Detailed analysis
VIDEO_TDR_FAILURE (0x116) is the single most common BSOD for gamers. TDR stands for Timeout Detection and Recovery — Windows gives the GPU driver 2 seconds to respond, and when it doesn't, Windows kills the driver and you get this blue screen. The faulting driver is almost always nvlddmkm.sys (NVIDIA) or atikmpag.sys (AMD).
What it means in plain English: Your GPU stopped talking to Windows. The graphics driver either crashed, froze, or was too overloaded to respond in time. This is NOT a sign your GPU is dying — in the vast majority of cases it's a software or configuration issue.
The most common causes (in order)
- Outdated or buggy GPU driver. This is the #1 cause. NVIDIA and AMD release frequent driver updates that fix TDR crashes for specific games. A driver even 1-2 months old can be missing critical stability patches. Always DDU clean install — don't just 'update' over the old driver.
- GPU overheating. When the GPU hits its thermal limit (usually 83-90C depending on the card), it throttles aggressively. If throttling can't cool it fast enough, the driver times out. Open HWiNFO64 and monitor GPU temps while gaming. Anything over 85C sustained is a problem — clean dust from fans, improve case airflow, or repaste the GPU thermal compound.
- Overlay software conflicts. Running multiple overlays simultaneously is a classic TDR trigger. Discord overlay, GeForce Experience overlay, Steam overlay, MSI Afterburner OSD, and Xbox Game Bar all hook into DirectX. When two overlays fight over the same DX hook, the driver can freeze. Disable all overlays and re-enable one at a time.
- Unstable GPU overclock. An overclock that passes FurMark can still crash in games because games create different GPU load patterns. This includes memory overclock — unstable VRAM clocks cause TDR constantly. Reset MSI Afterburner to stock and test.
- Insufficient PSU power. When your GPU demands more wattage than the PSU can deliver (especially during load spikes), the driver can crash. Minimum PSU requirements: RTX 3070 = 650W, RTX 3080 = 750W, RTX 4070 Ti = 700W, RTX 4080 = 750W, RTX 4090 = 850W, RX 7900 XTX = 800W.
How to fix it
- DDU clean install your GPU driver. Download DDU (Display Driver Uninstaller) from guru3d.com. Boot into Safe Mode (hold Shift + Restart > Troubleshoot > Startup Settings > Safe Mode). Run DDU, select 'Clean and restart.' Install the latest driver from nvidia.com or amd.com immediately after reboot.
- Monitor GPU temperature. Download HWiNFO64 and run it in sensors-only mode while gaming. Watch 'GPU Temperature' — if it exceeds 85C, clean the GPU fans, improve case airflow, or repaste with quality thermal paste (Thermal Grizzly Kryonaut or Noctua NT-H1).
- Disable ALL overlays. GeForce Experience: Settings > In-Game Overlay > Off. Discord: User Settings > Game Overlay > Off. Steam: Steam > Settings > In-Game > uncheck 'Enable Steam Overlay.' Xbox Game Bar: Windows Settings > Gaming > Xbox Game Bar > Off.
- Remove GPU overclock. Open MSI Afterburner and click the reset button. Also reset the memory clock — unstable VRAM overclock is more likely to cause TDR than core clock instability.
- Check Event Viewer for pattern clues. Open Event Viewer > Windows Logs > System. Look for Event ID 4101 ('Display driver stopped responding and has successfully recovered') — if you see many of these before the BSOD, the driver has been struggling for a while.
- Last resort: increase TDR timeout. Open Registry Editor, navigate to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers. Create a DWORD value called 'TdrDelay' and set it to 8 (seconds). This gives the GPU more time to respond, which can fix TDR crashes caused by brief driver stalls during shader compilation.
When to suspect hardware failure: If DDU clean install + stock clocks + all overlays disabled still crashes, test the GPU in another PC or test with a different GPU in your PC. If the crash follows the GPU, it may be a hardware issue (failing VRAM, degraded die). Run OCCT's VRAM test to check for VRAM errors.