Confirmation of an issue discovered in Kernel 7.0.0-38 dealing with the screen flickering during high performance. The following are the detail of my situation and why i reverted back to kernel 7.0.0-34 which resolved my problem. post has been made to Ubuntu as the issue has also been reported there.
(https://bugs.launchpad.net/ubuntu/+s.../2169181/+edit)
Hardware: Beelink SER4 mini PC, AMD Ryzen 7 4800U (Renoir iGPU, DCN 2.1), 32 GB RAM
Display: 1920x1080 @ 60 Hz over HDMI (HDMI-A-2)
OS: Kubuntu 26.04, KDE Plasma (Wayland)
Problem: With kernel 7.0.0-38-generic, the screen randomly flashes whenever the
CPU is under sustained heavy load (HandBrake encoding, or stress-ng --cpu 0).
Flashes appear within a few minutes of load starting.
Booting 7.0.0-34-generic on the same system, with the same settings, gives no
flashes over 25+ minutes of full CPU load (HandBrake) plus ~50% GPU load.
Reproduce: stress-ng --cpu 0 --timeout 20m on 7.0.0-38.
Ruled out:
- HDMI cable and ports (swapped)
- Temperature (CPU Tctl ~75°C at full load)
- Power draw (still flashes with CPU capped to 1.4 GHz, PPT ~10 W)
- CPU scheduling (still flashes with stress-ng at nice 19)
- Clocksource (TSC was falling back to HPET; tsc=reliable did not change it)
- GPU clock switching (power_dpm_force_performance_level=high did not change it)
- amdgpu.sg_display=0 is set on both kernels
No amdgpu/drm errors appear in dmesg when the flashes occur.
This looks like a regression between 7.0.0-34 and 7.0.0-38.
(https://bugs.launchpad.net/ubuntu/+s.../2169181/+edit)
Hardware: Beelink SER4 mini PC, AMD Ryzen 7 4800U (Renoir iGPU, DCN 2.1), 32 GB RAM
Display: 1920x1080 @ 60 Hz over HDMI (HDMI-A-2)
OS: Kubuntu 26.04, KDE Plasma (Wayland)
Problem: With kernel 7.0.0-38-generic, the screen randomly flashes whenever the
CPU is under sustained heavy load (HandBrake encoding, or stress-ng --cpu 0).
Flashes appear within a few minutes of load starting.
Booting 7.0.0-34-generic on the same system, with the same settings, gives no
flashes over 25+ minutes of full CPU load (HandBrake) plus ~50% GPU load.
Reproduce: stress-ng --cpu 0 --timeout 20m on 7.0.0-38.
Ruled out:
- HDMI cable and ports (swapped)
- Temperature (CPU Tctl ~75°C at full load)
- Power draw (still flashes with CPU capped to 1.4 GHz, PPT ~10 W)
- CPU scheduling (still flashes with stress-ng at nice 19)
- Clocksource (TSC was falling back to HPET; tsc=reliable did not change it)
- GPU clock switching (power_dpm_force_performance_level=high did not change it)
- amdgpu.sg_display=0 is set on both kernels
No amdgpu/drm errors appear in dmesg when the flashes occur.
This looks like a regression between 7.0.0-34 and 7.0.0-38.