DXGI_ERROR_DEVICE_HUNG: How to Fix the GPU Crash

DXGI_ERROR_DEVICE_HUNG (hex code 0x887A0006) means your GPU took longer than Windows’ ~2-second Timeout Detection and Recovery (TDR) window to respond, so Windows reset the graphics driver and crashed the game. It is almost always an unstable GPU — a bad overclock, a corrupt driver, or insufficient power — not a game bug, which is why the same crash follows you across different titles. Work the fixes below in order; the first three clear the large majority of cases.

What DXGI_ERROR_DEVICE_HUNG actually is

Windows watches your GPU with a feature called Timeout Detection and Recovery (TDR). If the GPU doesn’t finish a graphics instruction within roughly 2 seconds, Windows assumes the driver has locked up, forcibly resets it, and the running game dies with DXGI_ERROR_DEVICE_HUNG. The error is a symptom of the GPU stalling — your job is to find why it stalled.

The two errors people confuse:

Error	Hex code	What Windows thinks happened	Most likely cause
DXGI_ERROR_DEVICE_HUNG	0x887A0006	GPU stalled past the TDR timeout	Unstable overclock, driver bug, brief power dip
DXGI_ERROR_DEVICE_REMOVED	0x887A0005	GPU disappeared from the system bus	Power delivery, failing card, driver crash

If you see both errors intermittently, treat it as one problem — the fixes below apply to both, with a note where they differ.

Fix 1 — Clean-install your GPU driver with DDU (do this first)

A corrupt or half-updated driver is the single most common software cause. A normal “install over the top” does not remove the broken files — you need a clean wipe.

1. Download the latest stable driver for your card from NVIDIA or AMD, plus Display Driver Uninstaller (DDU).
2. Boot into Safe Mode and run DDU → Clean and restart.
3. Install the fresh driver. Choose a clean install if the installer offers it.

Full step-by-step with the Safe Mode details: How to clean-install GPU drivers with DDU. If the crash started right after a driver update, roll back to the previous stable branch instead of the newest one — new “Game Ready” drivers occasionally regress on specific engines.

Fix 2 — Reset your overclock (the number-one hardware cause)

An unstable overclock is the top cause of DEVICE_HUNG, and this includes factory-overclocked cards that are fine at the desktop but crack under a demanding game engine.

• If you use MSI Afterburner or a similar tool, set the core and memory offsets back to 0 and lower the power limit to 100%.
• If you never overclocked manually, your card may be running an unstable factory OC or an aggressive undervolt curve. In the NVIDIA Control Panel, set the power management mode and try enabling Debug Mode (right-click the desktop → NVIDIA Control Panel → Help menu → Debug Mode), which forces the card to reference clocks and disables the factory overclock.
• Test at stock for a full session before re-applying any tuning. See our GPU overclock guide for how to find a stable offset once the crash is gone.

If stock clocks stop the crash, your previous overclock was the problem — dial it back and re-test in small steps.

Fix 3 — Give the GPU enough power

A GPU that momentarily starves for power will stall long enough to trip TDR.

• Desktop: make sure each PCIe power cable is seated (use separate cables, not a single daisy-chained one on high-draw cards) and that your PSU has the headroom for a modern GPU’s transient spikes.
• Laptop: plug directly into the wall charger, not a battery or a low-wattage USB-C dock. Laptops cap GPU wattage on battery and will crash under load. Set Windows power mode to Best Performance.
• Set the Windows power plan to High performance or Ultimate Performance so the GPU isn’t down-clocking mid-frame.

Fix 4 — Increase the TDR timeout (workaround, not a cure)

If the crash is intermittent and the fixes above haven’t fully stopped it, giving the GPU more time to recover from a brief stall can mask the remaining crashes. This is a registry edit — back up your registry first, and never disable TDR entirely.

1. Press Win + R, type regedit, and press Enter.
2. Navigate to: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers
3. Right-click the right pane → New → DWORD (32-bit) Value, name it TdrDelay.
4. Double-click TdrDelay, set Base: Decimal, and enter 8 (eight seconds).
5. Close the editor and restart your PC.

TdrDelay raises the timeout from ~2 seconds to 8, so a GPU that briefly stalls under a load spike gets time to finish instead of being reset. It helps DEVICE_HUNG far more than DEVICE_REMOVED. If it makes no difference, delete the value and move on — the stall is too long to be a timeout-window issue.

Fix 5 — Cap your frame rate and check your VRAM

Two smaller but real causes:

• Uncapped frame rates in menus or on loading screens can spike the GPU and trip the timeout. Cap FPS a few frames below your refresh rate with an in-game limiter or RivaTuner Statistics Server (RTSS). Enabling V-Sync as a test is a quick way to confirm this is the trigger.
• Failing VRAM or system RAM can cause a hang that no driver reinstall fixes. Run MemTest86 for system RAM and a GPU memory test if the crash survives every fix above.

Still crashing? Rule out the game and the heat

If DEVICE_HUNG only happens in one game, verify its files (Steam → right-click → Properties → Installed Files → Verify) and delete its shader cache — see when to clear the shader cache. If it happens across every game, it is your GPU, driver, or power, and the fixes above are the path. Finally, watch your GPU temperature: a card hitting its thermal limit can hang under sustained load, so confirm case airflow and clean the heatsink if temps exceed the mid-80s °C.

Related guides

• Out of video memory error fix
• How to clean-install GPU drivers with DDU
• Clear the shader cache to fix stuttering
• Best GPU overclock guide for gaming