WHEA_UNCORRECTABLE_ERROR: what causes it and how to fix it

What it means

WHEA is the Windows Hardware Error Architecture: the way the processor, the memory controller and PCIe devices report their own errors to Windows. Most errors are caught and corrected on the spot, and only logged. This one could not be corrected, so Windows stopped. Microsoft describes the code as typically a physical hardware failure — heat, a defective part, memory, or a processor beginning to fail — and its first advice is to switch off any overclocking and make sure the fans work.

Find out which part reported it

After the restart, open Event Viewer, then Windows Logs, System, and look for an error from WHEA-Logger at the time of the crash: event 18, “A fatal hardware error has occurred”. It says which part reported it:

• Processor Core, with error source Machine Check Exception: the processor. By far the most common.
• Memory: the memory or its controller.
• PCI Express Root Port or a PCIe device: the graphics card, an NVMe SSD or the slot it sits in.

Look further back too. WHEA-Logger event 19, “A corrected hardware error has occurred”, from the same part, is the warning that comes first: the part made a mistake and caught it, which it only does at the edge of stability.

The usual causes

• A processor overclock or undervolt, including motherboard settings that raise or lift the processor’s power limits, and AMD’s Curve Optimizer set too far.
• Intel 13th and 14th gen desktop processors running microcode older than Intel’s fix. The instability these processors develop shows up as hardware errors among other crashes — how to check yours.
• A memory profile (XMP or EXPO) faster than this PC can hold, especially with four sticks.
• Heat: a cooler badly seated or clogged with dust.
• Power: a power supply too small, failing, or with a loose cable.
• A PCIe device not seated properly, or a slot set to a speed the card or riser cable cannot hold.

How to fix it

1. Update the BIOS. Processor microcode, Intel’s 13th and 14th gen fix among it, and memory compatibility improvements arrive that way.
2. Go back to defaults for a few days. Load the BIOS defaults, or switch off XMP or EXPO and any overclock or undervolt. If the crashes stop, bring the settings back one at a time to find the one it cannot hold.
3. Test it under load. A stress test with temperatures shows quickly whether the processor and memory hold at full speed and full heat. One that fails at default settings points at the part itself, or at power.
4. If PCI Express reported it, reseat the graphics card and the SSD, and as a test set the slot to the previous PCIe generation in the BIOS.
5. Check the power supply against the parts with the PSU calculator, and every power cable at both ends.
6. On a 13th or 14th gen Intel processor that still crashes after the fix, it may already be worn: the fix stops the wear but cannot undo it. Intel extended the warranty on these processors by two years.