Fedora 44, kernel 7.1.8, BIOS 4.05 on AMD Ryzen™ 7 7840HS with Radeon RX 7700S
Hi, the bottom of my Framework 16 sometimes get surprisingly hot during relatively simple workloads (and, like others, I sometimes hear extremely loud fan noise when gaming), so I would like to learn how to objectively determine whether my hardware is working the way it should.
I did install the PTM product on my CPU some months ago but don’t know whether I did a good job.
I know how to run sensors but have no idea what items I should pay attention to or even which item is which. When I run sensors-detect and provide the default answers, it says “no” after pretty much every probe, and at the end it says
Sorry, no sensors were detected.
This is relatively common on laptops, where thermal management is
handled by ACPI rather than the OS.
which makes me wonder how useful the output of sensors actually is.
I also have run btop, htop, nvtopCockpit, and amdgpu_topbut am honestly feeling like I have too much information and not enough knowledge about how to interpret it.
I’ve also heard about ectool but don’t know if that would be helpful or just add to the confusion.
Thanks in advance for any input on how to understand my FW16 hardware better! Any input would be appreciated!
Historically, if you got a cinebench r23 score of ~13k then you were experiencing issues. However, that’s crude and only works on Windows. Those low scores were a result of the CPU bottlenecking at ~35W. What I’d do is run a stress -c 16 load to load the cores to 100% and then monitor both your temps and CPU power draw (I haven’t found a good tool for monitoring CPU power draw yet, but I know it’s possible because it’s reported in mangohud). If your CPU caps out at 35W and temps are at 100c, then you’re having issues. If your CPU is 45W+ and temps are less than 100c, you’re probably fine.
Forgot to mention, my system has an overhauled custom cooling solution, so you’ll have to ask what other people achieve regarding temps and power draw. I can break past 100W on the CPU in bursts and sustain somewhere around 80W+ at about 80c. Even with my setup and when I’m not loading my system heavy, the bottom of the device still gets hot to the touch. That’s just part of how the device is. I wish it wasn’t, but oh well.
Note: there’s no real performance gain from getting power draw that high.
notebookcheck has results of many benchmarks which you can try to reproduce for sanity checking. prime95 is quite easy to do on linux. but you need to sense “watts”. if power is 35w fixed and temp = 100 , my guess is it hits thermal limit and stops applying more power, i would take a look at the thermal interface, could be a dry out problem . before you remove for checking, make sure you are ready to re-apply replacement paste, because dried TIM once heatsink comes off, it will flake and peel = gone, do not put back and re-use (you need to apply new paste). if in doubt ask more questions.
I have already raised this issue and reported it to support. The area on the left side—where the motherboard’s voltage regulators are located—heats up first. After a while, the entire bottom panel gets hot because the air cannot escape. In a worst-case scenario, temperatures can exceed 60°C, which simply cannot be right.
Well I’m unsure what to make of this initial test: btop shows the CPU temp is 48C while idle, and the integrated GPU power draw at ~6W (this is the closest proxy I can find for CPU power draw at moment). Running stress -c 16 spikes the CPU temp right to 100C, and the integrated GPU is drawing ~49W. Both scenarios are with battery charging.
I’m guessing this is probably in the ballpark of acceptable performance?
Edit: unsurprisingly, the fans are annoying loud during stress test
. before you remove for checking, make sure you are ready to re-apply replacement paste, because dried TIM once heatsink comes off, it will flake and peel = gone, do not put back and re-use (you need to apply new paste). if in doubt ask more questions.
