I am having my laptop basically be hot at all times if I have something like steam/discord open. Is there a way to adjust the cooling system to be more effective or limit power usage to minimize the thermals? Here is an image which shows what I am talking about:
I don’t know why but the chassis around the middle bottom part of the laptop gets really hot to the touch and the palm rest gets really warm. It is okay if I only do light workloads but it definitely does not like any degree of multitasking.
I did see some posts about the liquid metal potentially having issues for the cooling system, but overall the laptop itself just generates a lot of heat without even doing things more intensive like gaming.
Kind of want to help improve battery life and reduce heat if that is able to be achieved. OS is NixOS stable. Can provide a list of software that is installed, but I doubt it would have much to do with the situation.
If you are willing to sacrifice some performance, you can change your power profile to balanced or even power saver. This makes the CPU draw way less power therefore more battery and less heat.
You can also check out fw-fanctrl for manual fan control on Framework Laptops.
It is an Hardware Problem, most People do not even seem to care about. Uneven CPU thermals! Is a great Documentation off the Heat and Performance Problems the FW16. When the Heatsink is fixed the System is really Silent and only about 44-50C on light use. Messing with the Fan Control does not help you in any means. Do a proper Test about what TDP your System is pulling under Load. If its less than 50w and hitting 100C in High Performance Power Profile (Windows or Linux) Your System is also affected. And due to our Research a Board RMA doesnt help as any shipped Board has the same Problem.
Yes There are alot of Pictures and Ideas in the Thread i linked. Yes it is really alot to read. But i posted something yesterday in there. If you read the Last Week almost everything you need is there talked about and even linked.
Thermal Paste instead of the Liquid Metal is a temporary Fix for testing. PTM7950 is the best longterm Solution. Replacing the Original Shim is the best best overal Performance Fix, if the Performance is still low after replacing the Liquid Metal.
This sounds like a lot of work for something Framework should easily be able to fix. Have they said anything? Why did they use liquid metal if it is so much worse than other tried and true options?
Until now they deny its a widespread Problem. But look how many Posts are about System Noise and Heat.
We are trying to spread awareness of this Issue by Pointing to the “Uneven Thermals” Thread and the Work and Investigation we did. We want People to Test their Framework 16s and Reach out to Framework to RMA. If its more and more they are forced to deal with this Issue.
Until now they are walking through Tests and Replace everyones Mainboard. Wich is Okay and costs alot of Money (which is not great for such a Small and new Company), But everyone who replaced his Mainboard got the same Issues or worse Performance as before after some Time and it is not resolved. @Destroya or @nrp
I suspect most people are unaware of the problem and not getting the performance they paid for. As with me, I just assumed the fans ramping up all the time was normal. Even setting it on my bed and browsing the internet the fans will ramp up and down pretty loudly so I finally started looking around here to see if it was really normal or not.
Also, using Linux, was not entirely sure which, if any, of the temps reported were per core temps.
I would love to improve the performance of my Framework 16, but it isn’t as if I haven’t tried asking Framework support if they can resolve the issue. Unfortunately, even a chain of 100+ emails hasn’t led to a 16500+ (R9 7940HS) score in Cinebench R23.
I have already gone through a mainboard replacement and sending it to the Repair Center (which took a month and they said there was no issue with the mainboard). Now the maximum score I’ve gotten is around 15650, and they said that this is the expected score and that 100C is expected. I’m pretty exhausted from this back and forth.
Here is the screenshot of the Cinebench scores that I got:
This is the response from Framework after I sent them the screenshots of my Cinebench scores:
I would love to be able to get a more performant mainboard, but if they say this is the expected benchmark scores, I really don’t have a way to argue with them (it would be great if there was a way).
The Framework 16 is my most performant PC right now though, and I’m unwilling to risk possibly damaging my $2500+ device and voiding my warranty for the extra performance. It also requires buying many materials that I don’t have and do not want to shell out for.
If Framework is willing to better address this issue, that would be great, but unfortunately for the people like me, we’ve tried but to no avail. They likely won’t take any responsibility if you break your own mainboard trying to upgrade the thermal solution.
If your laptop has relatively high idle temps, is the CPU clocking down?
Mine goes down to 400 MHz when idle, and temps can get down to around 45c for me. Then again, I’m on Fedora most of the time. I ran Cinebench in WIndows 11.
I know they aren’t wrong, but it’s just really sad that I’m left with an underperforming computer due to thermal issues that could be pretty easily addressed.
The temps are low when idling, the main thing is this is completely underperforming for the R9 7940HS.
Unfortunately, there is no good way as the customer to address this issue sadly.
If Framework was willing to provide replacements with an updated thermal system to its first gen adopters that would be great, but it would be a once in a blue moon moment considering the financial burden that would put on the company.
I think the problem is the thermal conductivity of the CPU, not necessarily the cooling system. I don’t have a 16 so I can only use the 13 as an example. When stressing all cores the temps are between 85C to 100C, a little bit unbalanced, but the performance is as expected (3.3 to 3.6GHz, 28W). However when using a single thread the temps are 55C to 93C, a whopping 38C difference, if the heat exchange from core to core is that slow, the heat exchange from core to heat pipe could be slower. In this case even water cooling doesn’t help that much. According to other threads, the unbalanced thermal seems to be occurring randomly, which higher possibility in 16s than 13s with AMD.
I think the practical way for the company is to thermal test the mainboard before final assembly and only pass the boards when the power reaches the TDP.
Nope its the assembly. My Heatsink is modified by myself as seen in the other Thread and my CPU can pull 78w in Peak TDP now and 58w sustained. Right after i got my Board replaced the Peak TDP was 56w and sustained about 40w.
Edit and i do not hit 100c on any Core anymore when i just use the Windows Power Profiles for TDP Control.
I believe the main issue is the poor liquid metal application. The FW 13 uses regular thermal paste between the heat sink and the CPU die. Unfortunately, the FW 16 uses liquid metal and the application was quite poor.
People such as @PSierra117 replaced the liquid metal with PTM7950 and saw scores in the 16500 range, but that is only for the adventurous people who are willing to risk damaging their devices.
QC definitely could’ve helped Framework in this case, but it won’t be able to solve already affected units. So even if they apply the better QC to newer boards, I’m unfortunately stuck with this board unless Framework is willing to swap first adopter boards.
Thanks for your and PSierra117’s clarification, so it’s not random thermal of the CPU’s quality but the liquid metal cooling. Maybe we should ask Framework to use conventional thermal paste instead of liquid metal on Laptop 16 for a stable performance.
Its two factors the Liquid Metal and the Heatsink. The Heatsink is misconstructed and affects the uneven Thermal Distribution. The Liquid Metal run off contributes to the overall Performance Degradation.
