Fix Display Driver Stopped Responding and Has Recovered on Windows 11

The screen goes black for a second, comes back, and a notification reads "Display driver stopped responding and has recovered." Sometimes the game closes, sometimes the whole PC freezes. This is TDR, Timeout Detection and Recovery, a Windows safety feature that resets the GPU driver (nvlddmkm.sys on NVIDIA cards) when it does not respond in time. TDR is the symptom, not the disease. The real cause is one of three things: a bad driver, a thermal or power limit, or unstable settings. Here is how to tell them apart.

Key takeaways

• TDR is Windows resetting a hung GPU driver, so something made the GPU stop responding, it did not fail on its own.
• A clean driver reinstall with DDU clears the most common cause: a corrupted or half-installed driver.
• Heat and power are the next suspects, especially crashes that only happen under load.
• HAGS (Hardware-Accelerated GPU Scheduling) causes TDR resets on some configurations; toggling it is a quick test.
• Persistent crashes after all of the above point to a failing card or unstable overclock.

Use the pattern of when it crashes to jump to the most likely cause:

When it crashes	Likely cause	First fix
Randomly, on the desktop or any app	Corrupted or half-installed driver	Clean reinstall with DDU
After a recent Windows feature update	HAGS scheduling conflict	Turn off Hardware-Accelerated GPU Scheduling
Only under load (gaming, rendering)	Overheating or power sag	Check temps, set high-performance power, test PSU
Only in one game, right after an update	Shader or engine bug	Verify game files, update game and driver
With on-screen artifacts before the reset	Failing card or unstable overclock	Reset to stock clocks, test card elsewhere

Step 1: Clean-reinstall the GPU driver with DDU

A half-installed or corrupted driver is the leading cause of TDR. A normal reinstall leaves the bad files in place; Display Driver Uninstaller (DDU) removes everything.

1. Download the latest driver for your exact GPU from NVIDIA, AMD, or Intel, but do not install it yet.
2. Download DDU.
3. Boot into Safe Mode (Win + R, msconfig, Boot tab, tick Safe boot, restart).
4. Run DDU and choose Clean and restart.
5. Back in normal mode, install the driver you downloaded.

If TDR stops, a corrupted driver was the cause. If it continues, the problem is heat, power, or settings.

Step 2: Disable Hardware-Accelerated GPU Scheduling

HAGS shifts scheduling work to the GPU and is a documented TDR trigger on some systems.

Go to Settings > System > Display > Graphics > Change default graphics settings and turn off Hardware-Accelerated GPU Scheduling, then restart. Test for a day. If crashes stop, leave it off.

Step 3: Rule out heat and power

If the resets only happen under load (gaming, rendering, video decode), the GPU is likely throttling or starving.

• Monitor temperatures with HWiNFO or the GPU vendor's app. Sustained temps near or above the card's limit mean a cooling problem. Clean dust from the fans and heatsink and confirm every fan spins.
• Set Windows to Best performance in Settings > System > Power and select High performance in Control Panel power options so the GPU is not forced into aggressive idle states mid-load.
• On a desktop, a marginal or aging power supply can sag under GPU spikes and trigger TDR. If the card is new or power-hungry, this is worth ruling out.

Step 4: Reset any overclock and check the cable

Factory or user overclocks that were stable last year can become unstable as a card ages.

• In MSI Afterburner or your tuning tool, set the GPU to stock clocks and voltage. Test. If stable, your overclock was the cause; dial it back in slowly.
• Reseat the GPU and its power connectors, and try a different DisplayPort or HDMI cable. A flaky cable can cause output drops that look like TDR.

Step 5: Increase the TDR delay (last-resort workaround)

This does not fix the cause but can keep heavy workloads from tripping the watchdog while you investigate. Only do this if you are comfortable editing the registry.

In regedit, go to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers, create a DWORD (32-bit) named TdrDelay, and set it to a decimal value such as 8 (seconds). Restart.

Frequently asked questions

What does nvlddmkm.sys mean in the error?

nvlddmkm.sys is the NVIDIA kernel-mode display driver. When it appears in a TDR event or a VIDEO_TDR_FAILURE blue screen, the NVIDIA driver is the component that stopped responding. A DDU clean reinstall is the right first move.

Why does it only crash in one game?

A single-title crash often means a shader-compilation or engine bug rather than a hardware fault, especially right after a game or driver update. Update both, verify game files, and lower in-game settings to test. Our guide to shader compilation stutter covers related GPU-load symptoms.

Could bad RAM or PSU cause this rather than the GPU?

Yes. Unstable system memory and a sagging power supply both produce load-dependent GPU resets. If a DDU reinstall and thermal checks do not help, test RAM with MemTest86 and consider the PSU, particularly on a desktop with a high-draw card.

Is my graphics card dying?

Possibly, if TDR continues after a clean driver, good temperatures, stock clocks, and a known-good cable. Artifacts (strange textures or colors) alongside the resets strengthen that conclusion. Test the card in another system if you can.

Quick recap

Clean-reinstall the driver with DDU first, then toggle HAGS off. If crashes only happen under load, check temperatures, set high-performance power, and reset any overclock. Use a higher TdrDelay only to diagnose, and suspect a failing card if everything else checks out.