What’s funny is other manufacturers have tried using LM on their laptops, Asus ROG is one example. They always end up with the same result so I don’t understand why Framework repeated all their mistakes.
1 Like
I’m disappointed it didn’t work out in the end, as many are, but I personally don’t fault them for having the passion to try. It was one of the things that made the FW 16 exciting for me.
1 Like
Framework tried another approach with their phase change liquid metal. Its hard and dry when the device is off and cold and liquifies when heated, but as it seems this increases the pumpout. They added the etching to minimize the pumpout but well we all know how that helped.
2 Likes
I can’t overly blame them for following the trend and trying to innovate over the issues. But given they’re a small company with limited budgets I personally think they should innovate in more targeted ways going forward. Their whole thing is repairably and long term use. Liquid metal doesn’t really fit into that category. It shows by how long they’re taking to come up with a fix and how many people are worried against repairing it themselves. Had this been a thermal pad or paste issue I think most would have done it themselves and it’d have been far easier. Then again neither is 16 screws to get into it 
that could have done with some innovative thinking.
I hope now they have their 12 & pc out they’ll refocus back towards the 16 and focus on pushing what the 12,13,16 and pc can do rather than introducing yet another line.
4 Likes
For some days I’ve noticed my FW16 performing slower than usual sometimes and then I recently took a look at amdgpu_top and I’m seeing in the GPU metrics section
Throttle Status: [PROCHOT_CPU, PROCHOT_GPU]
and sometimes the CPU frequency doesn’t go above 900MHz~
Yeah, I want Framework to get that thermal padding sorted ASAP…
1 Like
I have my RMA still on hold. Will ping support again in early April if they have finally exhausted their existing stock of boards with Liquid Metal. Last time I checked earlier March they still had some left and unable to guarantee my RMA would have the new board without the Liquid Metal… still have 4 months left on warranty to honor the RMA.
Mine goes into this state quite often. Seems only a reboot fixes it. Like it triggers into some protection mode that needs to get reset.
I did - but only because I wanted to test the application of the PTM on something before I broke open the CPU.
My kid’s laptop, my steam deck, and my F16 CPU and GPU all have PTM now.
Ill eventually do my wife’s F16 - just been too lazy 
1 Like
Probably because at the time people were trying LM on their FL13 units with success at the time that the FL16 was in development.
I don’t think Liquid Metal is an optimal thermal solution for laptops. Also the rapid degradation of LM is a concern.
Doesn’t seem the ptm fixed my issue 
Cools down really quickly after it’s finished it’s work load but doesn’t keep it down
It’s really weird because install something and 1 core flies but play a game and they stay fairly happy. If anyone has suggestions as to why I am open to them 
What in gods name are you doing that has 1 core at 90+ and the rest in the 50s?…
Are you sure the monitoring sensors are reading right?
Or that you fully covered the whole chip with the ptm? Might also check your mounting screws are all snug. At that kind of difference, I dont think any thermal material is the issue here…
I had the same issue before I did anything. It was the reason for changing things. In this particular case I was installing/compiling a bunch of amd graphics drivers. 1 core spikes and the fans go mental. After the change I still had that 1 spike up but under 80 until now
Not sure what you are referring to because nothing you posted is showing per-core temps. Those “acpitz” (acpi temperature zones) are not all cpu cores. In fact, I think only temp4 is the cpu core. Here is a list of the same things from my machine with different labels (using a different sensor module I think):
ambient_f75303@4d: +45.9°C
charger_f75303@4d: +45.9°C
apu_f75303@4d: +48.9°C
cpu@4c: +60.9°C
gpu_amb_f75303@4d: +54.9°C
gpu_vr_f75303@4d: +52.9°C
gpu_vram_f75303@4d: +53.9°C
gpu_amdr23m@40: +56.9°C
The only program under linux I have found to give per-core temps has been “amdgpu_top” monitoring the APU’s 780M. Wish I knew where/how it got those.
Here is a screenshot of mine stressin’:
Oh look, I managed to trigger that low power mode it gets stuck in until a reboot just by running “stress -c 16” for this screenshot.
1 Like
Could you use amdgpu_top to verify which core is the hottest?
2 Likes
Those arent cpu temps my guy.
Use amdgpu_top.
IF you want to see if your issue is resolved - you need to run a benchmark like Cinebench R23 - and compare with the results in this thread.
R23 is BY FAR the most used tool to verify F16 thermal issues by the community - so it is the most efficient way to see if you still have issues.
1 Like
Agree, Cinebench R23 is definitely the tool to use. I see my performance dropping steadily over the last 7 months as the Liquid Metal degrades. Can’t wait for the official fix kit… almost April and not a peep when it is expected… if ever!!
Cinebench is not in linux, is it? I had used geekbench and just tried stress-ng. And yes under amdgpu_top the core temps all stay roughly together. It was just me misunderstanding sensors Thanks for the info guys.
You can use cinebench via proton/wine. Its ±100points the same score and in cinebench it detects “windows 10”
3 Likes
As my thermals problem is worsening really bad, my CPU has started coil whining very loudly.
It started with a boot where the coil whine was a constant, since wave sound, audible through headphones.
I was stupid not to record that, because, from the next boot onwards, it turned into a really bad screeching that is much louder, but happens unpredictably. It’s like a very loud scratchy sound that sounds like something is being physically grinded.
Did this happen to anyone else? So far this is my n-th straw with this unit, combined with chassis bending and of course the CPU thermals issue, I am starting to think it is within my domain to politely request a full replacement.