Detailed analysis
"Display driver stopped responding and has recovered" means your GPU froze for more than 2 seconds, so Windows killed the driver and restarted it. Your screen goes black, comes back a couple seconds later, and you get that notification bubble in the corner. The fastest fix: DDU clean install your GPU driver and remove any overclock. But that black-screen-and-recovery is a symptom, not the disease — this guide helps you find the actual root cause so it stops happening for good.
What's actually happening
Windows has a watchdog system called TDR — Timeout Detection and Recovery. Every 2 seconds (by default), Windows checks whether your GPU driver is still responding. If the driver doesn't respond within that timeout window, Windows assumes the driver has hung, kills it, and restarts it. Your screen goes black during the restart, then comes back when the fresh driver loads.
From your perspective: you're gaming, the screen flickers black for 1-3 seconds, then everything's back but your game has crashed (or is frozen/artifacting). The Windows notification tray shows "Display driver stopped responding and has recovered." In Event Viewer, this gets logged as Event ID
4101.
The 2-second default timeout is actually aggressive. Demanding modern games routinely push GPUs close to that limit during complex shader compilation, ray tracing passes, or when loading massive textures. A GPU that's running a tiny bit slow — from overheating, power throttling, or an unstable overclock — can tip over that 2-second edge.
The most common causes (in order of likelihood)
GPU driver bug or corruption. A bad driver installation is the most common cause of TDR events. This includes partially installed drivers, drivers corrupted by Windows Update, or a driver version with a known bug for your GPU model. A DDU clean install fixes this.
GPU overheating. When your GPU gets too hot, it thermal throttles — slowing down the clock speed to reduce heat. A GPU that's throttling can become so slow that it can't respond to Windows within the 2-second TDR window. NVIDIA cards start throttling at 83C and can become unstable approaching 90C. AMD cards have a "junction temperature" that throttles at 90-110C depending on the model. If your GPU is hitting these temps, the TDR is a heat symptom, not a driver problem.
Unstable GPU overclock. This includes overclocks you set yourself AND factory overclocks from card manufacturers. An overclock that passes benchmarks can still trigger TDR in games because games hit the GPU with unpredictable workload patterns that synthetic tests don't replicate. A GPU overclocked 50MHz too high might run fine for hours and then TDR during one specific scene.
VRAM exhaustion. When your GPU runs out of video memory, it swaps textures between VRAM and system RAM, which is extremely slow. If a frame requires data that's been swapped out, the GPU stalls waiting, and that stall can exceed the TDR timeout. Most likely on 4GB and 6GB cards in modern games, or 8GB cards with Chrome running in the background.
Insufficient power delivery. Your GPU needs stable, clean power. If your PSU is too weak, the PCIe power cables are daisy-chained (one cable split into two connectors instead of two separate cables), or your PSU is old and degrading, the GPU can brown out under heavy load. The voltage drops, the GPU can't sustain its clock speed, and it crashes — which triggers TDR.
Overlay and background software conflicts. Discord overlay, GeForce Experience ShadowPlay, Steam overlay, MSI Afterburner's RTSS, and OBS game capture all hook into the DirectX rendering pipeline. When two of these fight over the same hook, the GPU driver can hang.
How to fix it
- DDU clean install your GPU driver. Download DDU (Display Driver Uninstaller) from guru3d.com. Boot into Safe Mode (hold Shift while clicking Restart > Troubleshoot > Startup Settings > Safe Mode). Run DDU, select your GPU brand, click "Clean and restart." Once rebooted, immediately install the latest driver from nvidia.com or amd.com. Don't let Windows Update install a driver first. This single step fixes TDR for about 40-50% of people.
- Check GPU temperature under load. Download HWiNFO64 (free), run it in sensors-only mode. Play the game that causes TDR. If you're hitting 85C+ on NVIDIA or 100C+ junction on AMD, your GPU needs better cooling. Clean dust from the fans with compressed air. If the card is more than 2 years old, replacing thermal paste can drop temps 5-15C.
- Remove any GPU overclock and test. Open MSI Afterburner and click the reset button. If TDR stops at stock clocks, your overclock was unstable.
- Increase the TDR timeout (buys you time while troubleshooting). Open Registry Editor (Win+R > regedit). Navigate to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers. Right-click > New > DWORD (32-bit) Value. Name it "TdrDelay" and set the value to 10 (decimal). Create another DWORD called "TdrDdiDelay" set to
- Restart your PC. This gives the GPU 10 seconds instead of 2 to respond. Important: this doesn't fix the root cause — it just prevents Windows from killing the driver prematurely.
- Check your VRAM usage. Open Task Manager > Performance > GPU while gaming. If "Dedicated GPU Memory Usage" is above 90% of your total, lower texture quality, close Chrome, and turn off hardware acceleration in Discord (Settings > Advanced > Hardware Acceleration > Off).
- Check your power delivery. Use separate PCIe power cables from the PSU to your GPU — not one cable with two connectors daisy-chained. Each 8-pin connector should have its own dedicated cable. If your PSU is close to the GPU's minimum wattage recommendation or is more than 5 years old, it might not deliver clean power under heavy load.
- Disable overlay software one at a time. Turn off Discord overlay, GeForce Experience overlay, Steam overlay, and Xbox Game Bar. Test after disabling each one. When TDR stops, you've found the culprit.
Is this a hardware problem or a software problem?
Here's the diagnostic path: If a DDU clean install at stock GPU clocks with good temperatures (under 80C) fixes the TDR, it was software. Done.
If TDR persists after clean driver + stock clocks + confirmed good temps, you might have a hardware issue. Run OCCT's GPU test and VRAM test for 30 minutes. If OCCT crashes or shows errors, your GPU may have failing VRAM modules or a degrading GPU core. Test in another system if you can borrow one — this rules out PSU and motherboard issues.
Also check your PCIe slot: power off, remove the GPU, clean the gold contacts with isopropyl alcohol, and reseat it firmly. A loose GPU can cause intermittent TDR events that look like driver issues.
When to use the TDR registry fix vs. when NOT to
Use it when: your GPU is healthy but a specific game occasionally pushes past the 2-second limit during shader compilation. Unreal Engine 5 games (Fortnite, Palworld, ARK: Survival Ascended) have shader spikes that trigger TDR on perfectly healthy GPUs.
Don't use it when: you're getting TDR every 15-30 minutes. That means something is actually wrong. Increasing the timeout just means longer black screens before the crash. Fix the root cause.
Frequently asked questions
Q: Does "display driver stopped responding" mean my GPU is dying?
A: Usually no. The vast majority of TDR events are driver issues, overheating, or overclock instability. Hardware failure is possible but unlikely if this is a new problem. Try software fixes first.
Q: Why does this only happen in one specific game?
A: Different games stress GPUs in different ways. A game that maxes out shader units might trigger TDR while a game that's more memory-bound runs fine. Check that game's subreddit — there's often a known driver version that fixes TDR for that specific title.
Q: I increased TdrDelay but now my screen goes black for 10 seconds before recovering. Is that normal?
A: Yes. You told Windows to wait 10 seconds instead of
2. If your GPU is hanging for 5-8 seconds, you're masking the problem. Go back and address the root cause — temperature, driver, overclock, or power.
Q: This started happening after a Windows Update. Can I fix it without rolling back?
A: Try a DDU clean install of your GPU driver first — Windows Updates sometimes partially overwrite GPU driver files. If that doesn't work, uninstall the recent update: Settings > Windows Update > Update History > Uninstall updates.