Fedora 43 on LG Gram / Ryzen 5 7530U: TSC marked unstable, system falls back to HPET and becomes laggy
Hi,
I’m having a performance/stutter issue on Fedora 43 with an LG Gram laptop using an AMD Ryzen 5 7530U.
Current kernel:
6.18.4-200.fc43.x86_64
Clocksource:
current_clocksource: hpet
available_clocksource: hpet acpi_pm
Relevant dmesg:
TSC synchronization [CPU#0 -> CPU#2]:
Measured 1287282640 cycles TSC warp between CPUs, turning off TSC clock.
tsc: Marking TSC unstable due to check_tsc_sync_source failed
clocksource: Switched to clocksource hpet
Unstable clock detected, switching default tracing clock to "global"
I accidentally booted a +debug kernel before, but I already fixed that. I’m now on the normal Fedora kernel, and the TSC issue still happens.
Boot args:
root=UUID=... ro rootflags=subvol=root rhgb quiet amd_iommu=off
No clocksource=, tsc=, nohpet, or hpet=disable parameters.
Questions:
- Has anyone seen this on Ryzen laptops or LG Gram models?
- Could this be a Fedora kernel regression, BIOS/UEFI issue, or ACPI/power management problem?
- Should I test older kernels like 6.17 or 6.14?
- Is testing
tsc=reliable clocksource=tsctoo risky with this kind of TSC warp?
Any advice would be appreciated.Hi,I’m having a performance/stutter issue on Fedora 43 with an LG Gram laptop using an AMD Ryzen 5 7530U.Current kernel:6.18.4-200.fc43.x86_64Clocksource:current_clocksource: hpet
available_clocksource: hpet acpi_pmRelevant dmesg:TSC synchronization [CPU#0 -> CPU#2]:
Measured 1287282640 cycles TSC warp between CPUs, turning off TSC clock.
tsc: Marking TSC unstable due to check_tsc_sync_source failed
clocksource: Switched to clocksource hpet
Unstable clock detected, switching default tracing clock to "global"I accidentally booted a +debug kernel before, but I already fixed that. I’m now on the normal Fedora kernel, and the TSC issue still happens.Boot args:root=UUID=... ro rootflags=subvol=root rhgb quiet amd_iommu=offNo clocksource=, tsc=, nohpet, or hpet=disable parameters.Questions:Has anyone seen this on Ryzen laptops or LG Gram models?
Could this be a Fedora kernel regression, BIOS/UEFI issue, or ACPI/power management problem?
Should I test older kernels like 6.17 or 6.14?
Is testing tsc=reliable clocksource=tsc too risky with this kind of TSC warp?Any advice would be appreciated.
