Uneven CPU thermals!

Guys, please stop asking Destroya. Framework will most probably not give any estimate in the Forum and no possible changes will be released until after maybe a DeepDive or official Announcement. The Call in here was purely to calm us down and assure us we are heard. This is fine. And after all, also i got some hope back after the blatant dissapointment. Lets see what will happen.

8 Likes

What?

Action speaks louder than words.

I do have an Intel laptop (not Framework) and on that s-tui shows core0 core1 core2 etc etc, in addition to acpi that shows temperature of somewhere else.

Is it possible to get 54W on thermal paste used on FL13?

Intel and AMD don’t have the same sensors names

Be interested to know if the “fix” will be made available to everyone including those that have already had RMAs.

But will wait for the official word.

It would also be wonderful to know if Framework will be proactive in getting the fix to people who have 16s. Emails to customers and such to notify them to check if they’re having the issue and replacement of the part, or just replacement of the part in general.

All of us here are aware something is amiss with the cooler, but there are still lots out there that may not even know something is wrong.

3 Likes

Has anyone compared the performance here in terms of cpu core thermals against that of another laptop with the same chip?

It’s not really a valid comparison, the issue is that some cores on the FW16 are hitting 100c and causing thermal throttling. And the difference between cores is ~30c in some cases.

Clearly its a CPU and heatsink thermal interface issue.

My RMA board (so far) has not even touched 100c and the core temps are much more uniform.

That’s exactly why I asked. Has anyone demonstrated that this is not seen on other devices?

You’re not understanding the issue. Uneven temps would we a problem with the cooling solution. It would be a completely different design on another brand laptop and therefore pointless.

You are essentially asking someone to check a BMW for weak power steering because your Toyota has the same steering wheel design.

Yes, because I what I want to know is if this is a common issue that many manufacturers struggle with, or if it’s unique to Framework/Toyota.

I’ll take the attacks on the question as a “no, no one has checked”.

My opinion is that it is a function of the design of the cooling solution and the the design choice to let the chip run hard. It seems that there is an adjustment that needs to be and presumably will be made, which those of us running into problems will need to apply to our systems.

For me personally, I’d love to have the liquid metal be gone, due to the asymmetric risk/reward. I doubt that that will be part of the solution, but if I find that whatever refit solution doesn’t resolve things, I likely will have it removed professionally so that I can hack at the machine with less concern about destroying it due to liquid metal getting out of its cage.

Nope its a 100% the Cooling Solution.
When i run my PPT within Frameworkspecs with my modified Heatsink i do not hit Thermal Limit anymore and have 54wTDP sustained with 97C on the hottest Core. And thats without Liquid Metal.

2 Likes

I agree that it is the cooling solution. I was also pointing out that Framework does not limit how hard the chips can run, which some other manufacturers do, although I can’t say if there are examples of other manufacturers doing it with this particular chip.

If whatever solution Framework comes up with does not work out well, my plan is to have the liquid metal professionally removed and replaced with PTM. Then, if I want to, I can proceed with your solution of removing the existing shim and making a PTM-shim-PTM sandwich.

Thanks for your effort, it is good to know that the cooling solution is capable of allowing the machine to run hard when it (the cooling solution is functioning properly, even if that means modifying it.

Well partly no. Framework does limit how hard this Chip can run. It can go to 65w for i think 10-15s (havent measured) and than it is locked to 54w under Full Load (best performance Power Profile).
Other Manufacturers like Asus in the G16 let this Chip run up to 80w and its getting cooled just fine, even with a GPUshared Coolingsystem.
The full Systempotential gets unlocked when overriding the PPTs with x86 Universal Tuning Utility (with proper functioning Cooling)
If i remember correctly its throttled to 50w tdp in balanced.

3 Likes

Good to see an official response about the issue. Given they also contacted people who bought the 16 before the keyboard deflection issue was completely fixed and offered to send a kit for users to install, I believe they’ll make good on letting people know what the problem is and offer the solution to us. I just hope CoolerMaster are the ones taking the hit on poor build and quality control rather than Framework.

4 Likes

Framework will have to do some number crunching and math what makes most financial sense, either put the money and resources into a fix of the current board or include the fix with the Ryzen 9000 series boards(and offer discounts to existing customers)? I am assuming that Framework is skipping Ryzen 8000 series or we would have heard something about it by now…again just a lot of speculation at the moment…just have to wait until a decision is made how to address this issue.

There is no Ryzen 9k mobile. Those are the Ryzen AI X 3XX and well those are best performing with lpddr5 but ddr5-5600 is supported. We will see what route they will go or if they possibly go to LPCAMM2 Modules.
For the Price the Devices currently Cost they have to bring the fix to every User. Else it would be a serious hit in the Trust Factor.

2 Likes

Agreed, I do hope Coolermaster is accepting it’s share of the blame as a business
partner and helping with a fix.

Now I’m wondering just how much this chip could be pushed to do in the 16 with good, high-quality cooling, even in this form factor. If a lot of the issues are due to use of glue and such instead of solder, or improper amounts of solder, plus the LM runoff, it’d be really nice to see a serious improvement.

1 Like

Interesting, I didn’t realize that, thank you for the correction and explanation.