Couple of points. FW seems to be using PTM rather than liquid metal in the traditional sense. From the linked deep dive:
“Liquid metal is exactly what it sounds like: a 100% metal sheet made up of indium, tin, and bismuth that turns from solid into liquid at around 58°C as the CPU heats up”
The “etched” patterns previously mentioned above are the support columns / wick columns for the vapor chamber internals, and likely having nothing to “holding the liquid metal in” as someone speculated.
The etched Pattern has no Contact to the vapor chamber internals. I Lapped them down to bare Copper. And it is exactly what we suspected as they mentioned it. Liquid Metal Deep Dive “Next, an etched pattern in the surface of the vapor chamber holds the liquid metal through tension.”
The Columns and Wicks are inside the vapor chamber and heatsinks. it would be fatal if they are exposed, as those Components are as Vacuum as possible to have the Liquid inside the vapor chamber and Heatpipes beeing “vaporised” at the lowest possible Temperature to evict the Heat from where its generated.
And they are using liquid metal, just differently formulated. Yes it is a “Phase Change” Type, but still Conductive. I will bet it outperforms the PTM7950, but it is not enough Material to fill the etched Gaps.
I see where they say the etched pattern helps to hold in their liquid metal. I missed that, and time will tell if this was a better solution than a compressible mask or similar around the die. Vapor chambers often do have the imprint of the support structures visible which is what I was referring to - yes, it would be fatal if they were exposed.
I view PCM and LM as different due to differences in surface mating, longevity, reactivity with copper, etc. PCM is more forgiving. Kyrosheet might be a good alternative, but obviously if one is lapping in an attempt to get stock performance/temps rather than to maximize performance/temps something is definitely wrong with the implementation of the cooling solution.
Yep, after all my effort im still just in the Lower End of the Median 7840hs. This is well… Frustrating if i consider i paid 1000€ more to get less Performance than the Asus TUF 16 Advantage Edition and its still a lot Louder.
I am struggling with this purchase. I really appreciate the Philosphy and Ideas from Framework. And the 13 was a Golden Nugget. I really hope they release a statement or a DIY Fix, to all Users. I would even pay for it, because i like the Rest of the Laptop.
What did you use to test with under Linux? From what I gathered, Cinebench is not available under Linux.
Or did you run it through a translation layer like Wine or Proton?
I just did a Geekbench 6 run under NixOS and didn’t have any issues. The fans didn’t even turn on during the entirety of the benchmark and I didn’t feel any noticeable heat coming from the laptop itself. The highest temp I saw using btop was a spike of 70C, but after around a second or so that was back down to 50C .
My score seems to be quite a bit above the average consensus for the chip.
Here are my results:
CPU: Framework Laptop 16 (AMD Ryzen 7040 Series) - Geekbench
GPU (igpu) : Framework Laptop 16 (AMD Ryzen 7040 Series) - Geekbench
Here is the average score for the Ryzen 9 7940HS
https://browser.geekbench.com/processors/amd-ryzen-9-7940hs
Maybe I am just lucky and wasn’t affected by the issue? I am batch 2.
In The Reviews many Reviewers mentioned, that Framework is redoing the Cooling System with a “cheaper to manufacture” Version. Maybe Batch 2 has a Cooling System, which is before that change.
If you have steam, just add the cinebench exe as a non steam game, launch, profit.
That would really suck. Hopefully we can get some more confirmation on this and if it’s true.
anyone else getting an average decibel reading of 90+ with an average of 95 degrees while running any sort of program/games that is a tad bit load-intensive?
I really wish I could just return this. It is definitely not worth at this moment in time…
Mine makes significant noise under those conditions from the fans going full speed (like a hairdryer on low or medium) but nowhere near 90 DB. I’d contact support.
Thanks,
I did the registry update in windows 11, so the context menu show me the full list.
It solved the delay issue. May be some logic is trying to run.