Been following this thread and finally got my PTM in the mail today. Just did a swap this evening and booted up right after. Let idle for maybe ~10 mins and figured why not, lets see where we are at. This was literally right after the swap. Was around 47 cpu package power the majority of the 10 min test, but toward the end it was consistently around 49 and even stuck on 50 every few seconds. Differences between the cores are also MUCH better. Was seeing ~20 degrees difference between the coolest/hottest cores previously. Needless to say, fairly pleased with the result right after the swap. (Was in the 14,000s before swap). Power plan set to performance and battery actually locked in bios at 80%. HWiNFO probably ate some of the score but I know I was reading from some folks that performance should go up after ~10 cycles for that burn in period. Time will tell.
Edit: Forgot to mention that this is not in over clocked mode and I am running a -20 curve undervolt..
A -20 curve optimizer will actually increase clock speed (not directly, but one of the big factors in the cpu clock speed is how much voltage is being used - see https://youtu.be/c45KXBnbbpk?feature=shared)
I normally rock a -30 all-core (tested stable) and it adds a significant performance boost in synthetic workloads, and it makes my laptop boost to around 5.4/5.5 ghz.
One tool I’ve used to test stability for windows systems is core cycler (sp00n.) I haven’t found a Linux one to use, though.
When I first did an undervolt on mine I was able to get -30 as well. A few days after, stability tanked and I was getting reboots so I had to dial it back. Maybe I’ll mess with it again..
Just got back into town again and my laptop has been off for a few days.. Decided to do another test (2nd boot & test since repaste with ptm). Very pleased with the product Honeywell has created.
Edit: Snapped that literally after test finished and cpu was still coming down which is why current temps were still a bit high.. And again, this was without any modification to the heat sink like many others have done.
If they don’t, you can hunt one down for Ubuntu (or other distro), in the “.deb” package format, and install it manually.
You can certainly try to do that.
Psneosr look pretty cool. But like, does s-tui not suffice?
A overheating (and properly throttling) chip should not prevent the normal functionality of the computer.
Furthermore, PTM is non-conductive, so it should be ok.
None.
If you were on the old liquid metal and 1. didnt clean it up well or 2. Yanked on the heatsink without letting the old liquid metal melt to room temperature, you can break things that way.
I removed the LM day 2 of using the framework, so to make sure not to run into those problems.
I have a spare heatsink (modified or not), so maybe. But I do not think (bad thermal) is the issue.
Try reseating the GPU interposer, if anything. It’s somewhat finicky, sometimes I have to do it a few times to get both SSDs (in the dual m.2 carrier) to show up.
I didn’t find any relevant damage on the dgpu interposer pads (on the gpu, the mainboard, or the interposer itself). I did find a little blue film near the threads but not on a pad - probably loctite or something similar from the threads. Cleaning and reseating the interposer did not help unfortunately even with multiple attempts at reseating - everything lines up and goes in without issue.
I didn’t particularly find any debris or damage on the display cable either (male or female ends).
It seems if I issue a restart once I get a boot with no issues I will continue error free until an actual shutdown is issued. I’m not sure if this really implies some sort of damage to a capacitor or something similar, but I was pretty careful and slow about my process and haven’t observed any physical damage at this point. Unfortunately the crash dumps just seem to point to amdfendr.exe which seems to be AMD crash protection (ironic).
I appreciate all the suggestions so far. I may just need to limp along until upgrades are announced and start replacing parts at that point to become error free again.
I finally bit the bullet to contact support and get my replacement PTM.
So today I headed to my local hacker space to mainly get the mental support and know-how if I mess up.
After opening everything, I knew why my FW was constantly hot: Basically no LM left on the CPU.
The scariest part of the process was peeling off the heatsink, and the few long seconds when booting up the computer again.
So far my FW feels much cooler, which is really no surprise looking at how bad the LM looked.
what software have you been using to undervolt I noticed from the image your running windows and I just wasnt aware of any good ways to undervolt any framework 16 processors on windows at the moment
I would recommend rotating the cooling assembly a bit back and forth and not prying. The die can stick to the cooling assembly and you may risk issues with ball grid array underneath.
Lets see the first reviews when the ai 300 drops. No Vaporchamber and no shim on the new Heatsink. Can be good or bad. I need reviews asap and as the ryzen ai uses the same footprint as ryzen 7x4x does this heatsink fit on the old model? Questions over Questions
I’ll be a part of batch one to be able to report but technically it’s already in the 13 isn’t it? Are they facing any issues? I don’t read 13 forums haha
Done, additionally i took a look at the fans and thought, which Part could whistle, well i copied the mylar/Vinyl Strip from the Cover and coverred the small hole where the ribbon runs through.
don’t mind the electrical Tape, thats an “old” Try of mine to reduce turbulence and even the pressure loss.
Overall subjectively, Yes that helps to reduce the whistle and whine on the gen 1 dGPU Bay. When the new Fan comes to avail, i definitly think about getting a pair and compare it.
and i know its not uneven cpu thermals related, but as my whole heatsink mods are here i wanted to keep em here And yes the dGPU has also been swapped to PTM long long ago.
It changes the pitch a little. The high pitch gets reduced, but is still there. Its just a little reduced. The overall noise is the same, but different. Can’t fully describe it. Would be nice if someone else tries it to confirm my test.
I managed to get quite good results following your methods. I taped over the wire hole on both fans, then taped over the first four inner slits on the bottom fan grates, and finally taped over the inner two exhaust holes on each side of the dGPU and the whine is nearly gone. It’s much more tolerable to play a game or perform some other CPU intensive task now. That whine grated on my ears.
EDIT: Forgot to mention that I also applied tape where you had kapton tape sealing the fans to the dGPU radiator fins.
Another win for you! Your experimentation and guides have really helped me get this laptop closer and closer to what it should have been from the factory. Thank you.
Count me in as a happier customer after the PTM replacement. Like others, all of my liquid metal leaked out and there and what was left just covered 1/4ths of the CPU:
I can highly recommend this procedure to anyone else suffering from loud fans on minimal workloads. It also took months for the PTM replacement to arrive - press FW support for action if you’re waiting for more than a week.
and as the ryzen ai uses the same footprint as ryzen 7x4x does this heatsink fit on the old model? Questions over Questions 
but technically it’s already in the 13 isn’t it? Are they facing any issues? I don’t read 13 forums haha
