“Cooler Master” if you happen to be reading this thread we are looking at you!!! I do want Framework to suceed but at the same time I did drop $2000+ on this machine and looking forward to a proper fix to this issue.
That is so true. I noticed that my CPU is capable of running at 54W, but due to uneven thermal dissipation, it can only sustain 38W when running on its own. In contrast, when the integrated GPU is active (e.g., during gaming), the system can consistently draw 54W with stable temperatures around 85°C. After about 10 minutes, the power draw drops to 45W—not because of a thermal limit on the CPU, but likely due to power restrictions set by the Framework. Interestingly, if I unplug and reconnect the charger, the CPU can once again temporarily pull the full 54W for another 10 min.
1 Like
you can try running hwinfo in the background and monitor the Limits, When it is throttling to 45w not due to Thermals it is possibly the STAPM Limit or the PPT Slow Timer.
How can you bypass this?
Hey there,
First and for most: thank you for this megathread. I’ll be as concise as possible to share my experience.
I had no clue about a wide spread issues, all I knew was that my laptop was very very noisy but thought it was normal as I was used to the GT76’s noise prior to my FW16.
Opened a ticket end of october. Got RMAd. Great support experience dispite my less than ideal response time (I really want to emphasize that, I know not everyone got the same experience but mine was astonishingly pleasant).
Recieved a new board last friday, installed it: it was DAY AND NIGHT. Dead silent even in performance mode (could not use performance mode in meeting even if only web browsing). Went from high 11s on CBR23 (max tdp of 30W) to mid 15s and a TDP of 48w.
Now a week later and started to see an increase on the fan noise. Ran few CBR23 today and I can’t break the 14700 barrier.
Here’s my last and best run.
Just wanted to share as I believe having as many data points as possible is a must regarding this issue. Will keep you updated on the degradation if any.
I don’t trust myself on repasting or doing anything related to LM. I’m not tech savvy, just a guy that liked the concept of the FW16.
B4 - R9 7940HS - RX 7700S - 64G
6 Likes
Thanks for sharing your test results and experience. Framework is aware of this issue but there’s no verified/official fix yet. The more people report and share feedback hopefully we will get more traction for a fix sooner rather than later. I am confident but it is not getting fixed overnight we have to wait patiently as to what Framework decides.
I did some Thermal Limit Tests with my Mod.
65C 80C 90C 100C set as Temp Limit and watched the CBR23 Scores and the maximum TDP.
Great Result, this is now my Baseline for my next Mod, as i will try to solder a Shim to the Vaporchamber.
4 Likes
Its been a few days with the replacement RMA board, so its had a few thermal cycles.
Latest R23 test - 15.6K
Held about 47-49W through the 10 min test, and peaked at 60W. Still a 15c difference across core temps 85-100c and core 3 being the hottest, but its better.
Since then, I have to “re-paste” twice. I have a weird “issue” with my paste where it would “bubble” away from those 4 really hot cores, leaving mostly a void.
I assume it might be due to the absorbed moisture. It is really old paste, and the tube of it did the same thing (expand) when I had it on top of the radiator. They rate it at 200C, so it shouldn’t be it decomposting.
But assuming this, I added on a lot extra during the first re-paste, so it will get all around the CPU, getting warm and hopefully letting out the moisture. The second time I just scrape those back onto the CPU. 44, 45 watts on Cinebench.
It is not ideal, but this is not exactly high-performance paste.
The temperature spread is a lot better. 100 on hottest, 85 on lowest. Although, this might indicate lower performance, the 85 core might have not had its heat wicked away to 80, and the 100 didnt bosot as high.
14512 points or some, with discord and edge open.
That doesn’t sound typical for Thermal Paste.
The Ryzen X 7X40HS has favored Cores, but thats mostly for Single Core Loads, when you use a Full Load, all Cores stay in the same Range to each other in Terms of Frequency. So yeah that big of a Difference is still unlucky.
My modded Setup is still in the High 15k Range, but it degraded slightly from the first Burn ins.
This week i will attempt to solder a new Shim to one of my Spare Heatsinks. I Also bought Liquid Metal. Tomorrow should my Reflow Station arrive, than i can start the modding attempt 
This is true. but the hotter cores draw significantly more power. I don’t know why, but that’s a thing.
Thats really wierd, mine did stay ±50mhz before RMA After RMA and even now after my Mod.
A little update to mine as I tweaked it more. I’m at 16148 points multi core on a warm boot with my 7940hs (just spent the last hour compiling a kernel and using the laptop for various things.) I’m on Ubuntu 24.10 with kernel 6.12. Tomorrow morning I will run a cold boot test but so far so good.
I’ve not made any changes to my heatsink or cooling, just software stuff.