively stable, would say that it would be around 80-110 FPS for a ultra-wide 2560x1080, slightly higher FPS than the version provided by Framework. Nothing exactly deal breaking lately compared to the older latest versions that were horrendous. However for seemingly older games like FiveM seems to cause the browser to visually tear or glitch on Youtube playback, randomly crash when trying to tab out of the game to view the browser with a driver timeout, crashing with DXGI_ERROR_DEVICE_REMOVED/_DEVICE_HUNG, and unexplainable FPS drops in which when tabbed out, game FPS looks smooth, but going back into the game drops below 30 FPS. The in-game overlay says the GPU and CPU are only being used up to 40-50%. These kinds of games that break/disable AMD’s hotkey ALT+R to try to configure graphics make it hard to toggle frame-gen or anti-lag features still struggle but have been using that AMD Bug Report tool every time that it full on hangs/crashes to hopefully release another update in the future that’ll improve it. I’ve hadn’t tried changing to dx10 or dx10.1 yet for FiveM.
As for a further update as I’m playing around with what driver versions to use.
Decided to just use the AMD Cleanup Utility in safemode after that painstaking bitlocker key and needing to give the computer internet through my android phone with usb tethering to ‘skip’ the pin safeboot issue.
Went ahead and just used Windows Update to install AMD drivers itself, no framework drivers or AMD setup exe and FiveM runs 200+ fps on DX11.
AMD Driver version 32.0.12064.1002
Really being a mixed bag with AMD at this point.
Sadly, I don’t see how to extract any report, so I dug deeper myself. In the Windows Event Management, I checked and found this:
It certainly matches roughly the times when either my PC went into a freeze or when the application in question crashed.
But I’m not sure why it would crash. The temperatures of the APU hardly ever exceeded the low 60s. So either the CPU can’t handle the workloads of games like Cyberpunk or the AMD PMF Service seems to be faulty.
@k_peter_06 Could you check, if that’s also something that happens on your end? Maybe it’s something we have in common here. If you have windows, hit the Windows Button + X and select the Event Viewer
Doesn’t mean it’s the source of error, but I feel like it does indeed have something to do with AMD but I have no idea in what capacity. Flawed cooling/power handling, perhaps?
I don’t have anything in the event log when the GPU drivers crash; I get just these errors:
But I got the following ones when my FW Desktop completely freezes in games or restarts, or when it cannot resume hibernation, restarts, and boots up freshly.
And also occasionally see this one, but I don’t think this one matters at all.
Overall, I’m pretty sure these are all mostly driver issues.
What gets me about this is… one would think we would have our errors be more uniform since we are using the Framework Desktop. Getting the AMD report/Cyberpunk one more often would be putting me a bit more at ease. But in my case, experiencing hard freezes that force me to hard reset are far more common. I have to say, I’m a bit disappointed, and I doubt that framework did test their machines as extensively as they make it seem on the front page.
Run some stress tests and see what happens. Here are the ones that I run What tests should I run to check if everything is as it should be? - #5 by k_peter_06
Next, I will let it run overnight to run some memtests to be sure all 128GB of memory is fine.
I’m pretty sure all the issues are mostly because of the drivers. I keep seeing more and more reports about it on every forum and social media.
