Random reboots at idle, driving me nuts!

Issues & Assistance
1

Since a couple months I have random reboots on my system, can be twice in 3 hours or not at all after 12hrs (so not even every day), everytime while idle, afk or doing light office/browsing stuff

Usually I got 0 logs in journalctl, but lately I have the lines below “GPU lost from bus” (falled from the bus lol ?)
Updated BIOS
Removed all OC/tweaks
Reseated GPU and power connections
Removed OpenRGB as it crashed the system when first launched without -noautodetect

Tried LTS kernel
No luck
ChatGPT keeps telling me it’s a low power distribution fault likely from the PSU, I’m quite dubious but have no clue…

Is some genius here able to help ?

Specs : AMD R9 7950X3D CPU/ Asus ROG STRIX X670E-E board/ 2x32GB G-Skill Trident Z Neo 6000CL30 RAM / ASUS TUF Gaming AMD Radeon RX 7900 XTX OC Edition GPU/ Phanteks P600S case / Arctic Liquid Freezer III 360 ARGB cooler/ 2TB WD SN850 NVme + 2TB Crucial T500 NVme + 4TB Toshiba X300 HDD / Corsair RM850x PSU/

Log :

mai 18 23:05:46 cachyos kernel: BIOS-provided physical RAM map:
– Boot 296fa72905044025ae7df5ca2957eac7 –
mai 18 23:05:21 pdf-CachyOS kwin_wayland[2371]: With the output of ‘sudo dmesg’ and ‘journalctl --user-unit plasma-kwin_wayland --boot 0’
– Boot c16359c5b5b9478693e4d051dab7142f –
mai 18 23:05:46 cachyos kernel: Command line: quiet nowatchdog splash rw rootflags=subvol=/@ root=UUID=5b60956b-e1ef-4ec7-91b4-96ac643ddb3a
– Boot 296fa72905044025ae7df5ca2957eac7 –
mai 18 23:05:21 pdf-CachyOS kwin_wayland[2371]: Please report this at Making sure you're not a bot!
– Boot c16359c5b5b9478693e4d051dab7142f –
mai 18 23:05:46 cachyos kernel: Linux version 7.0.8-1-cachyos (linux-cachyos@cachyos) (clang version 22.1.5, LLD 22.1.5) #1 SMP PREEMPT_DYNAMIC Fri, 15 May 2026 13:22:28 +0000
– Boot 296fa72905044025ae7df5ca2957eac7 –
mai 18 23:05:21 pdf-CachyOS kwin_wayland[2371]: Pageflip timed out! This is a bug in the amdgpu kernel driver
mai 18 23:05:21 pdf-CachyOS kernel: amdgpu 0000:03:00.0: Failed to export SMU metrics table!
mai 18 23:05:21 pdf-CachyOS kernel: in params:00000005
mai 18 23:05:21 pdf-CachyOS kernel: amdgpu 0000:03:00.0: SMU: bus error for message: TransferTableSmu2Dram(18) response:0xFFFFFFFF
mai 18 23:05:21 pdf-CachyOS kernel: amdgpu 0000:03:00.0: device lost from bus!
mai 18 23:05:20 pdf-CachyOS kernel: amdgpu 0000:03:00.0: Failed to get fan speed(PWM)!
mai 18 23:05:20 pdf-CachyOS kernel: amdgpu 0000:03:00.0: Failed to export SMU metrics table!
mai 18 23:05:20 pdf-CachyOS kernel: in params:00000005
mai 18 23:05:20 pdf-CachyOS kernel: amdgpu 0000:03:00.0: SMU: bus error for message: TransferTableSmu2Dram(18) response:0xFFFFFFFF
mai 18 23:05:20 pdf-CachyOS kernel: amdgpu 0000:03:00.0: device lost from bus!
mai 18 23:05:20 pdf-CachyOS kernel: amdgpu 0000:03:00.0: Failed to export SMU metrics table!
mai 18 23:05:20 pdf-CachyOS kernel: in params:00000005
mai 18 23:05:20 pdf-CachyOS kernel: amdgpu 0000:03:00.0: SMU: bus error for message: TransferTableSmu2Dram(18) response:0xFFFFFFFF
mai 18 23:05:20 pdf-CachyOS kernel: amdgpu 0000:03:00.0: device lost from bus!
mai 18 23:04:59 pdf-CachyOS kdeconnectd[3225]: No uuids found for “D0:5A:00:01:D7:36”
mai 18 23:04:55 pdf-CachyOS pika-backup[3446]: 2026-05-18T21:04:55.116293Z INFO pika_backup_common::config::loadable:16: Loading file “/home/pdf/.config/pika-backup/schedule_status.json”
mai 18 23:04:55 pdf-CachyOS pika-backup[3446]: 2026-05-18T21:04:55.116271Z INFO pika_backup_common::config::loadable:85: Reloading file after change Some(“/home/pdf/.config/pika-backup/schedule_status.json”)

2

To me, this smells of either a thermal problem, a dying power supply or faulty RAM. Did you check your temperatures? And did you run Memtest86+ yet?

PS: sorry, I didn’t get that “GPU lost from bus”. Maybe it’s something else completely.

PPS: sometimes, the simplest solutions are the best: check if your graphics card is firmly in its slot and the safety latch (or whatever that thing is called) is clicked in, see step 2 here.

3

Thermals ? At idle/browsing everything is sub 40c

GPU has been firmly reseated twice

RAM issue ? Wonder how it can take 12hrs+ to crash and memtest won’t help

Yeah PSU maybe but again why so unfrequently ?

GPU lost from bus here :slight_smile: :

mai 18 23:05:21 pdf-CachyOS kernel: amdgpu 0000:03:00.0: device lost from bus!

4

It is a little difficult to get a proper idea of what your setup really is like. Like, are you using more than one monitor?! If you feel comfortable with it you can run sudo cachyos-bugreport.sh and share the link it provides here for people to better diagnose it.

I do not know what kernel updates might have subtly done in the background, but to me it looks like the PCIe link for your GPU is getting fully dropped which is possibly due to power saving controls of some sort. It has been a long time since I last looked at it, but I would look for “Power Supply Idle Control”(I think it is called) in the BIOS options and switch it away from auto and test it out. Or wait until someone else has a better idea of what is going on.

5

Think it’s solved now, no reboot in 3 days, fix was some kernel params - got it from a guy on the discord :
pcie_ports=native pci=noaer amdgpu.dcdebugmask=0x12

So a compatibility issue with the GPU pcie link, not a hardware problem

6

Something similar was already discussed here not too long ago, although to me it seems that with

pci=noaer

you just disable some error logging which is not a fix to anything but might make debugging issues actually harder (because, well, you are disabling some logs), so I’d maybe re-enable that and see what happens.

7

My issue wasn’t the same, no freeze but black screen then reboot, with either no logs or “GPU Lost from bus”
Indeed chatgpt also told me noaer just disable some error logging, will remove it