My PC recently started crashing with FAULTY_HARDWARE_CORRUPTED_PAGE and SYSTEM_SERVICE_EXCEPTION while I was only browsing the web. MemTest86 found no RAM errors, I checked the SSD, reinstalled the graphics drivers, and reviewed the crash logs. The system had been stable for about eight months.
As a test, I lowered my memory speed from DDR4-3200 to DDR4-2933, and the crashes appear to have stopped. I'm using 32GB as two 16GB modules rated at 16-20-20-40 and 1.35V, with no manual voltage adjustments or other overclocking.
An AI assistant suggested that the Ryzen 7 5700X's integrated memory controller may have degraded. Is that realistic, or could this instead be a motherboard, storage, driver, or Windows issue? The processor is officially rated for DDR4-3200, although many people run similar systems at 3600 MT/s. How could a previously stable setup suddenly become unreliable at its rated speed?
2 Answers
Be careful with the terminology: DDR4-3200 is commonly called 3200 MHz, but 3200 MT/s is the technically correct data-transfer rating. DDR4-3600 is also an overclock for this processor, while DDR4-3200 is the official maximum under the supported two-module configuration.
A system can run an overclocked memory profile for months and then become unstable because of temperature, BIOS behavior, aging components, or a marginal memory kit. However, if two modules at the official speed are failing, I would also investigate the motherboard, socket contact, BIOS settings, and the SSD or page file rather than assuming the CPU is worn out.
FAULTY_HARDWARE_CORRUPTED_PAGE indicates a memory-related fault, but the error may be detected when Windows reads something from the page file. That means the underlying cause can involve storage, RAM, the memory controller, drivers, or corrupted system files. SYSTEM_SERVICE_EXCEPTION is even less specific.
MemTest86 passing is useful, but it does not completely rule out intermittent problems under Windows. Check the crash dumps with a debugger, test the SSD and page file, run Windows system-file checks, and look for WHEA or storage errors in Event Viewer. If the crashes only disappear at 2933, instability in the memory path is still a strong possibility, but the dumps and hardware testing are needed before calling it IMC degradation.
Related Questions
Lenovo Thinkpad Stuck In Update Loop Install FilterDriverU2_Reload