How to Fix NVIDIA Driver Crashes and Recovery Errors

Published 2026-10-16 · Marion Remote Fix

Your screen goes black for a moment, then recovers, and Windows shows a notification: "Display driver nvlddmkm stopped responding and has successfully recovered." Or your games crash to desktop with a driver error, the screen flickers repeatedly, or you get a blue screen with VIDEO_TDR_FAILURE or nvlddmkm.sys. These are NVIDIA driver crashes, and they are one of the most common GPU issues on Windows. The cause is usually a driver problem, a power delivery issue, overheating, or a GPU hardware failure. Here is how to fix each cause.

Clean install the NVIDIA driver

The most common fix for NVIDIA driver crashes is a fresh driver installation. Residual files from previous drivers cause most instability:

  1. Download DDU (Display Driver Uninstaller): download the latest version of DDU from the developer's website (Wagnardsoft). Extract it to a folder on your desktop
  2. Boot into Safe Mode: Settings → System → Recovery → Advanced startup → Restart now → Troubleshoot → Advanced options → Startup Settings → Restart → press 4 for Safe Mode
  3. Run DDU in Safe Mode: open DDU → select GPU → NVIDIA → click "Clean and restart." DDU removes all NVIDIA driver files, registry entries, and configuration. The computer restarts in normal mode
  4. Install the latest driver: download the latest NVIDIA driver from NVIDIA's website (not GeForce Experience). During installation, choose Custom → check "Perform a clean installation." Install only the Graphics Driver and PhysX — uncheck GeForce Experience, HD Audio, and USB-C if you do not need them
  5. Test by running the application or game that was crashing. If crashes stop, the previous driver installation was corrupted

Try an older stable driver

If the latest driver causes crashes, the issue may be a bug in the new driver version:

  1. Go to NVIDIA's driver archive page and download a driver version from 2-3 months ago. Newer is not always more stable
  2. Use DDU to remove the current driver (same Safe Mode procedure as above), then install the older version
  3. Check gaming forums and Reddit for reports about your specific GPU model and the current driver version. Known buggy driver versions are widely reported
  4. If an older driver is stable, you can stay on it. Disable automatic driver updates: Device Manager → Display adapters → right-click your GPU → Properties → Driver → check for a "Do not search for drivers" option, or disable driver delivery through Windows Update

Adjust power management settings

NVIDIA power management can cause crashes when the GPU rapidly changes power states:

  1. Open NVIDIA Control Panel (right-click desktop → NVIDIA Control Panel)
  2. Manage 3D settings → Global Settings → Power management mode → set to Prefer maximum performance. The default "Optimal power" or "Adaptive" mode aggressively downclocks the GPU to save power, and rapid clock transitions can trigger the TDR error
  3. In the same panel, set Threaded Optimization to Auto (or Off if crashes persist). Some applications conflict with NVIDIA's threaded optimization
  4. Windows power plan: Control Panel → Power Options → set to High Performance or Balanced. The Power saver plan limits GPU clocks
  5. For laptops: in NVIDIA Control Panel → Manage 3D settings → set the preferred GPU to your NVIDIA GPU for the crashing application instead of Auto. Optimus GPU switching during load can cause crashes

Increase the TDR timeout

TDR (Timeout Detection and Recovery) is Windows checking if the GPU is responding. If the GPU takes too long to complete an operation, Windows resets the driver. The default timeout is 2 seconds, which can be too short for heavy GPU loads:

  1. Open Registry Editor (Win+R → regedit)
  2. Navigate to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers
  3. Right-click in the right pane → New → DWORD (32-bit) Value → name it TdrDelay
  4. Double-click TdrDelay → set the value to 8 (this gives the GPU 8 seconds to respond instead of 2)
  5. Also create TdrDdiDelay with value 8
  6. Restart the computer. This does not fix the underlying cause but prevents Windows from killing the driver during heavy loads. If crashes stop, the GPU is overloaded but functional — consider reducing graphics settings in the affected application

Check GPU temperatures

Overheating causes the GPU to crash as a self-protection measure:

  1. Download HWiNFO64 or MSI Afterburner. Monitor the GPU temperature under load
  2. Normal gaming temperatures: 65-80°C. Thermal throttling begins around 83-90°C (varies by model). Above 90°C, crashes and shutdowns are expected
  3. If temperatures are too high:
    • Clean the GPU fans and heatsink of dust. Compressed air through the heatsink fins makes a significant difference
    • Check that all case fans are working and airflow is not blocked
    • Repaste the GPU: remove the cooler, clean old thermal paste, apply fresh paste. GPU thermal paste can dry out after 3-5 years
    • Check the fan curve: in MSI Afterburner or your GPU manufacturer's software, set a more aggressive fan curve to keep temperatures lower

Check GPU power delivery

Insufficient power causes driver crashes, especially under load:

  • Power supply: check that your PSU meets the GPU's recommended wattage. An NVIDIA RTX 3060 needs a 550W PSU; an RTX 4070 needs 650W; an RTX 4090 needs 850W. These are NVIDIA's recommendations, not theoretical minimums — an underpowered PSU causes intermittent crashes under peak load
  • Power cables: use separate PCIe power cables from the PSU to the GPU, not a single cable with daisy-chained connectors. Daisy-chaining limits the current available to the GPU
  • 12VHPWR connector: RTX 4000-series cards use a 12VHPWR or 12V-2x6 connector. Make sure it is fully seated — a loose connection causes intermittent power drops and crashes. Check the connector for any signs of melting or discoloration
  • Aging PSU: power supplies degrade over time. A PSU that met specifications 5 years ago may not deliver its rated wattage reliably today. If crashes happen only under heavy GPU load, the PSU is a suspect

Undo overclocking

If you have overclocked the GPU or it came with a factory overclock:

  1. Open MSI Afterburner or your GPU manufacturer's overclocking utility
  2. Reset all clocks to default: core clock offset to 0, memory clock offset to 0, power limit to 100%
  3. If the GPU is a factory-overclocked model (e.g., ASUS ROG Strix, MSI Gaming X), try underclocking slightly: set the core clock offset to -50 MHz. If crashes stop, the GPU's silicon is not stable at the factory overclock speed — this happens with silicon lottery variation
  4. Also reset any voltage adjustments to default. Overvoltage can damage the GPU over time, and undervoltage causes crashes under load

Check for Windows and system issues

Driver crashes can be triggered by Windows-level problems:

  1. Windows Update: make sure Windows is fully updated. Some driver crashes are fixed by Windows servicing stack updates that improve how Windows communicates with the GPU
  2. SFC and DISM: run sfc /scannow and DISM /Online /Cleanup-Image /RestoreHealth from an elevated command prompt. Corrupted system files can cause driver communication failures
  3. Check RAM: faulty RAM causes random crashes that can look like GPU driver failures. Run Windows Memory Diagnostic (search "Memory" in Start) or MemTest86 to test
  4. PCIe slot: if your motherboard has multiple PCIe x16 slots, try moving the GPU to a different slot. A damaged PCIe slot causes intermittent communication errors that trigger TDR crashes

Remote NVIDIA driver troubleshooting

If your NVIDIA driver keeps crashing and the fixes above have not resolved it, Marion Remote Fix can diagnose the issue remotely — performing clean driver installations, adjusting power and TDR settings, checking temperatures, and testing GPU stability through a secure remote session.

← Back to Blog