Fix 'Graphics Driver Has Stopped Responding' on Windows 10/11

I know that message — "Display driver stopped responding and has recovered" — feels like a slap in the face. Your game freezes, then a black screen, then you're back to desktop. I've seen this on RTX 3080s, RX 6800s, and even integrated Intel graphics. The fix isn't always what you think.

Try This First: Increase the TDR Timeout

Windows has a built-in watchdog called TDR (Timeout, Detection, and Recovery). If your GPU takes longer than 2 seconds to respond (like during a heavy shader compile or a voltage spike), Windows kills the driver. The real fix? Give the GPU more time.

1. Open Registry Editor (Win+R, type regedit).
2. Go to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers.
3. Right-click on the right pane > New > DWORD (32-bit) Value.
4. Name it TdrDelay.
5. Double-click and set the value to 8 (decimal).
6. Create another DWORD named TdrDdiDelay and set it to 8 (decimal).
7. Close regedit and reboot.

This gives the driver 8 seconds to recover instead of 2. Doesn't fix every crash, but it stops the false positives where a game just needs a moment.

Why That Works (and When It Won't)

The TDR mechanism was designed to catch truly hung drivers. But modern games, especially Unreal Engine 5 titles like Hogwarts Legacy or Fortnite with Nanite, can spike GPU load beyond the 2-second limit. I've seen this on a clean Windows 11 install with fresh NVIDIA 537.13 drivers — the game loaded fine, but during the first lighting compilation, boom, TDR reset.

If the crash happens during gameplay (not just loading), the TDR delay might mask a deeper problem. Look for Event ID 4101 in Event Viewer under Windows Logs > System. That's your cue.

Less Common Fixes That Actually Work

1. Disable GPU Hardware Acceleration in Browser

I've seen Edge or Chrome trigger TDR crashes when you're watching YouTube while alt-tabbing out of a game. Go to browser settings > System > toggle "Use hardware acceleration when available" off. This stops the GPU from juggling multiple workloads.

2. Roll Back to a Stable Driver (Not the Latest)

Stop installing the newest driver the day it drops. For NVIDIA, version 537.13 (August 2023) is rock solid for most RTX 30-series cards. For AMD, version 23.9.2 or 23.12.1 are safe bets. Use DDU (Display Driver Uninstaller) in Safe Mode to wipe old drivers, then install the stable version.

3. Increase GPU Voltage (If You're Overclocked)

Factory overclocks or your own manual OC can cause voltage droop under load. Use MSI Afterburner: bump the core voltage slider to +25mV, or set a custom voltage curve. This stops the TDR crash on RTX 3070s with aggressive factory OC from ASUS or Gigabyte.

4. Check Your PSU's PCIe Cables

I once spent two weeks chasing a TDR crash on an RX 6700 XT only to find the daisy-chained PCIe power cable couldn't deliver enough current. Run two separate cables from the PSU to your GPU. If your PSU is older than 5 years, test with a known good unit.

Prevention for the Future

Once you've got things stable, stop the crashes from coming back. Keep your Windows updated (the 2024 Update fixed some scheduler bugs that stressed GPUs). Use a stable driver branch — not Game Ready, but Studio drivers for NVIDIA, or optional drivers for AMD only if needed. And if you're on a laptop, disable CPU turbo in power settings (set max processor state to 99%) — that stops the heat build-up that triggers driver crashes on thin-and-light gaming laptops.

That TDR timeout increase? It's a bandage. But for 80% of cases, it's the bandage that works. If you still crash after all this, you might have a failing GPU — run FurMark and check for artifacts. But start with the registry tweak. It's saved me more times than I can count.