Fix DirectX Error 0x887A0006 (DXGI_ERROR_DEVICE_REMOVED)

You're mid-match, the action's at peak, and then boom—the screen freezes for a second, drops to black, and the game closes with an error dialog saying something like "DXGI_ERROR_DEVICE_REMOVED 0x887A0006." This usually pops up in DirectX 11 or 12 titles—think Cyberpunk 2077, Forza Horizon 5, or even lighter games like Fortnite—especially after a driver update or when your GPU gets pushed hard under load. I know this is infuriating, especially when you've just spent an hour tweaking settings.

The root cause is simpler than most forums make it out to be. The error means Windows or the DirectX runtime has detected that your graphics card has 'gone away' from the system's point of view. This isn't your GPU physically falling out—it's the driver crashing or a timeout window being exceeded. Windows has a built-in watchdog called TDR (Timeout, Detection, and Recovery) that kills the driver if it takes longer than a certain amount of time (default is 2 seconds) to respond to a command. When the GPU is busy rendering something heavy and the driver hangs for just a moment, TDR kicks in, resets the driver, and the game loses its connection to the device. That's your 0x887A0006.

There are a few common triggers: overheating, unstable overclocks, outdated or corrupted drivers, and occasionally a game-specific bug. But don't panic—most of the time, it's a driver issue, and you can fix it without replacing hardware.

Step-by-Step Fix

Update or rollback your GPU driver. Head to NVIDIA's or AMD's site, not the Windows Update catalog. Use the latest Game Ready or Adrenalin driver. If the error started right after an update, roll back via Device Manager → Display adapters → your GPU → Properties → Driver → Roll Back Driver.
Increase the TDR delay. This gives your GPU more time to respond. Open Registry Editor (Win+R, type regedit), go to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers. Create a DWORD (32-bit) called TdrDelay, set it to 10 (that's 10 seconds). Restart. This doesn't fix the root cause, but it stops the premature timeout so you can finish a match and debug properly.
Remove any GPU overclocks. If you use MSI Afterburner or EVGA Precision X1, set the core and memory clock sliders back to zero. Even factory-overclocked cards can be unstable under load—try setting a slight underclock (say -50 MHz) to see if stability improves.
Check your temperatures. Download HWiNFO64 and run it in sensors mode. If your GPU hits above 85°C under load, clean the fans, reapply thermal paste if it's an older card, or check your case airflow.
Run the game in DX11 mode if possible. Some games let you force it via launch options. For Cyberpunk, add -dx11 to the Steam launch options. It's not ideal, but it narrows down whether the issue is specific to DX12.
Disable hardware acceleration in your browser or apps. This is a sneaky one—if you have YouTube or Discord running with hardware accel on, it can interfere with the GPU's state. Turn it off and test.

What If It Still Fails?

If you've done all that and the error persists, here's what to check next:

Power supply. A dying PSU can cause voltage dips that drop your GPU. If you hear coil whine or your PC reboots randomly, this is likely it.
BIOS settings. Make sure PCIe link speed is set to Auto or Gen4, not forcing Gen5 if your card doesn't support it.
Another GPU to test. Borrow a friend's card or run integrated graphics if your CPU has it. This isolates the problem to the card itself.

The TDR registry tweak is your most reliable quick fix, but don't ignore the signs of a dying card. If you get this error even on light games, it's probably not a driver issue—it's hardware. Consider underclocking via MSI Afterburner by -100 MHz to see if that stabilizes things while you plan a replacement.

I've seen this error hundreds of times in my help desk days, and 80% of the time it's a driver version mismatch or an overclock gone wrong. Start there, be patient, and you'll get back to gaming.