Quick answer for advanced users
Disable TDR (Timeout Detection and Recovery) via registry, or increase the timeout value. This stops Windows from resetting the driver when a frame takes too long.
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers]
"TdrLevel"=dword:00000000
"TdrDelay"=dword:0000003cSet TdrLevel to 0 disables TDR entirely. Set TdrDelay to 60 (decimal) gives 60 seconds instead of 2. Reboot after either change.
Why this error happens and what it means
This error pops up when your graphics card takes longer than 2 seconds to respond to Windows. The OS thinks the driver is frozen and resets it. It's not a hardware failure nine times out of ten — it's a timing problem. I've seen this on everything from old GTX 970s to RTX 4090s, especially when running GPU-intensive games like Cyberpunk 2077 or rendering in Blender. The error shows a black screen for a second or two, then "Display driver has stopped responding and has recovered" in the system tray. The event log will show event ID 0x00000117 or 0x00000116. The real trigger is usually a driver that's hanging on a shader compilation or a power state transition that Windows doesn't wait for. Overclocking, overheating, and flaky power delivery make it worse, but the core issue is the timeout.
Fix 1: Clean install your graphics driver
Skip the "Update driver" button in Device Manager. It doesn't help here. You need a clean removal.
- Download Display Driver Uninstaller (DDU). It's free, well-tested, and I've used it for years.
- Boot Windows into Safe Mode (hold Shift while clicking Restart).
- Run DDU, select your GPU vendor (NVIDIA, AMD, Intel), and choose "Clean and restart."
- After reboot, download the latest driver from NVIDIA, AMD, or Intel. Not from Windows Update. Pick the "Studio" or "Game Ready" driver — whichever matches your workflow, but don't install the optional ones if you're not gaming.
- During installation, choose "Custom (Advanced)" and check "Perform a clean installation."
This wipes out leftover files from previous drivers that can cause conflicts. Does it always fix it? No. But it's the first thing to try because it eliminates the most common variable.
Fix 2: Increase the TDR timeout (most effective)
This is the fix that works when a clean driver install doesn't. Windows 10 and 11 let you change how long the OS waits before killing the driver. By default it's 2 seconds. That's too short for modern games with heavy shader compilation.
- Press Win+R, type
regedit, hit Enter. - Go to
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers. - Right-click in the right pane, choose New > DWORD (32-bit) Value. Name it
TdrDelay. - Double-click TdrDelay, set Base to Decimal, and enter
10(10 seconds). - Also create a DWORD named
TdrLevel, set to0if you want to disable TDR entirely, or leave it alone if you just changed the delay. - Close Regedit and restart your PC.
I've seen this fix stuttery games that crashed every 20 minutes. Setting TdrDelay to 10 gives the GPU breathing room for shader compilation without disabling the safety net entirely. If you're overclocking and getting crashes under load, set it to 30 seconds instead.
Fix 3: Disable GPU hardware acceleration in browser (if it's browser-related)
If the error happens only when watching YouTube, streaming video, or using Chrome/Edge, the GPU acceleration in the browser is the usual suspect.
- Chrome: Settings > System > Turn off "Use hardware acceleration when available."
- Edge: Settings > System and performance > Turn off "Use hardware acceleration when available."
- Firefox: Settings > General > Performance > Uncheck "Use recommended performance settings" and uncheck "Use hardware acceleration when available."
This moves video decoding back to the CPU. It uses more CPU but stops the driver from hanging on video decoding tasks that aren't optimized.
Alternative fixes if the above don't work
- Check your power supply. If your GPU draws more power than the PSU can deliver, the voltage drops and the driver gets confused. Use a tool like HWMonitor to check 12V rail voltage under load. Below 11.6V is bad.
- Lower your GPU overclock. If you've overclocked, reset to stock. Even factory-overclocked cards can be unstable at certain boost frequencies. Drop the core clock by 50 MHz and memory by 100 MHz, test again.
- Update your motherboard chipset drivers. This is often overlooked. A stale chipset driver can cause PCIe link issues. Go to your motherboard manufacturer's site and grab the latest chipset driver.
- Disable Windows Game Mode. Open Settings > Gaming > Game Mode and turn it off. Game Mode can mess with GPU scheduling on some systems.
Prevention tip to avoid this happening again
Once you've fixed it, don't install every optional driver update that pops up. Stick to stable Game Ready or Studio drivers. Use DDU every time you upgrade the driver, not just when something breaks. Also, keep your system clean of dust — a GPU running at 85°C is more likely to hit TDR than one at 65°C. And if you're on a laptop, don't play games on battery. The GPU will throttle and the driver will timeout faster.