I have an Intel Core i7-14700K running Windows 11, with Arch Linux installed on a separate drive, and I've recently started having crashes when launching demanding games such as Cyberpunk 2077. The games close during loading without showing a blue screen, while lighter games like Slay the Spire work normally. The same general problem occurs in Linux.
To narrow it down, I ran Cinebench 2026. The single-core and single-threaded tests complete successfully, but the multithreaded test crashes almost immediately. Other component tests and benchmarks haven't shown any obvious problems. The system is about two and a half years old, and the issue has become noticeably worse within the past six months.
I haven't manually overclocked the processor, although XMP is enabled for the memory. Temperatures remain reasonable during the crashes, and the applications themselves close rather than shutting down the entire computer.
Is there a reliable way to determine whether the CPU is degrading without having another processor available for comparison?
3 Answers
The quickest isolation test is to disable XMP and return the memory to standard settings. You can also temporarily reduce the CPU multiplier or maximum turbo frequency by roughly 500 MHz and rerun the multithreaded benchmark. If the crashes stop only after reducing CPU speed, that strongly points toward an unstable or degraded CPU, though memory or motherboard power delivery can still be involved.
Because this is a 13th- or 14th-generation Intel desktop processor and the problem has gradually appeared over time, CPU degradation is a real possibility. First update the motherboard BIOS to the latest version, including the current Intel microcode, then test again with BIOS defaults. If the problem remains, Intel warranty support may be able to replace the processor.
Since the programs are closing rather than the whole PC powering off, heat and PSU protection are less likely, although they shouldn’t be ruled out completely. Run a CPU stress test with XMP disabled, check Windows Event Viewer for hardware errors, and look for files in C:WindowsMinidump. If there are no blue screens or dump files, an application crash may not generate one. Also test the RAM separately, because an unstable memory profile can make demanding multithreaded workloads fail in a way that looks like a CPU problem.
I haven’t changed CPU voltages or overclocked it, but XMP is enabled. The crashes only close the applications, and temperatures stay well within normal limits.