Fixes / Windows · Checked

Windows VIDEO_TDR_FAILURE: what it means and how to fix it

Short answer

VIDEO_TDR_FAILURE (stop code 0x116) means your graphics card stopped responding. Windows tried to reset the display driver, the reset failed, and Windows restarted the PC. The file shown under "What failed" names the graphics driver that crashed: nvlddmkm.sys (NVIDIA), atikmpag.sys or amdkmdag.sys (AMD), or igdkmd64.sys (Intel). The most likely fix is a clean reinstall of the latest driver from that graphics maker.

What this message means

Windows has a safety feature called Timeout Detection and Recovery (TDR). If the graphics card (GPU) doesn't finish a task within 2 seconds, Windows resets the display driver. Usually this works and the screen just flickers, sometimes with the message "Display driver stopped responding and has recovered." You get the VIDEO_TDR_FAILURE blue screen (bug check 0x116) when that reset fails, or when the GPU hangs more than five times within one minute. It is almost always tied to the graphics driver or the graphics hardware (its power supply, heat, or overclock). A normal app bug rarely causes it.

Common causes

CauseHow to tell
Corrupted, outdated or badly installed graphics driverThe blue screen lists nvlddmkm.sys, atikmpag.sys, amdkmdag.sys or igdkmd64.sys under "What failed". The crashes started after a driver update, a Windows update, or switching graphics cards.
GPU or system overclock, including factory overclocks and undervolts that aren't fully stableCrashes happen only under heavy load (games, video export, 3D rendering) and stop when you reset clocks to stock in MSI Afterburner, AMD Software or the BIOS (XMP/EXPO memory profiles count as overclocks too).
Overheating graphics card or laptopGPU temperature is close to its limit (often 83 to 90°C or higher) just before the crash. The fans are loud or clogged with dust, and crashes come after a period of gaming rather than right away.
Not enough power to the graphics card (weak or failing power supply, loose PCIe power cable, aggressive PCIe power saving)The PC crashes on sudden load spikes, such as loading a game level or starting a render. The power supply is old or close to the GPU maker's recommended wattage, or reseating the power cables temporarily helps.
Faulty RAM or a failing graphics cardCrashes continue after a clean driver install at stock clocks. Windows Memory Diagnostic reports errors, or you see artifacts, flickering or colored blocks on screen before the crash.

How to fix it

1. Find out which driver failed

  1. On the blue screen (or after restarting), note the file listed under "What failed": nvlddmkm.sys = NVIDIA, atikmpag.sys or amdkmdag.sys = AMD, igdkmd64.sys = Intel.
  2. If you missed it, press Windows key + R, type eventvwr.msc and press Enter.
  3. Go to Windows Logs > System and look for "Display driver ... stopped responding and has successfully recovered" warnings (Event ID 4101) shortly before the crash. The driver name tells you the graphics maker.
  4. On laptops with two graphics chips, this tells you whether the Intel/AMD integrated chip or the NVIDIA/AMD dedicated card is the one crashing. Fix that one first.

2. Clean reinstall of the graphics driver

  1. Download the latest driver for your exact graphics card and Windows version directly from NVIDIA, AMD or Intel. Laptop users can also check the laptop maker's support page.
  2. NVIDIA: run the installer, choose Custom install and tick "Perform a clean installation". AMD: in the AMD Software installer, choose the Factory Reset option. Intel: uninstall the old driver first in Settings > Apps > Installed apps (Windows 11) or Apps & features (Windows 10).
  3. If a normal clean install doesn't help, the widely used alternative is Display Driver Uninstaller (DDU) run in Safe Mode, then installing the fresh driver. Disconnect from the internet while doing this so Windows Update doesn't install its own driver in between.
  4. Restart and run the game or app that crashed before.

3. Roll back a driver that started the crashes

  1. Right-click Start and select Device Manager.
  2. Expand Display adapters, right-click your graphics card and choose Properties.
  3. Open the Driver tab and click Roll Back Driver. If it's greyed out, install an older version manually from the maker's website instead.
  4. Pick a reason, click Yes, then restart.
  5. If the old driver is stable, wait for the next driver release before updating again.

4. Remove overclocks and check temperatures

  1. Reset GPU clock, memory clock and voltage to default in the tool you used (MSI Afterburner, AMD Software > Performance > Tuning, NVIDIA app performance tuning).
  2. In the BIOS/UEFI, disable XMP or EXPO temporarily and load optimized defaults if you changed CPU or memory settings.
  3. Watch the GPU temperature while gaming. On Windows 10 and 11, Task Manager > Performance > GPU shows it for dedicated cards.
  4. Clean dust from the GPU fans and case filters. On laptops, use a flat hard surface and check the maker's cooling settings.
  5. If crashes stop at stock settings, your overclock (or factory OC) is unstable. Stay at stock, or slightly lower the clocks or power limit.

5. Turn off PCI Express power saving

  1. Press Windows key + R, type control and press Enter to open Control Panel.
  2. Go to Hardware and Sound > Power Options > Change plan settings (for your active plan) > Change advanced power settings.
  3. Expand PCI Express > Link State Power Management and set it to Off (for both On battery and Plugged in on laptops).
  4. Click Apply, then OK. On some Windows 11 laptops that use Modern Standby, this option is hidden. Skip this fix if you can't see it.

6. Test the memory and power hardware

  1. Press Windows key + R, type mdsched.exe, press Enter and choose Restart now and check for problems.
  2. After it finishes, check the result in Event Viewer > Windows Logs > System (source MemoryDiagnostics-Results). Any error means a RAM stick needs replacing.
  3. On desktops, power off, then reseat the graphics card and its PCIe power cables. Use separate cables from the power supply for each 8-pin connector rather than one daisy-chained cable.
  4. Compare your power supply's wattage with the graphics card maker's recommendation. If you can, test with a different power supply or put the card in another PC.
  5. If the card still crashes at stock settings on a known-good system, contact the card maker for warranty service.

Next time, skip the search

Crade is an AI assistant for Mac and Windows that sees your screen. When a message like this pops up, ask “what does this mean?” and it explains the exact error in front of you, step by step.

Download Crade

Still not fixed?

If the crash continues after a clean driver install, stock clocks and a clean memory test, the graphics card, power supply or motherboard is the likely cause. Collect the crash dump files from C:\Windows\Minidump and post them on Microsoft Q&A or a hardware forum. Anyone with WinDbg can run "!analyze -v" on them to confirm the faulting module. If the PC is under warranty, contact the PC or graphics card maker and list the steps you've already tried. If the crashes began right after a Windows update, restore to an earlier restore point from Settings > System > Recovery (Windows 11) or Settings > Update & Security > Recovery (Windows 10).

Questions people ask

Is VIDEO_TDR_FAILURE a hardware or software problem?

Usually the graphics driver causes it, so start with a clean driver reinstall. If it continues at stock settings with a fresh driver, suspect hardware: overheating, a weak power supply, bad RAM or a failing graphics card.

What is nvlddmkm.sys and why does it cause a blue screen?

nvlddmkm.sys is the NVIDIA graphics driver. It shows up on the blue screen because it's the driver that stopped responding, which doesn't always mean the driver itself is faulty. Heat, an overclock or a power problem can make any GPU driver time out.

Can VIDEO_TDR_FAILURE damage my PC?

The blue screen itself does no damage. It's Windows stopping safely. You can lose unsaved work, though, and frequent crashes caused by heat or power problems should be fixed before they get worse.

Should I increase TdrDelay in the registry?

Microsoft documents TdrDelay and the other TDR registry values as settings for driver developers only and says end users shouldn't change them. Raising the timeout hides the symptom, and a PC that really hangs just freezes for longer instead.

Why does it only happen when I play games or edit video?

The timeout triggers when the GPU is too busy to respond within 2 seconds, which almost only happens under heavy load. That's also when heat, power draw and overclock instability are at their worst.

Sources

Related errors