DXGI_ERROR_DEVICE_HUNG Fix for All Games

What is DXGI_ERROR_DEVICE_HUNG?

DXGI_ERROR_DEVICE_HUNG (error code 0x887A0006) means your GPU failed to complete a DirectX command within the Windows TDR (Timeout Detection and Recovery) window. The GPU effectively froze, and Windows reset the display driver. You see messages like *"DirectX function 'GetDeviceRemovedReason' failed with DXGI_ERROR_DEVICE_HUNG"* or the Unreal Engine 5 variant: *"LowLevelFatalError [File:Unknown] [Line: 686] DXGI_ERROR_DEVICE_HUNG."*

This error affects NVIDIA, AMD, and Intel GPUs across dozens of games. It is especially aggressive in Unreal Engine 5 and RE Engine titles because their DX12 code paths are highly sensitive to marginal GPU clock stability. Factory-overclocked GPUs are the most frequent offenders.

When does it occur?

• During gameplay in DX12 games — especially with ray tracing or DLSS/FSR enabled
• In Unreal Engine 5 games: ARC Raiders, Marvel Rivals, THE FINALS, Fortnite
• In RE Engine games: Monster Hunter Wilds (extremely widespread)
• After GPU driver updates or Windows feature updates
• On factory-overclocked GPUs (partner models like STRIX OC, FTW3, etc.)
• When Hardware-Accelerated GPU Scheduling (HAGS) is enabled

Common causes

• Factory GPU overclock exceeding stable limits — partner cards ship above reference clocks, causing hangs under demanding DX12 workloads
• Outdated or corrupt GPU drivers — stale driver installations accumulate conflicts
• Windows TDR timeout too short — default 2-second timeout is insufficient for heavy DX12 compute shaders
• Hardware-Accelerated GPU Scheduling (HAGS) — can destabilize frame scheduling on some GPU/driver combinations
• Corrupt DirectX shader cache — stale compiled shaders cause GPU hangs
• DirectX 12 rendering path bugs — some game engines have unstable DX12 implementations
• Insufficient GPU power delivery or thermal throttling — power spikes or overheating cause the GPU to stop responding

Step-by-step fixes

1. Clean install GPU drivers with DDU — Download DDU from guru3d.com. Boot into Safe Mode. Run DDU, select your GPU vendor, click "Clean and Restart." Install the latest driver from NVIDIA/AMD/Intel with a clean install option. This is the single most effective fix across all GPU vendors.

1. Underclock GPU by 50-100 MHz — Open MSI Afterburner. Reduce Core Clock by -50 MHz. If crashes persist, reduce to -100 MHz. For Monster Hunter Wilds, users report needing -300 MHz. For NVIDIA users: a quicker alternative is NVIDIA Control Panel > Help > Debug Mode, which forces reference clocks.

1. Increase TDR timeout in registry — Open Registry Editor (regedit). Navigate to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers. Create a new DWORD (32-bit) named TdrDelay and set it to 8 (decimal). Create another DWORD named TdrDdiDelay, set to 60. Restart your PC. This gives the GPU more time to respond before Windows declares it hung.

1. Disable Hardware-Accelerated GPU Scheduling — Go to Settings > System > Display > Graphics > Change default graphics settings. Toggle OFF "Hardware-accelerated GPU scheduling." Restart. Note: this also disables DLSS/FSR Frame Generation.

1. Force DirectX 11 — On Steam: right-click game > Properties > Launch Options > enter -dx11. On Epic: Settings > scroll to game > Additional Command Line Arguments > enter -dx11. Not all games support this — DX12-only titles will ignore it.

1. Clear DirectX shader cache — Go to Settings > System > Storage > Temporary files. Check "DirectX Shader Cache" and click Remove files. This forces the GPU to recompile shaders cleanly on next launch.

1. Set per-game GPU to High Performance — Settings > System > Display > Graphics > Browse, add the game's .exe. Select it, click Options, choose "High performance." This prevents Windows from accidentally routing the game to integrated graphics.

1. Reduce in-game ray tracing and reflections — DXGI_ERROR_DEVICE_HUNG is most common with ray-traced reflections on High or Very High settings. Lower ray tracing quality to Medium or disable it entirely. Also disable DLSS Frame Generation if enabled, as it significantly increases GPU workload.

If it still doesn't work

For persistent crashes, try setting TdrLevel to 0 in the same registry path (HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers) as a DWORD value — this disables TDR entirely. Warning: instead of recovering from a GPU hang, your screen will freeze completely and require a hard reboot. Use this only for diagnosis.

If the error occurs across multiple games on a fresh driver install with underclocked GPU, the GPU itself may be failing. Run FurMark or 3DMark stress tests to verify hardware stability. For AMD users specifically: disable Anti-Lag and Radeon Chill in AMD Software, and consider rolling back to Adrenalin 24.5.1 which is reported as more stable for DX12 workloads.