Quick answer

WHEA_UNCORRECTABLE_ERROR (0x124) means hardware reported a fatal error Windows couldn’t contain; it doesn’t name a failed part. Return any overclock, undervolt or XMP/EXPO profile to default, check that fans and the cooler work, and install Windows and BIOS updates. Then read the WHEA-Logger event in Event Viewer, or the dump in WinDbg, to see which branch to test.

Fix steps

6 steps

Diagnostic workbench
Contents

At a glance

Applies to
Windows 11 and Windows 10, stop code 0x124
First fix
Overclock, undervolt and XMP/EXPO back to default
Intel 13th/14th
Intel Default Settings and a BIOS with microcode 0x12F or later
No tools needed
Event Viewer, System log, WHEA-Logger source
Replace a part
Only when the record or a vendor test pins it

Provenance

  • Oct 3, 2026Published
  • Nov 18, 2026Last substantive update

WHEA_UNCORRECTABLE_ERROR means hardware raised a fatal error that Windows could not contain. Stop code 0x00000124 (usually shortened to 0x124) is the operating system halting before corrupted state hits disk.

Don’t buy a replacement CPU yet. Seeing AuthenticAMD.sys or GenuineIntel.sys inside a crash viewer does not prove the silicon is dead. Windows Hardware Error Architecture (WHEA) simply forwards the packet raised by a CPU core, memory controller, PCIe root port, or NVMe endpoint. Read that packet first.

Microsoft’s 0x124 reference gives you the two fields that route the investigation:

  • Parameter 1 identifies the kind of error source that reported the problem.
  • Parameter 2 points to the WHEA_ERROR_RECORD that describes it.

The same page’s cause section is the short list to start from: the bug check is typically related to physical hardware failure, it can be heat related, and Microsoft says to disable overclocking, confirm that cooling fans work, and test system memory. It calls a driver a less likely but possible cause.

I log every crash timestamp, BIOS offset, and test pass inside the Hardware Diagnostic Workbench so one bad memory profile or per-core Curve Optimizer tweak doesn’t get confused with a dead motherboard.

1. Make sure 0x124 was the crash, even on silent reboots

Photograph the blue screen if it stays visible. Did the PC black-screen and reboot instantly? Open Event Viewer > Windows Logs > System and find the Kernel-Power event 41 from the restart. Microsoft’s Event ID 41 page says its BugcheckCode field records the stop code in decimal, so 292 means 0x124, and BugcheckParameter1 to 4 hold the four bug-check parameters. The random-restart guide walks the rest of that event.

Sometimes C:\Windows\Minidump is empty and Event 41 logs BugcheckCode 0. Microsoft’s Event ID 41 page lists two causes worth checking here. Power was interrupted, or the power supply is underpowered or faulty; the PSU failure guide covers that branch. Or no dump file was configured: the default dump file is the page file, and a volmgr event 46 (“Crash dump initialization failed!”) points to the page-file configuration.

When a machine reboots silently into the BIOS screen with the boot drive missing until you flip the PSU switch off and on, stop chasing CPU voltage. Power down, unplug AC power, reseat that M.2 NVMe SSD (checking that no thick thermal pad is bowing the PCB), and check the drive’s firmware before testing anything else.

Keep corrected warnings distinct from fatal stops too. Microsoft’s WHEA error-processing documentation says a corrected error is logged in the system event log once it passes the error source’s threshold, while a fatal uncorrected error makes the kernel generate a bug check. Both matter, and Event Viewer connects them.

2. Put deliberate tuning back to default

Start with BIOS defaults. It’s the cheapest step on this page and it doesn’t need any tool. On ASUS boards the option is Load Optimized Defaults (F5 in the BIOS, per ASUS’s support FAQ); other board makers label it differently, so check your board manual.

If you altered a Curve Optimizer offset, CPU core voltage, load-line level, or XMP/EXPO memory profile, write down the current numbers and return only those deliberate tweaks to documented stock defaults. The Curve Optimizer guide and the XMP and EXPO guide show where those settings live. Then use the PC normally for about 48 hours, including the exact game or idle workload that crashed, before touching a second knob. That duration is the same rule of thumb the 0x1A and 0x50 guides use, not a pass mark.

Confirm the cooling works while you’re in there: every fan and AIO pump spinning, and CPU temperature under the crashing workload within the chip’s limit (the safe CPU temperature guide has the numbers).

13th or 14th Gen Intel Core desktop chip? For the Vmin Shift Instability issue on some of those processors, Intel’s support article recommends the relevant Intel Default Settings in BIOS and the latest BIOS containing microcode 0x12F or later. Intel has also extended warranty coverage by two years for eligible affected processors, up to five years from the original purchase date. If your PC has one of those chips, apply that before anything else on this page; the BIOS update guide walks through the flash. Microsoft’s advanced stop-error steps also list an up-to-date BIOS and firmware among their general checks for any stop error.

Check your BitLocker key before entering UEFI. Microsoft says a hardware change can bring up a recovery-key prompt at startup and that its support staff can’t retrieve or recreate a lost key, so verify access via Microsoft’s recovery-key instructions first and leave TPM or Secure Boot untouched.

If 0x124 stops at stock settings, the tuning was the likely cause; keep the defaults. If crashes continue, keep the defaults anyway and read the record below.

Microsoft Support’s consumer WHEA fix page adds Windows Update for the latest updates and drivers, and System Restore or Startup Repair when the error began after an update, a new driver or a new app.

3. Decode Event Viewer WHEA-Logger Event ID 18 vs Event ID 17

Even when no .dmp file survives, Windows frequently writes a hardware entry into Event Viewer -> Windows Logs -> System right as the machine comes back up (or in the seconds leading up to the crash). Filter the System log by Event source WHEA-Logger.

Open the Details tab (Friendly View or XML View) on any Microsoft-Windows-WHEA-Logger entry. Event ID 18 and Event ID 17 point to different hardware subsystems:

Event ID & severityKey fields inside Event DetailsWhat the record namesTargeted isolation action
Event ID 18 (Error — Fatal hardware error, Processor Core)Error Type: Cache Hierarchy Error + Processor APIC ID: <N>A cache error reported by the processor’s machine-check source, with the APIC ID of the logical processor that raised it.Write down the APIC ID from each crash. If you run a per-core Curve Optimizer offset or an Intel undervolt, return it to default first (Step 2) and see whether the errors stop.
Event ID 18 (Error — Fatal hardware error, Processor Core)Error Type: Bus/Interconnect Error or Memory Hierarchy ErrorA bus or memory-hierarchy error rather than one core’s cache.Disable XMP/EXPO and custom FCLK overrides in UEFI. Test stock JEDEC memory speed before suspecting the processor.
Event ID 17 (Warning — Corrected error, PCI Express Root Port / Endpoint)Primary Device Name: PCI\VEN_...&DEV_..., often in repeating entriesPCIe link-training, replay timer, or CRC errors on a GPU riser cable, graphics card slot finger, or M.2 NVMe socket. An uncorrectable PCIe error is the 0x124 case with Arg1 = 0x4.Copy the VEN_xxxx&DEV_xxxx string, match it in Device Manager -> View -> Devices by connection (Details -> Device instance path), bypass any PCIe riser cable, reseat the GPU or M.2 drive, and test stepping that slot from Gen 5/Gen 4 down one generation in UEFI.

I always check Processor APIC ID across three or four Event ID 18 crashes before touching BIOS. Seeing APIC ID 0 crash one afternoon and APIC ID 1 (0 and 1 share physical Core 0 on a 2-thread-per-core Ryzen chip) crash the next morning during idle YouTube playback is the textbook signature of an aggressive -25 or -30 Curve Optimizer offset—not a dead CPU, and never a reason to reinstall Windows.

4. Preserve the dump before changing more settings

Check C:\Windows\Minidump for a .dmp file matching the crash time, or C:\Windows\MEMORY.DMP for a kernel dump. Copy it to a separate folder immediately so Storage Sense or a repeat crash cannot overwrite it.

No dump folder at all? Open System Properties > Advanced > Startup and Recovery > Settings and confirm Small memory dump (256 KB) (%SystemRoot%\Minidump) or Automatic memory dump is enabled alongside an active system-managed pagefile on C:.

Jot down everything that changed prior to the first 0x124:

  • Curve Optimizer, Intel voltage offset, or load-line calibration edits
  • XMP or EXPO memory profiles
  • a recent UEFI/AGESA/microcode flash
  • a PCIe riser cable, new GPU, or secondary M.2 SSD
  • cooler mounting or pump header changes
  • whether crashes hit at idle or light desktop load versus sustained 100% load

5. Read Parameter 1 and the error record in WinDbg

Open the copied dump in WinDbg via File > Open crash dump and run:

!analyze -v

Locate WHEA_UNCORRECTABLE_ERROR (124) and copy Arg1 (Parameter 1) plus Arg2 (Parameter 2). Feed the Arg2 hex address directly into WinDbg’s WHEA extension, using the exact pointer from your own dump:

!errrec <Arg2 value>

Microsoft’s !errrec documentation shows the command displaying the contents of a WHEA error record, and the error-record reference describes its header, severity and section descriptors.

A small memory dump holds limited information. If it doesn’t give you a usable record, switch Startup and Recovery to Kernel memory dump via Microsoft’s recovery options guide so the next crash captures kernel memory.

Parameter 1 picks the branch, not the replacement part

Microsoft’s 0x124 table maps Parameter 1 (Arg1) to the reporting error source:

Parameter 1Error sourceWhat the branch establishes
0x0Machine check exception (MCE)A CPU machine-check bank raised the fault. Inspect Event ID 18 (Cache vs Bus/Interconnect) and !errrec before choosing core, IMC/RAM, or thermal tests.
0x3Nonmaskable interrupt (NMI)Platform NMI asserted. Check board/OEM hardware logs and firmware notes.
0x4Uncorrectable PCI Express errorA PCIe root port or endpoint reported an uncorrectable error. Check Event ID 17, GPU risers, and NVMe M.2 slots before the CPU.
0x5Generic hardware errorPlatform firmware reported a generic hardware error; read the !errrec section descriptor.
0x7 / 0x10Boot error / Device driver error sourceA firmware boot error record, or an error reported through a device driver. Read the record’s section to see which device.

Stay disciplined on Arg1 = 0x0. Microsoft’s hardware-error source reference says a processor’s machine check exception typically reports processor errors, cache and memory errors, and system bus errors, so a machine-check source alone doesn’t single out the cores. Check the error-record section first.

6. Follow the record’s branch

Machine-check or processor section

Check thermals first. Confirm every fan and AIO pump spins normally and compare peak CPU die temperature under the crashing workload against the chip’s published limit using the CPU overheating guide.

Did the error record implicate the memory hierarchy or follow an XMP/EXPO change? Follow the MEMORY_MANAGEMENT test sequence step by step: stock JEDEC memory clocks first, bootable memory testing second, and single-stick slot isolation only if errors appear. Because the machine-check source also reports cache, memory and bus errors, keep the .dmp file and !errrec output ready for the board or CPU vendor before treating a processor swap as anything other than a final isolation step.

PCIe section or Parameter 1 0x4

Pull the PCIe device ID from !errrec or Event ID 17 and match it against recent hardware changes—such as a vertical GPU riser cable, a newly added M.2 SSD, or a BIOS update. Shut the PC down completely and unplug AC power before reseating one card or M.2 drive at a time per the motherboard manual.

Never assume every PCIe fault is the graphics card. M.2 NVMe drives and onboard Wi-Fi/LAN controllers share the PCIe root complex too.

Generic, NMI or platform-specific section

Generic (0x5) and NMI (0x3) records come from platform firmware rather than a standard CPU core check. Save the complete !errrec output and run the motherboard or OEM vendor’s own hardware diagnostics instead of guessing at parts.

What not to do first

  • Do not reinstall Windows before reading the dump. Wiping the drive erases C:\Windows\Minidump and Event Viewer history without fixing a hardware fault.
  • Do not blindly raise core or SoC voltages. Adding voltage changes electrical behavior and masks which setting caused the fault.
  • Do not replace the CPU over AuthenticAMD.sys or GenuineIntel.sys. Machine-check banks report cache, memory and bus faults under the same module name.
  • Do not confuse corrected Event ID 17 warnings with fatal crashes. Log the warning count and device path, while keeping severity clear.
  • Do not swap three parts in one afternoon. Simultaneous swaps hide which change fixed the crash.

The stopping rule

RMA or replace a part only when the WHEA record, slot isolation, or a vendor diagnostic pins the fault to that specific component; for an eligible 13th or 14th Gen Intel desktop chip affected by the Vmin Shift issue, Intel’s extended warranty applies. No dump file yet? Hold stock BIOS defaults, verify pagefile and small-dump settings, and let one clean capture tell you the answer.

Other stop codes take their own route: KERNEL_SECURITY_CHECK_FAILURE (0x139) and CRITICAL_PROCESS_DIED have separate guides.

Sources

Frequently asked questions

What causes WHEA_UNCORRECTABLE_ERROR?
Microsoft says 0x124 is typically related to physical hardware failure: it can be heat related or come from defective hardware, memory, or a processor that is beginning to fail. Overclocking is on its checklist to disable, and it calls a driver a less likely but possible cause. The dump’s Parameter 1 and the WHEA record tell you which hardware source reported it.
How do I fix WHEA_UNCORRECTABLE_ERROR in Windows 11?
Disable any overclock, undervolt or XMP/EXPO profile, confirm that the fans and cooler work, and install the latest Windows updates and drivers. If the error started after an update, a new driver or a new app, Microsoft suggests System Restore or Startup Repair. On a 13th or 14th Gen Intel Core desktop chip, apply Intel Default Settings and a BIOS with microcode 0x12F or later. If it returns, read the WHEA record.
Does WHEA_UNCORRECTABLE_ERROR mean the CPU is failing?
No. Microsoft says one processor machine-check source can report processor, cache, memory and system-bus errors. A machine-check record narrows the branch, but it is not a CPU replacement verdict. Return changed clock, voltage and memory-profile settings to defaults, read the WHEA record, and isolate the branch before replacing a part.
How do I read a WHEA 0x124 crash dump?
Open the dump in WinDbg and run !analyze -v. Parameter 1 identifies the error-source type. Parameter 2 is the address of the WHEA error record; pass that address to !errrec. Read the record’s severity, notification type and section type together. Microsoft documents processor, platform-memory, PCIe, bus, device, NMI and generic record sections.
Is a corrected WHEA-Logger event the same as a 0x124 blue screen?
No. WHEA records corrected errors in the system event log after its reporting threshold is reached. A fatal uncorrected error is saved and Windows generates a bug check. Keep corrected events as evidence, but don’t count every corrected event as a WHEA_UNCORRECTABLE_ERROR crash.
Should I reinstall Windows for WHEA_UNCORRECTABLE_ERROR?
Not first. Microsoft defines 0x124 as a fatal hardware error report, and a reinstall erases the dump and event history without changing the hardware. Restore deliberate tuning to defaults, keep the dump, and test the branch the record supports.

Sources and corrections

Last updated
Methodology
See our methodology for research and review standards. It draws on 14 cited sources, listed below, each checked against the original page on the date above. No first-party crash or repair is claimed. The diagnostic branches and commands come from the Microsoft and Intel documentation below, first opened on 2026-10-03 and re-read on 2026-11-18. The order favors preserving the record and reversing deliberate settings changes before component isolation. Commands were not run against an owner WHEA dump for this article.
Update log
  • 2026-11-18 — Moved the reset-to-defaults step up front, added Intel’s 13th and 14th Gen guidance, and replaced uncited claims about NVMe dropouts and undervolt signatures with what Microsoft documents.
Corrections
Spotted an error? Email a correction with the affected line and what you found. The draft includes this page’s address. Verified changes are recorded here. Contact details.

About the author

I am Lowell K. Wood IV. I research, write and maintain everything on TechFuelHQ from St. Louis County, Missouri.