Hi folks - I’ve been having issues since building my current PC.
Sometimes my screens freeze for a few seconds and it recovers (sometimes it does this several times in a row). Sometimes my screens freeze and audio also stops working. I can wait several minutes - in this case, my computer never crashes, it just stays frozen and I have to hold the power button to get going again.
For the former issue, it was this error found in journalctl
fedora kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, but soft recovered
After much searching, this seems like it may have been an issue where my GPU clock speeds were too high by default, so I installed CoreCtrl and have limited the speed and power limit to what I think the “default” should be.
I haven’t had any of the former issues recently, so the CoreCtrl clock speed change may have fixed that issues, but I’m still occasionally having my computer completely freeze and be non-responsive for multiple minutes with no recovery/change. I read online in these cases you can use “Raise The Elephant” where you press SysRq key sequences to interact directly with the kernel. Unless I’m pressing the wrong keys, this doesn’t work. When the computer completely freezes, I also see no associated output from journalctl until the next boot starts.
Any ideas how I can debug what the issue is? It can occur while playing games, or simply watching YouTube.
Specs:
Fedora 40 KDE workstation
AMD Ryzen 7 7800X3D
AMD RX 7900 XTX
MAG B650 TOMAHAWK WIFI