'Display Driver Stopped Responding and Has Recovered'? How to Stop TDR Crashes

The screen goes black for a second, flickers back, and a notification reads "Display driver stopped responding and has recovered." Sometimes a game crashes with it; sometimes the whole desktop freezes.

This is Windows' Timeout Detection and Recovery (TDR) feature at work: when the graphics card fails to respond within about two seconds, Windows assumes it has hung and resets the driver to keep the system alive. The reset is the symptom, not the disease. The real cause is usually a buggy driver, an overheating or unstable GPU, or excessive load, and this guide isolates which one is hitting you.

Key takeaways

• TDR is Windows resetting a GPU that stopped responding, the fix is finding why it hangs.
• A buggy or outdated graphics driver is the leading cause; a clean reinstall with DDU beats an in-place update.
• Overheating, dust, and an unstable overclock are the next most common triggers.
• Raising the TDR timeout in the registry is a workaround for borderline cases, not a root-cause fix.

Step 1: Note when it happens

When the reset fires tells you a lot. Use this to skip straight to the likely fix:

When it happens	Likely cause	Where to start
Under heavy game or render load	Heat, power sag, or unstable overclock	Remove overclock, check temps and PSU
At idle or on the desktop	Driver bug or failing card	Clean-reinstall driver with DDU
Right after a driver update	Driver regression	Roll back to previous stable version
During GPU compute (rendering, AI)	Task legitimately exceeds 2s timeout	Raise TdrDelay as a workaround
Only in one specific game	Game or shader cache issue	Verify game files, clear shader cache

The trigger narrows the cause:

• Under heavy load (games, rendering)? Suspect heat, power, or an unstable overclock.
• At idle or on the desktop? Suspect a driver bug or a failing card.
• Right after a driver update? The new driver is the prime suspect, roll it back.

Fix 1: Clean-reinstall the graphics driver

An in-place update leaves old files behind that can keep causing hangs. A clean install with Display Driver Uninstaller (DDU) removes everything first.

1. Download the latest driver for your exact GPU from NVIDIA, AMD, or Intel, do not rely only on Windows Update.
2. Download Display Driver Uninstaller (DDU).
3. Boot into Safe Mode (hold Shift while clicking Restart, then Troubleshoot > Advanced options > Startup Settings > Restart, press 4).
4. Run DDU and choose Clean and restart.
5. Back in normal mode, install the driver you downloaded.

If the crashes only started after a recent driver, install the previous stable version instead of the newest, a regression in a fresh release is a common cause.

Fix 2: Fix overheating

A GPU that overheats throttles or hangs, which trips TDR.

• Monitor temperatures with a tool like HWiNFO or GPU-Z while gaming. Sustained temps above 85 C are a red flag.
• Clean the dust from the card's fans and heatsink with short bursts of compressed air.
• Improve case airflow and make sure intake and exhaust fans are working.
• On an older card, the thermal paste may have dried out, repasting can drop temperatures significantly.

This is the same heat-driven failure mode behind in-game stutter and frame-time spikes, a throttling GPU shows up as both micro-stutter and, at the extreme, full TDR resets.

Fix 3: Check power delivery

An underpowered or failing supply causes the GPU to drop out under load.

• Confirm both PCIe power connectors are firmly seated, not partially clicked in.
• If you recently added components, verify your PSU wattage still has headroom for the whole system.
• A degrading power supply can deliver clean power at idle but sag under a gaming load, exactly when TDR fires.

Fix 4: Reduce GPU load

If the card is simply being pushed past its limit, easing off stops the hangs.

• Lower in-game settings, especially resolution scale, shadows, and ray tracing.
• Cap the frame rate (in-game limiter or driver control panel) to keep the GPU from running flat out.
• Close background apps that use the GPU, browsers with hardware acceleration, recording software, and overlays all add load.

Fix 5: Increase the TDR timeout (workaround)

If a heavy compute or rendering task legitimately needs more than two seconds, raising the timeout prevents an unnecessary reset. This is a workaround, not a cure.

1. Press Win + R, type regedit, and press Enter.
2. Navigate to:

   HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers
   

3. Create a new DWORD (32-bit) value named TdrDelay.
4. Set its value to 8 (seconds).
5. Restart the PC.

Fix 6: Repair system files and test the hardware

If nothing above helps, rule out OS corruption and a failing card.

DISM /Online /Cleanup-Image /RestoreHealth
sfc /scannow

Then run a stress test such as FurMark briefly while watching temperatures and stability. A card that resets within seconds at stock clocks and safe temperatures may be physically failing and need replacement or warranty service.

If your crashes are tied to one game stuttering rather than full TDR resets, the cause may be a shader cache problem instead; our guide to shader compilation stutter covers that specific case.

What to do right now

Work this checklist top to bottom and stop when the resets stop:

• Set the GPU to stock clocks and disable any factory or manual overclock, then test.
• Clean-reinstall the graphics driver with DDU in Safe Mode, using a driver downloaded from the vendor.
• Monitor temps with HWiNFO under load; if you see sustained 85 C or higher, clean dust and improve airflow.
• Reseat both PCIe power connectors and confirm your PSU has wattage headroom.
• Cap your frame rate and lower the heaviest settings (resolution scale, ray tracing) to reduce load.
• Only if it still resets, raise TdrDelay to 8 seconds as a workaround, and run FurMark plus MemTest to check for failing hardware.

Frequently asked questions

What does TDR actually do?

Timeout Detection and Recovery is a Windows feature that watches the GPU. If the card does not respond within roughly two seconds, Windows assumes it has hung and resets the driver to recover the desktop instead of forcing a full crash. The "stopped responding and has recovered" message is that reset completing.

Will increasing TdrDelay fix it permanently?

No. Raising the timeout only gives the GPU more time before Windows resets it. That helps when a heavy task legitimately needs a few extra seconds, but it masks a real hang from a driver bug, overheating, or unstable hardware. Fix the underlying cause first.

Why did this start right after a driver update?

A buggy or regressed driver release is a frequent TDR cause. Roll back to the previous stable version, or clean-install it with DDU. If the newest driver consistently crashes, stay on the last known-good version until the vendor ships a fix.

Could it be the graphics card failing?

Yes, if the resets persist at stock clocks, safe temperatures, and with a clean driver. A card that hangs under any load despite all software fixes may be physically degrading. Test it in another system if you can, and pursue warranty or replacement.

Quick recap

Note when the resets happen, remove any overclock, then clean-reinstall the driver with DDU. Address overheating and power delivery, reduce GPU load, and only raise the TdrDelay as a last-resort workaround. If the card keeps resetting at stock settings and safe temps, suspect failing hardware.