I was looking at the heatsink (Framework | Fix Consumer Electronics) and I can’t remember if it was the same photo before or if Framework added a new model 
It doesn’t look like the same one I bought,
2 Likes
GPU is purring. General temp at 64 and climbing, hotspot at 78. CPU now pushes an unbelievable 58 watts at cold start, dropping to 54, and temps don’t want to climb over 90 or so. Is this normal? Core utility is sticking around 115%. Just wanna make sure nothing weird happened.
Beautiful results. Your PTM-shim-PTM sandwich is lovely.
edit: Was able to push 100 using UXTU at a blinding SIXTY WATTS?!
2 Likes
Welcome to an ascended Framework 16 <3 
And Welcome to the tinkering, it’s like a Drug, you will find more Things you would like to modify 
Yep Temps sound good and yes you shouldn’t hit 100C anymore when you leave the PPT Stock and untouched.
3 Likes
AMD’s software seems to have realized it too, it’s giving me a peak TDP of 145 for the entire system - 100 for the dGPU and 45 for the APU.
I think I’m gonna avoid pushing any higher than 54W and just try to get the system to steadily hang around that. Notable is that I wasn’t able to get all the solder off of the heatsink, but it doesn’t seem to matter. I don’t have the cash to really recover if something tinkering-related destroys this machine, so I’m not gonna be trying any LM or other crazy modifications. For now, shim and PTM is more than enough for me, especially with having it on both main chips.
2 Likes
Perfect, did a good Job.
Keep it where you feel safe and mostly cool.
I did all my Tests to push and find the upper Limit, but in daily use i stay in balanced Power Profile or Limit the Thermal Limit to 70C to keep the Device relativly quiet.
1 Like
Weirdly low Cinebench scores that are probably just due to all the bloatware hanging around. Might try hunting it down first and then benching, but for now I’m happy with this.
Nope its the same Pictures as in August/September.
I suspect its the preproduction Heatpipe they had before Coolermaster “refined” their Shim Bonding 
@Redeye_43 Probably, i have 500 Points Difference alone when i close HWInfo. Do a Restart of the System and close every Tray App Thats helps with a Benchmark.
1 Like
I ran a couple benches without HWinfo after restarting and nuking tray apps. It’s a bit unstable, hovering around 16k, but fine to my satisfaction.
I can’t believe I lost 1500 points to a monitoring program. That’s ridiculous.
Either way, very happy with the result. This thing is gonna boil me alive if I overuse it in uncooled rooms, though, wow. Another addition to the list of positive experiences involving the PTM sandwich technique.
Sounds about right. Do you get near 16k with unlocked Limits?
HWInfo polls so many sensors and devices that it does have a pretty significant impact on the system.
The increase in chassis heat is actually the main thing stopping me from trying a sandwich mod lol.
Getting the full power worth out of the CPU would be great, but I actually don’t use full CPU power in my daily use. There’s a few games and activities that would benefit, but being able to use the machine wherever is also nice.
Maybe I’ll just pickup a spare heat sink I could swap back to if I ever felt like it…
The System runs cooler at the same Wattage, after the Mod. Yes it can dissipate more Heat into the Case, but at the same Performance you have now its cooler. (And quieter)
In this case(pun unintended) the CPU will run at lower power earlier as the STAPM cuts power sooner due to faster skin temperature rises thanks to better thermal conductivity.
1 Like
My STAPM has never risen into any Limit. 
Don’t know if stapm is really implemented.
Do the sandwich, and you can use a utility like UXTU to crank down the TDP if you want. Chassis heat increase is also generally negligible in my experience unless the system’s outright on your bare skin.
1 Like
Hey hey, don’t judge what I do on my bare skin…
2 Likes
I haven’t had time to read through the entire thread yet, but I wanted to add my benchmark scores & temps from my bone stock Framework 16 with Ryzen 9 7940HS under Windows 11 Pro just for reference. It looks like I have a fairly large gradient in temperatures across my cores with one of them hitting 100C at full load but the CPU power draw stays above 39 watts, so I don’t suspect I’m thermally throttling too much if any. I hope my information below serves as another data point for this thread and I look forward to reading the whole thread when I have more time and it’s not Thankgiving! BTW, Happy Thanksgiving! 
Cinebench R23:
Multicore: 13391
CPU-Z Benchmark v17.01.54
Multicore: 6351.1
Singlecore: 681.7
CPU-Z Benchmark v19.01.64 AVX-512 (Beta)
Multicore: 7090.4
Singlecore: 881.7
When I run a “stress test (v17.01.54)” under CPU-Z Benchmark multicore it will drop down slowly starting at around ~6600 dropping down rapidly to ~6300 and then more slowly dropping over several minutes until it finally levels off at exactly 6175.1 and it will stay there for however long I run the benchmark in a room with ambient temp of 73F. Every core stays between 4.166ghz & 4.215ghz throughout the stress test.
Max reported power & temps via HWMonitor once stress test (v17.01.54) stabilized
Core 0 : 79.8C
Core 1 : 88.6C
Core 2 : 91.3C
Core 3 : 98.4C
Core 4 : 96.0C
Core 5 : 100.1C (Drops to 100C and stays there for entire stress test)
Core 6 : 93.9C
Core 7 : 98.5C
Package: ~39.2W
Framework 16 Max Temps reported by HWMonitor during stress test (v17.01.54)
TZ00 : 50.8C
TZ01 : 53.8C
TZ02 : 60.8C
TZ03 : 99.8C
5 Likes
Thanks for the Reference. Looks as bad as expected. I habe read you want to make a Video Review on the Device. In all Fairness you could mention the Mods we did or try them youself.
Framework will provide PTM7950 instead of the Liquid Metal in new Batches and they will provide some for existing Users to switch from the Liquid Metal themselves. The LM is the Mainculprit for the degradation. The uneven Thermals are due to the Heatsink Manufacturer (Coolermaster) not fully bonding the Matingsurface of the Heatsink to the Vaporchamber. We as Users proved it can be “repaired” by putting a Coppershim there instead either by soldering or by a “PTM Sandwich”
3 Likes
I have the same Ryzen 9 Barnacules, you’re on the low end CB score wise. Since you haven’t read the whole thread (no surprise at the size it is right now) do note that you have it in performance power profile and you are using a power delivery source that is or compares to the FW power supply.
Otherwise your scores not comparable due to power constrains. (if it is, maybe you can edit post and add that for reference?)
With the modifications to the heatsink, lapping and replacing the shim with copper shim and PTM 7950 sandwich it is consistent in the 16400+ range.
After the mod all my cores are in the same temp range all the time with only a few degrees difference…
2 Likes
I just tried remove the LM and repasted with ptm. the removal of LM takes an hour. removing all LM with tweezers and pasted. now it is working with sustain 52w!
9 Likes
