Yeah, next time it happens I plan to write the relevant register to see if it unthrottles it. I couldn’t earlier today because Secure Boot has that locked down. I’ve disabled the Secure Boot checks so now I should be able to do a software override. I was worried about that causing the thing to not throttle correctly, but I figured since I was planning to get a new laptop anyway, worst case I just melt this one. Since you say the CPU can handle it itself, maybe it won’t matter.
I haven’t said anything but use case matters and we have resolved many cases. Please… let’s be transparent here. I dont know your system or use case and am not promising anything other than we will have eyes on it.
I cant legitimately promise anything without knowing the issue fully. To do so would be false. For all I know this may be a mystery box board, or CID. Not saying you did anything… but dude… no one knows as you haven’t provided info to support.
1 Like
Two questions off the top of my head:
- Why aren’t the troubleshooting steps available in a published KB/technote?
- You mean the use cases aren’t gracefully covered OOTB?
I agree with this.
2 Likes
I got it. I was replying to @Charlie_6 upthread. Next time it happens I am going to try the brutal software reset. I’ll post the results here.
Here’s the software reset procedure:
mokutil --sb-state
cat /sys/kernel/security/lockdown
You should see:
SecureBoot disabled
[none] integrity confidentiality
Then enable MSR writes:
sudo modprobe msr
echo on | sudo tee /sys/module/msr/parameters/allow_writes
When the CPU gets stuck at 400MHz:
orig=$(sudo rdmsr -p 0 0x1fc)
new=$(printf '%x\n' "$((0x$orig & ~1))")
printf 'MSR 0x1FC: 0x%s -> 0x%s\n' "$orig" "$new"
sudo wrmsr -a 0x1fc "$new"
In my case that changes the register value e4005b -> e4005a
According to AI: clearing bit 0 of IA32_POWER_CTL makes the processor ignore the external bidirectional PROCHOT request. But that could cause you to fry your laptop, so caveat emptor.
2 Likes
Thats a support escalation question, to be 100% honest. Ive seen the issue be resolved in multiple ways, but would have to go back through reports that I just cant do at the moment. Forums are not support.
1 Like
Update: that procedure doesn’t work. Bit 0 of IA32_POWER_CTL either doesn’t clear or is immediately reset back to 1. Whatever keeps reasserting PROCHOT is the problem.
PROCHOT could well be the problem.
There have been so many threads on this issue (another one here: Or Framework 13 12th Gen Intel i7-1280P sometimes running slow - suggestions for fixing the 400 MHz bug )
Seems that it doesn’t matter what OS people are running (Windows or Linux) the issue rears it head rather randomly.
Interestingly, when searching in Google about the CPU itself, the issue has appeared in laptops from other manufacturers.
It almost feels like it is very much a CPU issue which could well be due to things like manufacturing, etc.
If this is indeed the case a replacement mainboard seems to be the only way forward as that’s the only way to replace the CPU in laptops (and this has resolved it for others).
2 Likes
Different causes, so some should be in the KB.
Resolutions Ive seen: Mainboard replacements, BIOs / firmware updates, fan / sink replacements, OS tweaks… and some that are well beyond my understanding.
It all comes down to the overall symptoms and cause. Although some issues sound the same, or have the same end symptom… not all are the same cause.
3 Likes
Not every account with the 400mhz bug was me 
The vast majority of posts I ever made were wiped anyway.
But I agree with @Vikram , he’s been around a majority of the time I have while battling this issue. Contrary to @Eagle 's claims, the only way i’ve ever seen it solved is with a completely new board. (if there are any public facing stories of any other troubleshooting steps fixing the issue, i’d love to be proved wrong). I personally haven’t seen it in almost 4 years of complaining.
Point is, nobody should have ever had to have jump through any hoops, dug into code like @Jonathan_Baxter or in any other way had to modify or tweak their computer just to get it to work how it should have out of the box. Framework, from the first instance, should have just said “prove X, Y and Z to us, and we’ll replace your defective board”. I’ve never seen the words “known issue” uttered by anyone tangentially related to Framework. I shouldn’t have to troll for 4 years to get any admission, but even now it’s the same non-committal script.
To @Eagle 's credit, he was willing to engage with me despite me being a total nuisance on the forums and reddit. But in my defence, I just want a computer that works for the money I paid.
As an aside, I actually considered just giving up my crusade, but as @Jonathan_Baxter noticed, it got SO much worse in the last months, despite the PTM install and the fact I barely used any longer anyway (because it is so useless).
As I showed recently, pinning the cpu to 100% for a mere 3 minutes is enough to wreck the computer for 20+ minutes. Ingesting, converting files and building previews will trigger this behaviour, meaning by the time it’s time to work, i’m left with something worse than an rpi.
It’s always funny to see the neophytes stroll in and assume the issue is solvable though. @Squiggler , looks like you’ve only been here a few months We’ve been fighting this for years. Reddit is lousy with users that assume it’s user error (lol) It’s almost certainly not a cooling issue. As you can see below, it reaches a few degrees below 100c, does some traditional throttling for a few seconds, then dumps itself to 400mhz (which isn’t picked up by HWInfo as actual thermal throttling). Temps go down, clock stays at 400mhz. The full width of the window here represents 10 minutes.
I’m currently waiting for a response from Eagle in our support email chain. I hope he can do the right thing 
I understand why you’ve forgotten your manners - it’s the same reason why the OP started a hot thread in such a counterproductive fashion.
Of course, I am not going to read the whole of the forum to see every issue that has ever been had. This is a peer support forum and my questions were intended to encourage the OP to set out some context. Happily, I appear to have been successful in that endeavour.
I’m just a Pro 13 customer with a desire to help a small ethical company do well. I’ve been here a little while, and I witness that a certain category of complainant has given up with the support process, and they then blame front-line staff individually. From memory this seems to be the fourth such case I’ve noticed, and it’s not like I am enjoying finding them.
In such cases, staff have to wade through snide, hostile, needling, and passive-aggressive language directed at them in particular, and on occasions there will also be accusations of wilful omissions of information and/or deliberate dishonesty. I am of the view that these claims in general are unfair, and that such complainants are motivated by an animus against the company (or its agents) that stem from their consumer disputes.
These kinds of tone are counterproductive in terms of resolving the technical problem at hand, but moreover workers should not have to tolerate such an environment. Even if FW were exploitative or uncaring, their staff are not.
Yep. I have every confidence that you can do the right thing too. 
1 Like
Let’s not turn it into a seniority contest, please.
I get it, you have a faulty board. But there’s absolutely been a variety of other causes. I myself have fixed a board with the 400MHz issue. Working through this troubleshooting with support isn’t fun, but it’s necessary.
I personally haven’t seen a fix for 12th gen Intel. Even if there isn’t one now, they’re probably working on a patch now. Just last week, a BIOS patch solved a similar issue on FW16. (For various reasons, I’ll opt not to link to that issue here).
1 Like
Ok, I agree the title was inflammatory. I have already changed it. And my frustration is evident, yes. But I don’t see the thread as counterproductive. If the problem is resolved, it will be here for others who have the same issue. Nowhere else is a resolution listed or even an acknowledgement by framework that there is a problem.
I have troubleshot this issue to death: it looks very much like a design issue with the way the main board handles overheating. Based on the lack of public information from framework about this issue, I have less than 5% confidence my opened ticket will succeed without a major repair (e.g. motherboard swap). Even then, the correlation with recent change in my workload has me skeptical.
I’m just a Pro 13 customer with a desire to help a small ethical company do well.
Leaving a major design issue unresolved and unaddressed is not particularly ethical. I bought this laptop with much hope for a repairable, modular platform. I was rooting for framework as much as the next guy. Unfortunately, I have been disappointed by their thermals, battery, and overall build quality. And now this near-terminal issue.
It was an experiment for me, having bought laptops almost since they first existed (my first was a 486; I installed pre-1.0 linux on it with about 100 floppies. I was sad to see linux support for that chip ended).
I agree with you in part, Jonathan. I think things started off a bit unproductively in the thread, but it looks like it’s sailing now. I wish you and Support all the luck in the world - not least because the BIOS and/or CPU could all be a factor, and their respective technical departments may need to weigh in.
I don’t envy FW being in the middle of it all!
How? Even if it’s for a different model, the information may be helpful.
Simple, I had to chuck my HDMI card. This was almost 5 years ago, 11 gen Intel. It worked fine later when I got a replacement. moreaccountsthanmhz already tried this, though.
Yeah, I already bought new USB-C cards, since I saw a reference to that. No luck. I’ll remove the HDMI card but I never use that so I doubt it’s the issue.
Although I have had this problem intermittently for a while (at least 12 months), I can reliably trigger it now with a specific AI workload that exercises the graphics hard (WebGL).
This is what ChatGPT thinks the likely cause is, fwiw:
Heavy WebGL/iGPU load
→ high SoC or power-delivery load
→ board sensor, VRM, charger circuitry, or EC asserts PROCHOT
→ all CPU cores are forced to 400 MHz
→ assertion remains latched after the workload stops
2 Likes
I’m not. As they said, they didn’t read the whole forum. I’m just pointing out there’s a trend of people (whatever the neutral way of saying:) mansplaining (is), when they have literally no clue what it was like to deal with Framework in the first 2 years.
Whatever emergency state is triggered by whatever event on whatever board has every possibility to be fixed with whatever troubleshooting step. Searching PROCHOT gives many different clock speeds over many different boards (a design issue with FW’s underlying design/implementation/engineering?) But the point we are making, is that 12th gen is defective. I ran through every step support gave me, mods gave me, community members gave me, installed every BIOS update that claimed to fix it, and nothing worked. Wouldn’t you find it frustrating if the only support they could give you is a link to the mainboard section of the store? For a problem you didn’t cause?
They’ve had 4 years since it was first brought to their attention. We were told to wait for BIOS updates to solve the issue after board replacements didn’t work. The first update didn’t come until my warranty had expired, and didn’t solve the issue.
Instead of defending Framework, perhaps stand of the side of the users? Something something that quote about good men standing by. Allowing them to shaft users is going to circle around to everyone in the long run.
And what would that be?
Wouldn’t you be a little frustrated after 4 years of evasive, non-committal language and broken promises, and a computer you literally cannot use?
And this is where my main problem lies. A collection of words has never resonated with me so strongly. I took a very expensive chance on this then (then extremely) tiny company who seemed to be morally sound. Only for them to, not only leave us hanging in the initial stages of the board’s life, but put us on the hook for the replacement is literally abhorrent. You simply cannot get away with this when you market yourselves to enthusiasts. Just because it’s easy to replace, does not mean I want to pay for a problem I did not cause.
If Dell, Lenovo or another giant faceless company had told me to pound sand, yes, i’d be absolutely pissed. But when your website has the quote:
We’re here to remake consumer electronics to respect people and the planet. Unlike most products, ours are open for you to repair, customize, upgrade, and own at the deepest level.
I will absolutely not let you off the hook for this behaviour when I have multiple thousands invested.
Even if it was a solution though, why should one need to compromise their experience in order to have the machine operate how it should? Perhaps I want to use an external monitor while using the cpu at the same time. There’s no excuse for bad engineering at this price point.
Your having a valid consumer complaint, and your having a generally injurious persona, are apparently not mutually exclusive conditions. I hope FW is able to resolve your problem, but they will be able to concentrate on that task better when you learn to control your temper.
As I alluded to earlier - the fix here could involve the technical departments of FW, AMD, and Insyde. Cross-company collaboration is phenomenally complex and, while I agree that it would be nice if things worked OOTB, it is understandable when technical problems arise.
Hey! I found your support ticket. Would you please reply to it and put in all the troubleshooting you’ve already tried so we don’t ask you to redo things unnecessarily. All that came through in your ticket was ‘I have an issue with my laptop’, your OS version, Bios version, order number, and a link to this thread, but this thread doesn’t list all the troubleshooting you’ve tried.
This should save you a bit of time.
1 Like
In case it helps the OP - putting a full repro case on GitHub or some other public site can be very helpful. It also allows curious watchers (with the same hardware) to try the same thing and report their results. Plus it is also very easy for an investigator to send that link to collaborators outside their company. I’ve done this many times elsewhere - it can speed up the process.