The screen freezes for a few seconds, goes black, and comes back with a notification: the display driver stopped responding and has recovered. On Windows 10 this Timeout Detection and Recovery (TDR) event is the operating system’s way of avoiding a blue screen when the GPU stops answering — it restarts the graphics driver mid-flight and hopes for the best. Once in a while it is a hiccup; every day, it is a problem demanding a fix.

TDR triggers fall into three buckets. The driver bucket covers corrupted, mismatched, or outdated graphics drivers that mishandle commands. The environment bucket covers overheating GPUs, unstable overclocks, and power delivery problems that make the GPU stumble. The hardware bucket — the smallest but the most serious — covers a GPU that is actually failing. The order of investigation follows the odds: driver first, environment second, hardware last.

Below is the technician’s sequence for ending TDR crashes on Windows 10: clean driver reinstallation, timeout tuning, thermal and power checks, and the stress tests that give a final verdict on the hardware.

Why this happens

  • Corrupted or mismatched graphics driver: The most common trigger. A driver that miscompiles shaders or mishandles power states makes the GPU hang, tripping the TDR watchdog.
  • GPU overheating: Laptop GPUs that exceed safe temperatures throttle and then hang. Dust-clogged vents and dried thermal paste are the usual culprits on older machines.
  • Unstable overclock or undervolt: Factory overclocks on gaming laptops and user-applied tweaks can push the GPU past stability, producing TDR crashes under load but clean idle behavior.
  • Insufficient or unstable power: A failing charger or power circuit causes voltage droop under GPU load. The GPU hangs, Windows restarts the driver, and the cycle repeats.
  • TDR timeout too aggressive for the workload: The default two-second timeout can be too short for heavy scenes on modest laptop GPUs, tripping recovery on a GPU that would have finished the frame.
  • Conflicting background software: Overlays, screen recorders, and aggressive antivirus hooks can interfere with the driver’s command stream.
  • Failing GPU hardware: Degrading VRAM or a failing GPU core produces TDR crashes that survive every software fix and worsen over time.

How to fix it

Work through these in order — diagnose first, then act.

Fix 1: Clean-reinstall the graphics driver

The first and most effective step. A standard update over the old driver often preserves the corruption causing the crashes.

  1. Download the newest driver for your GPU from the vendor — NVIDIA, AMD, or Intel.
  2. Use Display Driver Uninstaller from Safe Mode with the Clean and restart option for a thorough removal.
  3. Install the fresh driver and restart. Test with the application that crashed most reliably.

Fix 2: Adjust the TDR timeout as a diagnostic

Lengthening the timeout distinguishes a slow GPU from a broken one. This is a diagnostic step, not a permanent fix for hardware issues.

  1. Open Registry Editor and navigate to HKEY_LOCAL_MACHINE, then SYSTEM, CurrentControlSet, Control, GraphicsDrivers.
  2. Create a QWORD value named TdrDelay set to 8 (seconds) and a QWORD TdrDdiDelay set to 10 if they do not exist.
  3. Restart and test. If crashes stop, the GPU was borderline-slow; still investigate heat and drivers, since a healthy GPU rarely needs this.

Fix 3: Fix GPU thermals

Heat is the second most common cause. Laptop cooling degrades silently over years.

  1. Monitor GPU temperatures during load with HWMonitor or similar — sustained readings above 90C under load are concerning.
  2. Blast dust from vents with compressed air and ensure the laptop sits on a hard flat surface.
  3. Consider a cooling pad and, on older machines, a professional repaste. Retest the crashing workload afterward.

Fix 4: Remove overclocks and test at stock

Stability problems that vanish at stock clocks indict the overclock, not the driver.

  1. Reset any GPU overclock or undervolt to factory defaults in Afterburner or the vendor utility.
  2. Disable automatic overclocking features in the laptop’s gaming software.
  3. Run the previously crashing application for an extended session. If TDR crashes persist at stock, continue down the list.

Fix 5: Rule out power delivery problems

Voltage droop under load mimics driver crashes perfectly. Test the power path before condemning the GPU.

  1. Use the original manufacturer charger with the correct wattage — underpowered chargers cause GPU instability under load.
  2. Test with the battery removed if the model allows, and with the battery in, to isolate battery-related droop.
  3. Try a different wall outlet to rule out mains issues.

Fix 6: Eliminate software conflicts

Background software that hooks the graphics pipeline can trip TDR on its own.

  1. Disable overlays: Discord, GeForce Experience, Xbox Game Bar, and similar.
  2. Temporarily disable screen recording and streaming software.
  3. Perform a clean boot and test. If crashes stop, re-enable programs one group at a time to find the culprit. For the newer-OS variant of these crashes, see TDR fixes on Windows 11 , and the 24H2-specific notes if the crashes began with that update.

Fix 7: Stress-test for a hardware verdict

If software is exhausted, controlled stress testing gives the final answer on the GPU’s health.

  1. Run a GPU stress test like FurMark while monitoring temperatures.
  2. Crashes within minutes at safe temperatures point to hardware failure.
  3. Artifacts — colored blocks, lines, texture corruption — alongside TDR crashes are the classic signature of dying VRAM. At that point, professional repair or replacement is the path; Microsoft’s Windows support pages document the built-in memory and hardware diagnostics for the remaining checks.

Closing

TDR crashes on Windows 10 are the GPU crying for help, and the help it usually needs is a clean driver, better cooling, or stock clocks — not a new laptop. Work the sequence in order and let the stress test deliver the verdict if software fixes fail. Most users never reach the hardware step, because the driver reinstall or the thermal cleanup ends the crashes first.

Frequently asked questions

What does display driver stopped responding and has recovered mean?
It means Windows detected the GPU had stopped answering within its timeout window and restarted the graphics driver to avoid a full system crash — a mechanism called TDR, Timeout Detection and Recovery. The screen flickers or freezes briefly, then comes back. Occasional recoveries are survivable; frequent ones signal a driver, heat, or hardware problem that needs fixing.
Is the TDR error caused by a bad graphics driver?
Very often, yes. Corrupted or mismatched graphics drivers are the most common trigger, because the driver is what translates Windows’ commands for the GPU. A clean reinstall of the correct driver resolves a large share of TDR crashes. Heat and failing hardware are the next most common causes when drivers check out.
Can increasing the TDR timeout fix the crashes?
Increasing the TdrDelay registry value gives the GPU more time before Windows declares it hung, which helps borderline cases where the GPU is slow but not broken — heavy scenes, slow laptop GPUs. It is a workaround rather than a cure: if the underlying cause is heat or a dying GPU, the longer timeout just delays the inevitable crash.
Does this error mean my GPU is dying?
Not necessarily, but persistent TDR crashes after driver reinstalls, cooling fixes, and underclocking point toward hardware. The distinguishing test is whether the crashes happen in all 3D applications and even at the desktop under load. GPU-specific stress tests and temperature monitoring separate heat problems from silicon problems.