You should read some of the posts on the forum about that
1 Like
It’s an intriguing concept but practical application yet to be validated.
If Thermaltake designed a better Heatsink, Noise wouldn’t be a problem in the fw16.
There is enough space for bigger heatpipes and wider fins. There was left alot of potential. We will see when a next gen arrives, they have that sorted out. (Hopefully)
The ion airstream system is an interesting concept, but not in the current devices and far to early in Dev for proper application.
Correction: Coolermaster not Thermaltake
3 Likes
potentially copper heatsink fins would be the next step up in this regard imo, a simple change to make that will increase performance without having to retool everything.
Yes that could work there is enough available space in the FW16 to accommodate that in my opinion
Copper heatsink fins would not require any extra space compared to the aluminum ones currently in use.
2 Likes
Update: I received an order confirmation for my official replacement thermal pad today. This will likely be my last update on this issue as the laptop is already using MODDIY PTM as of 2 weeks ago and no longer needs the official PTM replacement pad.
Just received my order confirmation as well…6+ months after requesting it and after my warranty has expired. That is quite insane but I guess it was finally done.
2 Likes
wasnt it coolermaster.
Also, heat really doesnt seem to be the problem. Even at 45W (stock factory trim), the heatpipes… they aren’t hot, they are warm.
Do you have a TL;DR version? I don’t understand the PDF as it seems to be writing the same thing g again and again. Is it some sort of passive cooling, magnetohydrodynamic drive or bladeless fan?
Yeah you are Right it was Coolermaster.
And That the Heatpipes are not getting Hot is the Main Problem. They have to heat up in Seconds. If not the Thermal Resistance before the Heatpipe is to high. They have to Transfer the Heat away.
2 Likes
Hello Framework 16 owners!
I wanted to share my experience with my uneven CPU temps and the framework support team.
Since I received my F16 (with graphics module), I’ve been very annoyed by the poor thermal management of the machine. Full blasting the fans whenever I would put windows 11 in anything else than “power saving” mode. I was certain it couldn’t be normal and had to be a hardware issue.
After finding this thread I immediately did some proper test and found out that one of the core of my CPU was jumping to 100 degrees whenever I would use the cpu more then 25%. And that’s enough for the motherboard to push the fans to the max.
I contacted support and followed the (long) procedure to justify a test&repair shiping ticket. It tonk almost 3 weeks to come back (they are overwhelmed in the Germany repair center apparently) but it’s now returned all fixed. You probably guessed it but they simply changed the faulty liquid metal with thermal paste.
I encourage most of you to do the thermal managment test (the one with Cinebench and HWinfo) and contact Framework support if needed. It really is night and day difference when it’s fixed. Sure it takes time and energy to do all the back and forth with the support team but in my experience it’s just procedure and they are not trying to minimize the issue.
Thanks
7 Likes
Just got my PTM7958 from Framework, followed the guide for replacement. My Liquid Metal had mostly run to one corner with a super thin layer with what looked like scorch marks on the CPU. Was meticulous about getting the CPU die and the heatsink clean. While I didn’t actually look at temps per core in HWInfo before, I do notice the fans ramp up more gradually. Running Cinebench R23 on my first run i got around 15500 (didn’t take a pic). Second run I had HWInfo running and got lower. But I noticed there are still a couple of cores with a pretty high temp delta. Something I should contact support about? Even though I’m prob out of warranty period?
The ghostly looking mark on the CPU die is before I did the final cleanup, it’s gone now.
4 Likes
Nicely done, much shiny!!!
How much power is it drawing? HWiNFO shows my CPU Package Power peaks near 60W when starting a benchmark and steady-state at 45W (at least, it’s steady-state at 45W for my 7940HS. I assume that’s the same spec for your 7840HS?).
If you aren’t getting 45W then I’d definitely contact support. If you are getting 45W and you’re getting roughly expected performance for a 7840HS, I don’t see much point in contacting support.
Anyone able to show the cpu frequency when doing a stress-test?
When I launch a stress -c 16, and doing so - my frequency stays around 3370Mhz for all cores.
Command I use is:
watch -n.5 "grep '^cpu MHz' /proc/cpuinfo "
Sadly sensors does not show me the core temp per se.
Just want to evaluate if I replace the metal-liquit with the replacement pad.
Try “amdgpu_top”.
It is on github.
Are you sure you’re in max power mode? Idk what it’s called in Linux. Nothing you’re showing so far indicates any thermal issues to me. But that’s because you aren’t looking at the right metrics.
If you aren’t getting 100C on any cores and/or aren’t drawing at least 45W in steady state, then I believe that means you’re in an efficiency or power saving mode. Could be due to using a 3rd-party charger too, at least for my system on Windows I won’t get max performance even in Windows Performance mode unless I’m specifically plugged in with the Framework charger and power cord. Third party ones should work with proper ratings, but you need proper ratings on not just the charger but also the cable.
I’m pretty sure, yes.
Using balanced mode provides power when required (my day to day use case when on power).
System is connected to a Dell WD19TB dock, with 180W power supply delivering 90W to non Dell systems. So that should be plenty enough.
I can try this WE (long WE) to plug in the 180 provided Framework power supply and putting it all in performance mode. But I remember doing that already with the same results some months ago.