Wide-spread DPC Watchdog violations on Framework AMD laptops - can Linus get involved?

[Ste]

Hey community. Long time listener, first time caller.

 

I love the Framework ethos and how they're bringing upgradability and maintenance to the laptop scene. As such I put in an order for an AMD-powered Framework 13 - and I really regret my decision.

 

I frequently get DPC_WATCHDOG_VIOLATION crashes in Windows 11. Going through support, I've done everything from switching back to Framework provided drivers (I was originally on these, had a load of crashes, wiped windows with AMD drivers, still got them, but less frequently, before going back to Framework drivers again), for some reason removing the SSD and RAM in slot 1, basically everything Support suggested. I've lost countless hours of work to these issues and I'm at my whits end. I want to support the movement but I can't afford to lose any more time or work to this.

 

Turns out I'm not the only one. There's loads of threads across the internet with the exact same thing, with no clear resolution. Isn't it time Framework did the right thing and acknowledged the issue? Is this Linus can do as an investor to ensure clear and transparent communication between them and their customers? Thought it was worth a shot anyway.

Here's some night time reading for y'all!

• https://www.reddit.com/r/framework/comments/17ita9n/ive_had_4_bsods_with_my_amd_13/
• https://community.frame.work/t/amd-bsod-dpc-watchdog-violation/42778
• https://community.frame.work/t/random-hard-freezes-fw13-amd7840u-win11/39677
• https://answers.microsoft.com/en-us/windows/forum/all/dpc-watchdog-violation-bsod-on-a-new-laptop/f26cbf96-2b0d-4587-9f07-c88547b4cbc6

[Levent]

Have you looked into crash dumps? Usually module that is suspect will show up there. Most likely a driver issue, which amd is renowned for.

 

Let me tag Linus for you. @LinusTech

[Ste]

Hey @Levent yep, gone through the dumps, DPC_WATCHDOG_VIOLATION, some mentioning dxgmms2.sys and amdgpio2.sys. It's the very same pattern observed across Reddit and Framework's community; all on AMD, running on battery, system hangs, eventual BSOD on DPC_WATCHDOG_VIOLATION

 

Like a lot of people have mentioned across the various forms, it happens with both framework's drivers and on AMD's too. Interestingly, it doesn't affect every AMD framework user. Those who it does affect, the issue seems to persist between re-installs of Windows, which stinks a bit like a bad batch..

 

Thanks for tagging Linus, I'm getting nowhere like many others with Framework support unfortunately

[venomtail]

Oh dear, looks like early adopters are paying beta testers again 

Constantly keeps happening again and again and again with every company, every tech product.

[seanondemand]

This is why businesses buy boring-ass hardware, it’s reliable. Being on the cutting edge means you have cutting edge problems.

 

Have you gotten to the point with support where they’ve offered to RMA the system?

[mg7134]

19 hours ago, venomtail said:

Oh dear, looks like early adopters are paying beta testers again 

Constantly keeps happening again and again and again with every company, every tech product.

So relatable, even today. Thinking of getting framework after dell has the same fiasco, but things don't look good.

[Ste]

20 hours ago, seanondemand said:

This is why businesses buy boring-ass hardware, it’s reliable. Being on the cutting edge means you have cutting edge problems.

 

Have you gotten to the point with support where they’ve offered to RMA the system?

Not yet, they've now asked me to run it with just one stick of RAM, for me that's 8gb. To say it's painful would be an understatement

 

I'm hoping they offer a RMA as a next step, otherwise I'll be seeking a refund unfortunately

[nrp]

(Cross-posted from the Framework Community) Just a note on DPC_WATCHDOG_VIOLATION specifically. As you can see from the assortment of debug paths and resolutions from this thread (referring to the Framework Community thread), we haven’t seen any single root cause, but rather a catchall in Windows for any of a number of possible root causes. If you are running into this, please reach out to support. (As OP has done)

 

Note that because there isn’t a single root cause, this makes the debug process require more steps with support than most other types of issues. We start with the least burdensome steps (which are also the ones that generate the least waste), and then go by process of elimination through each possible cause. This means we start with software and driver checks, then checking reseating memory and testing with one stick at a time to determine if one is bad or a socket on the mainboard is bad. The last step if none of those were the root cause is a Mainboard swap. (Which is the next step in the support flow for OP)

[Ste]

@nrp there is a disproportionate amount of us having the exact same issue, under seemingly same circumstances, going through the same laborious to-and-fro with support. My email chain is now 14 messages deep. It's the lack of transparency that is starting to frustrate many of us. You cannot deny that this is not an isolated case, each day there's a new person popping up with the same issue.

 

Just put a page somewhere acknowledging the issue. Even if you don't know the cause of the issue that so many of us are experiencing, let us know what you are doing to investigate it. Consumers have a right to know what they're in for before they buy. Customers deserve to know what steps are being taken to investigate/diagnose the issue, and what the likely resolution is going to be.

 

I, and I'm sure many other customers, have busy lives and full-time jobs, we don't have the time to keep taking apart our laptops, and reporting our findings to support - all the while still losing time and work to crashes. I've had to borrow a laptop to conduct my work because my Framework is so unreliable. It's easy to blame AMD's drivers, but this being my third Ryzen-powered laptop, I've not had issues in the past with them.

 

Like I've said before; I want to support Framework  I love what you guys are doing and taking 'right to repair' to the next level, but I feel you're handling this very poorly and you're alienating your paying customers.

 

cc @LinusTech

[karanssh]

wow this is disappointing to read, they should do right by you and either refund or replace the device...

[Qyygle]

I feel this is more akin to a throwback to computing days of old.

Have you never encountered a weird RAM combo causing BSOD issues on desktop builds before? Nowadays it's almost expected that almost everything is compatible with everything else, assuming the form factors and interfaces fit, but it's not always been the case.

 

I've had plenty of old work desktops refuse to boot on the basis of mixed RAM combinations that by all means should've worked. On others, certain sticks would cause BSODs after a while for apparently no reason. The same sticks would then work fine in the idential config next cube down, running the same driver set, Windows, etc.

Laptops are often more complicated than desktops to boot. Desktops don't even have to run the same power saving plans and device poweroffs, sleep levels as a laptop. Outside of overclocking, voltage controls in desktops mostly push for more power not less usually resulting in more stability too.

 

This isn't helped by AMD's relative lack of experience in helping board partners develop software/hardware integration. AMD's not been well known for it's reliability in the past, it's only recently that they've finally reached a 'mainstream' audience and market share, helped by Intel's lagging in the same space. Check around the rest of the internet for BSOD's on recent AMD devices in general, even desktops. Look at Linus and Luke's own issues running just AMD GPU's in their computers. 

 

Building on this is the fact that Framework emphasizes DIY kitting of their machines. For a Dell or HP, they're sourcing their components from a few sources, buying in bulk to minimize costs. They're able to optimize drivers and hardware at the factory on a limited number of pieces. This ensures better stability out of box for most of their sold machines. Framework on the other hand, while selling their own components, also lets you bring your own kit. 

It's obviously impossible for FM to validate on all hardware on the market, that would be a herculean task and pointless. By letting the end user have this freedom, the number of edgecase incompatibilities will creep up, though this affords you other benefits like not having to deal with a hardware whitelist, firmware locks, etc. You gain some, you lose some. 

 

Both AMD's internal issues and the huge pool of possible user hardware configs, leading to an increased number of driver/hardware incompatibility is my bet of what's happening the threads you've posteed. This also jives with the fact that there seems to be a large pool of people who don't have any issue, people who found that changing power settings fixed their issue, or that swapping around ports, or RAM, etc. etc.

In the end, if you're willing to slowly work through FM's help desk process, it's possible they might fix your issue.

If you're not patient though and have the ability, try just going to a microcenter or look around on Amazon, buy ~3 different kits of 'compatible' RAM. Test for a day with each one and see if your stability improves. If it does, keep that set and return the others. You could also try the same test with other items like the SSD or Wifi card or any combination of the 3.

 

Having repairable devices is a worthy goal, but getting there isn't necessarily going to be easy. If it was, there'd be many many FM's out there already. If rock solid stability and ease of use is your main concern, perhaps you'd really be better suited with say, a Mac? 

 

[Ste]

On 1/19/2024 at 3:29 PM, Qyygle said:

I feel this is more akin to a throwback to computing days of old.

Have you never encountered a weird RAM combo causing BSOD issues on desktop builds before? Nowadays it's almost expected that almost everything is compatible with everything else, assuming the form factors and interfaces fit, but it's not always been the case.

 

I've had plenty of old work desktops refuse to boot on the basis of mixed RAM combinations that by all means should've worked. On others, certain sticks would cause BSODs after a while for apparently no reason. The same sticks would then work fine in the idential config next cube down, running the same driver set, Windows, etc.

Laptops are often more complicated than desktops to boot. Desktops don't even have to run the same power saving plans and device poweroffs, sleep levels as a laptop. Outside of overclocking, voltage controls in desktops mostly push for more power not less usually resulting in more stability too.

 

This isn't helped by AMD's relative lack of experience in helping board partners develop software/hardware integration. AMD's not been well known for it's reliability in the past, it's only recently that they've finally reached a 'mainstream' audience and market share, helped by Intel's lagging in the same space. Check around the rest of the internet for BSOD's on recent AMD devices in general, even desktops. Look at Linus and Luke's own issues running just AMD GPU's in their computers. 

 

Building on this is the fact that Framework emphasizes DIY kitting of their machines. For a Dell or HP, they're sourcing their components from a few sources, buying in bulk to minimize costs. They're able to optimize drivers and hardware at the factory on a limited number of pieces. This ensures better stability out of box for most of their sold machines. Framework on the other hand, while selling their own components, also lets you bring your own kit. 

It's obviously impossible for FM to validate on all hardware on the market, that would be a herculean task and pointless. By letting the end user have this freedom, the number of edgecase incompatibilities will creep up, though this affords you other benefits like not having to deal with a hardware whitelist, firmware locks, etc. You gain some, you lose some. 

 

Both AMD's internal issues and the huge pool of possible user hardware configs, leading to an increased number of driver/hardware incompatibility is my bet of what's happening the threads you've posteed. This also jives with the fact that there seems to be a large pool of people who don't have any issue, people who found that changing power settings fixed their issue, or that swapping around ports, or RAM, etc. etc.

In the end, if you're willing to slowly work through FM's help desk process, it's possible they might fix your issue.

If you're not patient though and have the ability, try just going to a microcenter or look around on Amazon, buy ~3 different kits of 'compatible' RAM. Test for a day with each one and see if your stability improves. If it does, keep that set and return the others. You could also try the same test with other items like the SSD or Wifi card or any combination of the 3.

 

Having repairable devices is a worthy goal, but getting there isn't necessarily going to be easy. If it was, there'd be many many FM's out there already. If rock solid stability and ease of use is your main concern, perhaps you'd really be better suited with say, a Mac? 

 

The resolution with support is a mainboard swap. With each passing day there's more and more people on both Framework's community and on Reddit with the exact same issues. As with mine, many are with the Framework-provided memory. After testing my Framework-provided RAM with a friend's HP, and with me testing a new Kingston Fury (supported spec), I've ruled memory out as an issue.

 

As for using a Mac, this is exactly what I don't want to do. I've been stung in the past where a single component failure has meant the whole laptop becomes e-waste, plus I don't want to pay the apple tax for the privilege. I don't mind the odd crash, it happens, but a daily occurrence, often multiple in a day is completely unacceptable. From my research, the intel FW laptops appear to be rock-solid, I do feel FW have proved their mission as an achievable concept. Like you say, AMD don't have the experience that Intel have with their partners, so it's an opportunity if anything else.

I'm hoping it's just a bad batch, there does seem to be many AMD FW laptops that are happy, with the stability you'd expect. Really hoping the MB swap sorts it, aside from the crash, I'm really happy with the quality and battery, screen, keyboard etc.

[thomas ponzo]

I am from batch 4 and this happened to me I can't boot my framework laptop anymore: 

 

[voltagex]

Anyone finding this, update to BIOS 3.03b or newer.