Ryzen 5 7640U FW13 randomly reboots under GPU load

  • Which OS (Operating System)?

Linux

  • Which release of your OS?

Fedora 42 KDE

  • Which Framework product and which generation?

Framework 13, AMD 7040 (Ryzen 5 7640U)

Whenever I do anything involving sustained GPU load, the laptop reboots within 30 minutes. Any significant gaming, any sustained generative AI tasks, and any photo/video editing consistently consistently cause the laptop to reboot.

The reboot behavior is identical to if I had selected “Restart” from the KDE applications menu. The Framework logo flashes briefly on the screen before the power light blinks off, then turns back on as the laptop reboots. There is no LED storm from the mainboard.

“Normal” usage (web browsing, watching videos, text editing) and sustained high CPU-only activities (running LLM tools compiled without AMD GPU support, forcing software rendering on 2D games and avoiding the GPU) don’t seem to trigger this.

Temperatures and fans do not appear to be an issue. Fans are running full blast when the reboot occurs. I logged sensors output once per second prior to the most recent reboots and the output indicated nothing over 91C, with amdgpu reporting 72C:

Wed May 14 10:09:33 PM PDT 2025
amdgpu-pci-c100
Adapter: PCI adapter
vddgfx:        1.20 V  
vddnb:       764.00 mV 
edge:         +72.0°C  
PPT:          36.23 W  (avg =  30.07 W)

cros_ec-isa-0000
Adapter: ISA adapter
fan1:            6874 RPM
local_f75303@4d:  +60.9°C  
cpu_f75303@4d:    +55.9°C  
ddr_f75303@4d:    +50.9°C  
cpu@4c:           +90.8°C  

ucsi_source_psy_USBC000:003-isa-0000
Adapter: ISA adapter
in0:           5.00 V  (min =  +5.00 V, max =  +5.00 V)
curr1:         0.00 A  (max =  +1.50 A)

spd5118-i2c-20-50
Adapter: SMBus PIIX4 adapter port 0 at 0b00
temp1:        +62.5°C  (low  =  +0.0°C, high = +55.0°C)  ALARM (HIGH)
                       (crit low =  +0.0°C, crit = +85.0°C)

ucsi_source_psy_USBC000:001-isa-0000
Adapter: ISA adapter
in0:          20.00 V  (min =  +5.00 V, max = +13.20 V)
curr1:         3.00 A  (max =  +3.16 A)

BAT1-acpi-0
Adapter: ACPI interface
in0:          17.41 V  
curr1:       548.00 mA 

k10temp-pci-00c3
Adapter: PCI adapter
Tctl:         +91.8°C  

mt7921_phy0-pci-0100
Adapter: PCI adapter
temp1:        +54.0°C  

ucsi_source_psy_USBC000:004-isa-0000
Adapter: ISA adapter
in0:           0.00 V  (min =  +0.00 V, max =  +0.00 V)
curr1:         0.00 A  (max =  +0.00 A)

spd5118-i2c-20-51
Adapter: SMBus PIIX4 adapter port 0 at 0b00
temp1:        +57.2°C  (low  =  +0.0°C, high = +55.0°C)  ALARM (HIGH)
                       (crit low =  +0.0°C, crit = +85.0°C)

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)

nvme-pci-0200
Adapter: PCI adapter
Composite:    +47.9°C  (low  =  -5.2°C, high = +89.8°C)
                       (crit = +93.8°C)

acpitz-acpi-0
Adapter: ACPI interface
temp1:        +60.8°C  
temp2:        +55.8°C  
temp3:        +50.8°C  
temp4:        +90.8°C  

Battery is fully charged and the laptop connected to the stock power adapter when the reboots occur. Journalctl not only fails to log any errors prior to the reboot, it doesn’t log any SIGTERM or reasoning for the reboot.

No system updates are available when the reboot occurs; it’s not an update triggering an autoreboot in the background.

Sounds like a hardware issue, like a bad CPU pad that disconnnects once heated up enough.

If this happens since you bought it, could be a manufacturing defect or shipping damage. If this is new, it could be damage from a fall.

The behaviour is not the same as when closing the lid right?

Might be RAM temperature. I have noticed on my Framework 16, even. The RAM closest to the CPU can go to excess of 90 degrees.

ddr_f75303@4d is fine, but I don’t .. quite believe that?

It is not the same as a lid closure (and I’ve had problems with spurious lid closure reports with this FW13 in the past also). It is a hard reboot, the same as selecting “Restart” from the menu.

Either way, if that’s the case would this a mainboard replacement-level issue, or something I could feasibly fix? I’m assuming the use of “disconnect” here means that you’re referring to the CPU’s connection to the mainboard, and not thermal pad coverage.

Stray magnets have caused issues in the past by triggering lid close but if it doesn’t behave like that that’s probably not the cause.

If that were the case, that would likely be a mainboard replacement. A microsoldering repair company could theoretically diagnose and fix the mainboard (by resoldering the CPU), but as the Framework mainboards aren’t super expensive that’s probably not financially effective, like the repair probably costs around as much as a new mainboard and has a chance to fail.

1 Like

Have you ran Memtest86 (by passmark) yet?
If Memtest86 (by passmark) doesnt give you any error, chances are that the problem is with something else.

Problems related to RAM over-temperature can also be kind of hard to nail down.

@Xavier_Jiang No errors after one pass of MemTest86+. Temps maxed out at 100C during the test pass and stayed around 89-91C. When I have more time I’ll run it for a solid 20 minutes, which is about the average time to failure for sustained load.

I saw your post over on the similar FW13 AMD random reboots - #17 by Xavier_Jiang and I’ll investigate more cooling options, but I’m inclined to just move off the AMD line at this point.

@Adrian_Joachim Yeah, I’ve already had that problem in the past and disabled all lid closure behaviors as a result, and I also still log all lid closure events to their own file for posterity since journalctl doesn’t seem to log all of them. The last reboot occurred when the laptop was open and situated with the back panel fully unobstructed, with no magnets nearby and no pressure on the lid closure point that might cause a short that would trigger a lid closure event. No lid closure events were logged around the reboot.

I want the memory temperature. You can find that in HWInfo, I dont think basic hardware polling like Memtest will tell you that.

And, I do know, that my DIMM0 temp (in the 16) had got to beyond 90C at some point. This is worth investigation. Especially since GPU hit them harder than CPUs do. Specifically if you are crashing under GPU load.

AMD does have worse platform, but calling my laptop unuseable would be a gross understatement.

What RAM module are you using, by chance?

I’m afraid I’ve resolved the issue by swapping my former 12th-gen Intel mainboard back into the laptop chassis from its standalone case.

In the standalone case, the AMD board (with no changes in RAM or SSD) no longer exhibits shutdowns under sustained GPU load. In the laptop case, the 12th-gen Intel board no longer triggers spurious lid closure events and runs about 5C cooler under sustained load.

RAM modules on the AMD board are two 16GB A-DATA sticks, model AD5S560016G-B.

1 Like

What would you shift to? Intel core?

Seems like it has very few or these constant whack a mole firmware or other bugs but also rarely discussed here at forums.

Main issue is the core ultras (at least gen 1) are still super hot and cause crazy fan noise/throttling making them perform worse than 7x40u counterparts.

Are you under 1 year warranty on AMD board?

I am not. It’s a batch 1 board that shipped in October 2023.

Just received a new FW13 Core Ultra 155U DIY kit, put it together, and pulled the RAM and SSD from the 7640U mainboard and installed it on the Core Ultra.

I’ve been playing the same GPU-intensive games on it for the last 2 hours without issue. Temps reported by sensors have peaked at 71C. Fan hasn’t passed 4000 RPM.

I have new RAM and storage coming to put into the old Ryzen chassis to re-test.

Those temps are quite nice. Sounds like your fan is spooling much less than your potentially defective AMD board. Would be interested to see how a replacement does vs the Intel.