During the installation of the PTM thermal pad, I accidentally bent the heatpipe of the heatsink assembly a tiny bit. Since then I am noticing that a few sensor temperatures almost immediately show around 95c. I would like to confirm if this is normal or is caused by my heatpipe issue, before I order a new heatsink and replace it. But since I hadn’t run the test before applying the thermal pad, I don’t know what were my temperatures beforehand.
To help with this, I will be very glad if some community member who has Framework 16 with 7840HS can run s-tui on any Linux distro, when using performance mode (no battery saving options) and share the screenshot after a minute or so of s-tui test running.
That’s FW16? If yes, then I think somehow that is still in balanced or battery saving mode, because the CPU cores are stuck at around 3.5Ghz when they should boost to at least 4.2Ghz
Performance mode typically shouldn’t trigger CPU boost mode by itself. Performance mode primarily adjusts power management policies, thermal limits, and scheduling priorities, but boost mode activation depends on actual workload demand and thermal conditions.
Mine, when using default, is not stessing my laptop at all but if I can mimic your test, Ill share.
EDIT: NVM it was the stupid “locked at 900hz” bug. I had to do a quick reboot.
Here are my results after a minute. I have the TPM swap done, and using Cinebench R23, I stay at 95 degrees - here I did hit 100, so finally got to see some thermal throttling:
Thank you for the information about using amdgpu_top
I ran that and it shows Average power as 48W. Were you stressing GPU as well during the test when you hit 51W? I don’t think s-tui stresses GPU. At least not for me.
Edited: I am not sure if the average power is for the whole laptop or only the CPU. If it’s for the laptop itself, then my low screen brightness of 4% could be the reason for lower power usage.
It happens quite often for me. Like once every couple of days if I am moving my laptop around and plugging into various chargers (Framework 180, generic 100w, 65w) around the house. Have not been able to pin down what causes it or figure out any way to resolve it without rebooting.
When it happens, amdgpu_top will show Throttling PROCHOT_CPU, PROCHOT_GPU, and some EC_“something” I can’t remember right now. That EC throttling seems to be the key.
@jared_kidd , this seems consistent with a lot of the threads I have seen @James3 comment on with his testing.
EC, PD, and throttling issues have been kinda scattered in the forum lately. I wonder if we don’t look at creating a new post to sticky to the top with some “known issues”.
As well, @James3 I believe has a community fork of the EC that addresses a lot of these issues. But I lost the thread with the installation steps, and haven’t had a chance to hunt it down since I last saw it