My PC normally runs Fedora Linux, but it also boot-loops when I try to start Windows 11. Fedora displays these machine-check errors before rebooting:
mce: [Hardware Error]: CPU 2: Machine Check: 0 Bank 5: bea0000000000105
mce: [Hardware Error]: TSC 0 ADDR 1ffffc15c401c MISC d01a000000000000 SYND 4d000000 IPID 500b000000000
mce: [Hardware Error]: PROCESSOR 2:870f10 TIME 1787682678 SOCKET 0 APIC 4 microcode 8701034
The system had already been stuttering and restarting occasionally for several months, and the problem became more frequent during recent hot weather. Windows begins loading, then freezes and reboots, after which automatic recovery reports that it cannot repair the problem.
After being powered off for a few hours, the computer started normally again. Since the errors occur in both operating systems, I suspect a hardware or overheating problem. How can I determine whether the CPU is actually failing, and should I check the cooling system, fans, thermal paste, memory, BIOS settings, or another component?
3 Answers
Check actual temperatures instead of assuming heat is the cause. Record idle temperatures and temperatures under load with a hardware monitor. If the CPU quickly reaches its thermal limit, inspect that the heatsink is firmly mounted, the fan or liquid-cooler pump is running, and the case airflow is clear. Replacing dried thermal paste may help, but it will not fix a failed pump, fan, or other unstable component.
Since the processor is identifying the machine-check event, start by returning the BIOS to factory defaults and disabling memory overclock profiles such as XMP or EXPO. Then test one memory module at a time, check the CPU cooler, and inspect power and motherboard connections. If temperatures and memory tests are normal but the errors continue, testing with known-good parts or replacing the CPU may be necessary; the log alone cannot identify the exact failed component.
Because the crashes happen under both Fedora and Windows, a normal operating-system problem is unlikely. A machine-check exception is reported by the CPU, but it does not prove the CPU itself is defective; overheating, unstable BIOS settings, RAM, the motherboard, or the power supply can also trigger one. Reset any overclocking or undervolting, update the BIOS if appropriate, and run separate memory and CPU stability tests while monitoring temperatures.
A complete shutdown can sometimes clear a temporary hardware or firmware state, while a warm restart does not. That may explain why shutdown worked, but repeated crashes still need investigation rather than being treated as harmless.