Hopefully it keeps up with good performance. Let us know if and when you see it start to drop though. Of course, at that point bring the issue up to support again.
3 Likes
While I agree that it would be nice to get a response, remember that this is a member forum. Even though Framework employees do sometimes read and occasionally do respond to various threads in the forums, a lack of doing so doesn’t mean they are doing nothing. Most people in this thread have support tickets going with replacement mainboards and even replacement machines being sent out in some cases. I imagine Framework believes the issue is being handled at the moment. I wouldn’t expect them to give any kind of official response here until they have verified a widespread issue, and have a plan for what to do about it. If they pop in here and say “we agree there’s a problem and we’re working on it,” the forums would be flooded with people demanding updates and speculating about why they haven’t said more.
Hopefully a root cause is found and a definitive fix is implemented. In the meantime, the best thing people can do is report the issue to support. If Framework does still see it is an issue affecting a small percentage of units, the only way to change that perception is for everyone who discovers that their unit suffers thermal issues, report it to support.
11 Likes
Too bad I am in the US and only get a 1 year warranty and basically zero consumer protection laws. My warranty will be up in a few months sadly.
2 Likes
It’s actually pretty telling when a company doesn’t make the same warranty across the board. It means they are doing the absolute minimum possible to stand behind their product (or NOT standing behind it) while remaining within the law.
I think their standard warranty is 1 year and they only offer longer warranties in countries with stronger consumer protection laws. If they could offer a 1 year warranty in the EU and Australia they absolutely would.
Those of us in the US are used to this. With most other products, a warranty is nothing more than lip service that will be denied anyways. We assume there is no warranty on any new product.
At least Framework has provided support and replaced my board 3 times now. I will not knock their support team for actually providing a service in a country which hardly sees such service. I still think this has a happy ending once we get enough eyes on the problem. I could see them releasing a heatsink fix and shipping that to customers who request it, just like with the keyboard problem.
2 Likes
Sorry for the silence, everyone. We’ve been looking into this issue for a while now and believe we’ve found the root cause. We’ll make sure a solution is available for anyone facing performance degradation. Thanks for your patience so far!
35 Likes
Thank you for the update 
5 Likes
And the chad Destroya allays my worries at last.
Think the solution will bump perf up to 54W continuous or just hit the advertised 45W? Fine if you can’t say.
5 Likes
This is the news many of us in this topic were waiting for, thank you for the acknowledgment!
Can you hint or disclose what the solution is?
After facing the same issue again after board replacement I took matters in my own hands and experimented with different solutions.
I would like to understand if I would want to use the offered solution?
3 Likes
I’m not able to provide any additional information at the moment unfortunately 
12 Likes
Understandable.
We will continue to wait patiently… 
4 Likes
It happens. Good luck to y’all and your engineering team, here’s hoping the solution gives us the wild numbers people have been getting with their semi-bespoke setups.
4 Likes
actually fun story. Certain motherboards (might be older ones but might also ne new ones) actually have a BIOS setting to reduce CPU speeds for AVX 512, because they hit the CPU harder than others, potentially overwhelming the cooling solution.
Going into this knowing they specced a 45W cooler (actually more like 50W or 55W), I’m happy getting about that in cinebench.
It’s much better than some other thin and lights that will just throttle to hell and back with like a 30W solution. We have seen those.
It will be somewhat beneficial for framework to report the actual cooler designed power. Then we can extrapolate and see how much is lost in the thermal interface.
it’s just like Noctua rating their coolers for heatloads. Good rule of thumb. Though, the noctua ones you can always just get a 10,000 rpm fan (or 8,400) and get more cooling
1 Like
The irony in this case is that the cooler will smack 54W and stay there easily if the shim-vaporchamber bonding and the TIM are all good, as seen by others modifying their heatsinks. Might even do more with a little luck. I think it can do a lot more than the advertised 45W if the solution is good, and I’d be very happy to see it tap out the TDP reliably.
2 Likes
yeah. My temps are lower when I play a game, which hits the GPU, spreading the load out on a bigger area. Total package power is beyond 50Ws in a lot of cases.
Right. That silver thing isn’t built in, it’s stuck on.
Mm. Seems like framework is in quite a bit of trouble for this one. 45W and 54W is almost 20%. Oh well.
3 Likes
I mean, they ARE working on a solution. My honest hope is that they’ll find a way to make it give a steady 54W continuous- that’s the default top-end TDP for these CPUs anyway, so if they could make that work, we’d get full power out of our chips.
5 Likes
Thanks for acknowledging and letting us know Framework is aware and found the root cause. Yes, please let us know when a fix will be available, we will wait patiently.
Thanks once again and glad it is being looked into. We will wait patiently for the fix when it becomes available.
I think I agree with this, acpitz do not appear to be the cpu temps. Can someone confirm or deny why people are assigning those to cpu core temperatures? They also do not match up with what amdgpu_top is listing as actual cpu core temps. In my case all cores are pretty high, ranging between 80-100c when stressing them according to amdgpu_top.
My main worry is why I am only able to push to about 33W package power while stressing because it is thermal throttling so much.
I am currently working through troubleshooting steps with Framework Support and hope they can come back with a solution that works.
Would it be possible to get a timeframe to expect for the solution to this issue to be disclosed?
It is nice to see that Framework has gotten a grasp of the issue.