The fans come on with very little CPU load (10-20%) and are quite loud for a laptop: https://photos.app.goo.gl/1MBbfuiSQwmYvouH6
I’ve been dealing with FW support for several months about this, and even RMA’d the GPU unit - no luck.
Is this normal for Framework 16?
Yes, the dGPU fans can be quite loud. The expansion bay shell fans however are practically silent under even relatively high loads.
Take a look through this thread: Uneven CPU thermals!
It may be possible that if your CPU is using Liquid Metal compound it has pumped out and isn’t covering the cpu die correctly anymore.
If you run an all core workload such as cinebench or s-tui/stress what is the difference between the hottest for and the coolest core?
I’m not sure what’s the simplest way to show temperatures across all cores - s-tui didn’t seem to show that stat.
Here is what sensors says though, and if the ACPI temps at the bottom are the individual core temps, then they are indeed way out of line with each other:
$ sensors
amdgpu-pci-0300
Adapter: PCI adapter
vddgfx: 15.00 mV
fan1: 0 RPM (min = 0 RPM, max = 4900 RPM)
edge: +33.0°C (crit = +100.0°C, hyst = -273.1°C)
(emerg = +105.0°C)
junction: +33.0°C (crit = +100.0°C, hyst = -273.1°C)
(emerg = +105.0°C)
mem: +42.0°C (crit = +105.0°C, hyst = -273.1°C)
(emerg = +110.0°C)
PPT: 1000.00 mW (cap = 100.00 W)
spd5118-i2c-2-51
Adapter: SMBus PIIX4 adapter port 0 at 0b00
temp1: +50.5°C (low = +0.0°C, high = +55.0°C)
(crit low = +0.0°C, crit = +85.0°C)
ucsi_source_psy_USBC000:004-isa-0000
Adapter: ISA adapter
in0: 0.00 V (min = +0.00 V, max = +0.00 V)
curr1: 680.00 mA (max = +0.00 A)
k10temp-pci-00c3
Adapter: PCI adapter
Tctl: +89.6°C
cros_ec-isa-0000
Adapter: ISA adapter
fan1: 3389 RPM
fan2: 3223 RPM
ambient_f75303@4d: +53.9°C
charger_f75303@4d: +55.9°C
apu_f75303@4d: +55.9°C
cpu@4c: +90.8°C
gpu_amb_f75303@4d: +35.9°C
gpu_vr_f75303@4d: +37.9°C
gpu_vram_f75303@4d: +35.9°C
gpu_amdr23m@40: +32.9°C
nvme-pci-0400
Adapter: PCI adapter
Composite: +53.9°C (low = -40.1°C, high = +83.8°C)
(crit = +87.8°C)
Sensor 1: +70.8°C (low = -273.1°C, high = +65261.8°C)
Sensor 2: +53.9°C (low = -273.1°C, high = +65261.8°C)
BAT1-acpi-0
Adapter: ACPI interface
in0: 17.50 V
curr1: 1.23 A
amdgpu-pci-c500
Adapter: PCI adapter
vddgfx: 1.26 V
vddnb: 761.00 mV
edge: +56.0°C
PPT: 26.05 W (avg = 21.10 W)
ucsi_source_psy_USBC000:002-isa-0000
Adapter: ISA adapter
in0: 0.00 V (min = +0.00 V, max = +0.00 V)
curr1: 0.00 A (max = +0.00 A)
mt7921_phy0-pci-0500
Adapter: PCI adapter
temp1: +62.0°C
spd5118-i2c-2-50
Adapter: SMBus PIIX4 adapter port 0 at 0b00
temp1: +54.0°C (low = +0.0°C, high = +55.0°C)
(crit low = +0.0°C, crit = +85.0°C)
ucsi_source_psy_USBC000:003-isa-0000
Adapter: ISA adapter
in0: 0.00 V (min = +0.00 V, max = +0.00 V)
curr1: 0.00 A (max = +0.00 A)
ucsi_source_psy_USBC000:001-isa-0000
Adapter: ISA adapter
in0: 5.00 V (min = +5.00 V, max = +5.00 V)
curr1: 5.00 A (max = +3.00 A)
nvme-pci-0600
Adapter: PCI adapter
Composite: +48.9°C (low = -5.2°C, high = +89.8°C)
(crit = +93.8°C)
acpitz-acpi-0
Adapter: ACPI interface
temp1: +53.8°C
temp2: +55.8°C
temp3: +55.8°C
temp4: +90.8°C
temp5: +35.8°C
temp6: +37.8°C
temp7: +35.8°C
temp8: +32.8°C
Here are the fan temps when I put my minimal load on: Imgur: The magic of the Internet
Could you confirm?
Try blocking four inner holes of the dGPU exhaust and see if it improves the quality of the noise. It helped me and some others make the noise more bearable.
It does look like it could be related to the Liquid Metal issue. Temperature 4 could be related to core 4, which tends to be the one with the most significant issue.
The one where it’s at 90+C is with an all core workload? Is it with the stock charger and performance mode? What is the clock speed of the cores while it’s doing that all core workload?
This is what mine looked like when I first started looking into the issue: Uneven CPU thermals! - #269 by Obasav
I was having similar symptoms, with the fans going nonstop at a minimal load, and significantly degraded multi-core performance (using cinebench as an demonstration)
The one where it is at 90+C is a multithreaded workload where none of the cores go above 10-20% utilisation. Using stock charger, not performance mode. This is what makes this issue quite annoying.
How did you end up resolving this?
I did a replacement of the Liquid Metal compund.
I ordered some thermal grizzly phase sheet, and some 20mm x 20mm copper shims, about .8mm thick. Using sandpaper on the shim, I smoothed it out to a fine polish and cleaned the copper dust off of it.
I removed the battery from the laptop, as well as used an external mouse/keyboard and had the input deck and keyboard removed while I started the system connected to the AC.
I made sure to loosen the cpu heatsink screws, and using a towel I pressed down on it to give it mounting pressure while the system was on.
I ran the multi core workload for a few minutes, pulled the power cable from the system, and then lifted the heat sink off of the cpu just slightly to break the bond that the Liquid Metal compound would have when it cools.
After holding the cpu cooler up for a few minutes to let it cool and harden (I held for about five minutes), I removed it from the system and cleaned the compound from both. Tape works really well for picking up the metal flakes, you’ll want to be very careful and make sure all of them are picked up off the board.
Then I put a layer of phase sheet, one of the shims, and another layer of phase sheet, then reattached the cooler and put it all back together. Then I ran a benchmark for a bit to get everything warm.
After a few runs and letting the laptop cool, it is now completely silent regarding the fans while idle, unless cooling down from a previous workload.