STATUS_MCA_EXCEPTION (0xC0000713) – Machine Check Architecture Crash Fix

Cause 1: Memory Overclock (XMP / EXPO) – This is the #1 trigger

What's actually happening here is the CPU's Machine Check Architecture (MCA) is reporting an unrecoverable error from the memory controller or RAM. Most people hit 0xC0000713 after enabling XMP (Intel) or EXPO (AMD) to run their RAM at its rated speed. The rated speed—say 6000 MHz DDR5—is often beyond what the CPU's integrated memory controller (IMC) can handle without extra voltage or looser timings.

The crash usually happens during boot, video encoding, or heavy game loading. You'll also see Event ID 18 from WHEA-Logger in the System event log.

Fix: Dial back or manually tune the memory

1. Enter UEFI/BIOS (spam Del/F2 during boot).
2. Disable XMP/EXPO and set memory to JEDEC default (usually 4800 MHz for DDR5, 2133/2400 for DDR4). Reboot—if the crash stops, the overclock is the cause.
3. If you need the rated speed, don't just enable XMP. Manually set the DRAM voltage (1.35V for DDR4, 1.25–1.4V for DDR5 depending on kit) and loosen the primary timings by +2 from the XMP profile. For example, if XMP says CL30-38-38-96, try CL32-40-40-98.
4. Also set CPU VCCSA (System Agent) voltage to 1.25–1.30V on Intel, or VSOC to 1.20–1.25V on AMD. Many boards undervolt these when XMP is enabled, causing instability.

Why step 4 works: The memory controller lives inside the CPU. It needs enough voltage to drive the RAM sticks at high frequencies. Auto settings often starve it.

Cause 2: Faulty or mismatched RAM stick

Even at stock JEDEC speeds, one module might be physically failing. A single-bit error in a critical memory address triggers MCA. This is common with mixed kits—two sticks from different batches, or using four sticks when the board only supports two at high speeds.

Fix: Test each stick individually

1. Run MemTest86 (free version). Let it loop overnight—one pass isn't enough. Any red errors = bad stick.
2. If no errors but the crash persists, try running only one stick in slot A2 (second from CPU). Boot and stress-test (Prime95 large FFTs for 30 min). Swap sticks. If only one stick crashes, it's defective.
3. For four-stick setups: drop speed by 400 MHz (e.g., 3600 → 3200 on DDR4) and increase DRAM voltage by 0.05V. The memory controller struggles to drive four ranks.

Cause 3: CPU voltage too low under load (unstable undervolt or PSU issue)

Less common but real: the CPU itself is not getting clean power. This can happen after a manual undervolt (e.g., Ryzen Curve Optimizer set too aggressively) or a failing power supply that can't sustain voltage during transient spikes. The 0xC0000713 appears randomly under heavy load—compiling code, rendering video, or gaming.

Fix: Reset CPU voltage to stock, then check PSU

1. Reset BIOS to defaults. Disable any per-core offset or Curve Optimizer. If the crash stops, your undervolt was too aggressive.
2. If you're not undervolting, monitor +12V and +5V rails in HWInfo64 during a stress test. If the +12V rail drops below 11.8V under load, your PSU is failing. Replace it—don't keep using it, it can damage other components.
3. Also check CPU Vcore in HWInfo64. If it droops more than 0.1V below what you set in BIOS, enable LLC (Load-Line Calibration) to level 3 or 4. That reduces voltage drop under load.

The reason LLC helps: When the CPU draws sudden current, the voltage sags. LLC increases voltage under load to compensate, preventing the MCA error.

Quick-reference summary table

Skip the Windows reinstall—this error isn't software. It's your hardware telling you something is wrong. Fix the memory or power delivery first, and you'll never see 0xC0000713 again.