Fix DirectX 12 Error: Device Removed (DXGI_ERROR_DEVICE_REMOVED) in Games

You're in the middle of a ranked match or a long boss fight, and suddenly the screen freezes for a few seconds, then the game minimizes and you see a pop-up: DXGI_ERROR_DEVICE_REMOVED. Sometimes it's followed by a message about the GPU driver being removed or reset. This happens most often in DirectX 12 games like Cyberpunk 2077, Fortnite, Forza Horizon 5, or any title using ray tracing. The trigger is usually a heavy graphical load—ray tracing on, high resolution, or a sudden spike in GPU usage. But it can also happen on the desktop if you're mining or rendering video.

What Actually Causes This Error?

When Windows detects that your graphics driver hasn't responded within a certain time (usually 2 seconds), it kills the driver and resets the GPU. That's called a TDR (Timeout Detection and Recovery). When the driver comes back, the game loses the connection to the GPU and throws DXGI_ERROR_DEVICE_REMOVED. The root cause is almost always one of these:

• An outdated or corrupted GPU driver.
• An unstable overclock on the GPU, memory, or even the CPU (yes, CPU instability can cause this).
• Power delivery issues—your PSU can't supply enough stable power under load.
• DirectX 12's way of handling errors—it's stricter than DirectX 11, so it reports crashes instead of trying to recover.
• Windows' TDR timeout is too short for your GPU to complete a heavy task.

Let me be clear: you don't need to reinstall Windows. In 90% of cases, a clean driver install fixes it. If that doesn't work, we'll adjust the TDR timeout in the registry, but only as a last resort.

Fix 1: Clean Install Your GPU Driver

Don't just use the "Update Driver" button in Device Manager—that often installs the same broken driver. You need to remove the existing driver completely and then install the latest one from scratch.

1. Download the latest driver from your GPU manufacturer's website.
   • Nvidia: https://www.nvidia.com/drivers
   • AMD: https://www.amd.com/en/support
   • Intel (if you have an iGPU): https://www.intel.com/content/www/us/en/download-center/home.html
2. Once downloaded, run the installer but choose Custom Installation (Nvidia) or Factory Reset (AMD) when prompted.
3. Check the box that says Perform a clean installation or Reset to factory settings.
4. Let it install. The screen will flicker a couple of times—that's normal.
5. Reboot your PC after the install completes.

After rebooting, try playing the game. If the error is gone, you're done. If not, move to Fix 2.

Fix 2: Disable Overclocks and Set GPU to Stock

If you've overclocked your GPU or memory, that's the first suspect. Even a stable overclock in one game can crash in another because DirectX 12 uses the GPU differently.

1. Open your overclocking tool (MSI Afterburner, EVGA Precision X1, or AMD Radeon Software).
2. For GPU overclocks: Set the core clock offset to 0 MHz and memory clock offset to 0 MHz.
3. For CPU overclocks (if you have a K-series or Ryzen with PBO), go into BIOS and disable the overclock. Set everything to Auto.
4. If you're using XMP/EXPO for RAM, temporarily disable it to test—sometimes memory instability triggers this error.
5. Save and reboot.

Now test the game again. If it still crashes, we move to the TDR fix.

Fix 3: Increase Windows TDR Timeout (Registry)

Sometimes your GPU is fine, but it just takes longer than 2 seconds to finish a task under heavy load. Windows' watchdog kills the driver too early. We can give it more time.

1. Press Win + R, type regedit, and press Enter.
2. Navigate to this path:
   

   HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers

3. In the right pane, right-click and select New → DWORD (32-bit) Value.
4. Name it TdrDelay.
5. Double-click it, set the base to Decimal, and enter 10 (this gives the GPU 10 seconds instead of 2).
6. Click OK and close Registry Editor.
7. Reboot your PC.

After this, try the game again. If it still crashes, we need to look at power and temperature.

Fix 4: Check Power Supply and Temperatures

This error is a classic sign of a PSU that can't handle sudden power spikes from the GPU. If you have a 650W PSU with a powerful card like a 3080 or a 6900 XT, that's likely the issue. Here's what to do:

• Use a monitoring tool like HWInfo64 and watch the GPU power draw during a game. If it hits the card's max and the system crashes, it's power.
• Check your GPU temperature. If it's above 85°C (185°F), it might be thermal throttling and then crashing. Clean the fans, reapply thermal paste if you're comfortable, or set a more aggressive fan curve.
• Make sure your PSU is at least 80 Plus Bronze and has enough wattage for your system. You can calculate with a PSU calculator online.
• If you're using a pigtail cable (one cable with two connectors) for your GPU, try two separate PCIe cables if your PSU supports it. That provides cleaner power.

If you've done all this and the error persists, try running the game in DirectX 11 or Vulkan if it offers that option. Yeah, it's a workaround, but sometimes a game's DirectX 12 implementation is just buggy.

Still Crashing? Here's What to Check

If none of the above worked, you're in the small minority. Here's a few last things:

• Check Windows Event Viewer – Look under Windows Logs → System for any WHEA errors or Event ID 41 (Kernel-Power). These point to hardware issues beyond the driver.
• Run a stress test – Use FurMark for 10 minutes and see if it crashes. If it does, your GPU is likely physically failing.
• Test your RAM – Run Windows Memory Diagnostic (Win+R, type mdsched.exe) and let it reboot and test.
• Try a different game – If it only happens in one game, that game might have a known issue. Check its forums for hotfixes.

In rare cases, the GPU is just dying. If you've exhausted everything and the error occurs even in light 2D apps, it's time to RMA the card. But that's the exception, not the rule.