My seven-year-old PC suddenly shows a WHEA_UNCORRECTABLE_ERROR blue screen while playing demanding games such as Wuthering Waves and Genshin Impact. The system has a Ryzen 5 2600, RX 580 8GB, 32GB of 3000 MHz RAM, a 224GB NVMe SSD, and a 2TB SATA SSD. I have reseated the RAM, graphics card, and CPU, run several Command Prompt checks, and used DDU to reinstall the graphics drivers. These steps sometimes let me play temporarily, but the crashes return. I suspect the NVMe SSD might be failing and want to know how to check its health in Windows. Reseating the SSD also seemed to help briefly. What hardware or diagnostic tests should I try to identify the cause?
3 Answers
WHEA_UNCORRECTABLE_ERROR is a hardware-level error, so the SSD is only one possible cause. Since the crash happens during gaming, test the graphics card and its VRAM with tools such as OCCT or FurMark. If the system crashes or the tests report errors, the RX 580 may be failing or overheating. Also monitor temperatures and check that the card’s power connections are secure.
For the SSD, check its SMART health and temperature with a drive-monitoring utility from the SSD manufacturer or a trusted tool such as CrystalDiskInfo. You can also run Windows’ error checks, but a clean result does not completely rule out an intermittent drive, slot, cable, or motherboard problem. Test the NVMe drive in another slot if available, inspect its mounting, and run separate memory and CPU stability tests too.
Check whether Windows is creating crash dumps in C:WindowsMinidump. Those files can often show whether the failure points toward the CPU, memory, graphics card, storage, or another device. If there are no dumps, configure Windows to create small memory dumps and collect several crashes for comparison.
I’ll check for minidumps after the next crash. Reseating the NVMe drive let me play for about ten minutes without a failure, but I’m not sure yet whether that actually fixed anything.
The RX 580 came from a reputable local computer store and has been in this PC for years, so it is not an unfamiliar marketplace model. I’ll still run GPU and VRAM stress tests if the crashes return.