GPU Crash in Games – Causes & Complete Fix Guide
Few things are more frustrating than a graphics card crash mid-game — whether it's a black screen that recovers after a few seconds, a sudden BSOD, or a game that crashes to desktop with a cryptic DirectX error. GPU crashes during gaming have many potential causes: outdated drivers, thermal throttling, unstable overclocks, VRAM shortages, power delivery problems, and conflicting software. The good news is that the vast majority of GPU crashes are software or configuration issues that can be fixed without replacing any hardware. This guide walks through every type of GPU crash, how to identify your specific issue, and a proven step-by-step fix process.
Types of GPU Crashes
Black Screen Flash
Screen goes black for 1–3 seconds, then recovers. TDR event — driver restarted.
BSOD Crash
Blue screen with stop code (nvlddmkm.sys, atikmpag.sys). Full system restart.
Game Crash to Desktop
Game closes with DirectX or GPU error message. Often driver or game file issue.
Visual Artifacts
Colored dots, flickering polygons, distorted textures. GPU hardware or VRAM issue.
Most Common Causes of GPU Crashes in Games
- Outdated or corrupted GPU drivers — most frequent cause of in-game crashes
- GPU overheating — core or VRAM temperatures exceeding safe limits under load
- Unstable GPU overclock — even factory OC cards can be unstable with certain game loads
- Insufficient PSU wattage — GPU power spikes causing voltage drops and resets
- Corrupt or missing game files — causes DirectX errors in specific titles only
- Conflicting overlays — Discord, GeForce Experience, MSI Afterburner injecting into DX12/Vulkan
- Outdated Windows — missing DirectX patches delivered via Windows Update
- Faulty GPU hardware — failing VRAM, damaged PCB, or dying GPU chip
- PCIe slot contact issue — GPU not fully seated or slot damaged
How to Fix GPU Crashes in Games – Step by Step
- Monitor GPU temperature during gameplayInstall HWiNFO64 and run it alongside your game. Monitor GPU Core Temp and GPU Memory Junction Temp. Core should stay below 85°C; memory below 100°C. If exceeded, overheating is your primary issue — clean dust, repaste, or improve case airflow.
- Reset all GPU overclocks to stockOpen MSI Afterburner. Click the Reset button (circular arrow) to restore all values to default. Many "factory OC" cards (e.g., ASUS TUF OC, Gigabyte Gaming OC) run slightly outside their stable range and need to be underclocked to fix crash issues.
- Perform a clean driver reinstall using DDUDownload DDU (Display Driver Uninstaller) from guru3d.com. Boot into Safe Mode (hold Shift, click Restart → Troubleshoot → Advanced Options → Startup Settings). Run DDU, select your GPU brand, click "Clean and restart." Install the latest official driver from nvidia.com or amd.com.
- Disable all third-party overlaysDisable: Discord Overlay (Settings → Game Overlay → off), GeForce Experience in-game overlay (Alt+Z → Settings), MSI Afterburner/RTSS overlay. Some DX12 and Vulkan games crash specifically because of overlay injection. Test the game with all overlays disabled.
- Verify game file integritySteam: Right-click game → Properties → Local Files → Verify integrity of game files. Epic Games: Library → three dots → Manage → Verify. GOG Galaxy: Manage Installation → Verify. Corrupted shader cache or game DLL can cause GPU-level crashes in a single game.
- Check PCIe power cables and GPU seatingPower off completely and unplug. Reseat the GPU in the PCIe slot — press firmly until the clip locks. Remove and reconnect all GPU power cables. If your PSU uses a daisy chain, use separate cables for each GPU power connector instead.
- Run FurMark stress test to isolate hardware failureRun FurMark at 1080p for 15 minutes. If artifacts appear or the system crashes before 10 minutes, GPU hardware is suspect. Compare this to game crashes — if FurMark passes but games crash, the issue is software. If FurMark also crashes, consider GPU RMA or replacement.
- Try undervolting the GPUUsing MSI Afterburner's voltage/frequency curve editor (Ctrl+F), lower the GPU core voltage while maintaining clock speed. Undervolting reduces heat and often stabilizes GPUs with power delivery or thermal issues without reducing performance.
Most Common GPU Crash Error Codes
| Code / ID | Error Name | Severity |
|---|---|---|
nvlddmkm.sys | NVIDIA display driver TDR crash (BSOD) | Critical |
atikmpag.sys | AMD GPU driver kernel crash (BSOD) | Critical |
0x887A0006 | DXGI_ERROR_DEVICE_HUNG – GPU not responding | Critical |
0x887A0005 | DXGI_ERROR_DEVICE_REMOVED – GPU disconnected | Critical |
Code 43 | Windows stopped the GPU – driver/hardware fault | High |
TDR Event | Display driver stopped responding and recovered | High |
VIDEO_TDR_FAILURE | BSOD – GPU driver timeout (0x00000116) | Critical |
VIDEO_DXGKRNL_FATAL_ERROR | DirectX kernel fatal error (0x00000113) | Critical |
Frequently Asked Questions
Why does my GPU crash only in certain games?
Game-specific crashes usually indicate a driver compatibility issue, game file corruption, or a DirectX/Vulkan conflict. Verify game file integrity via Steam or Epic. Check if the game has known issues with your driver version — roll back to a previous driver if needed. Some games stress VRAM more than others, exposing an unstable memory overclock.
What is a TDR crash and how do I fix it?
TDR (Timeout Detection and Recovery) is a Windows mechanism that resets the GPU driver when it stops responding for more than 2 seconds. You'll see a black screen flash and a notification "Display driver stopped responding and has recovered." Fix by: updating drivers, removing GPU overclock, checking temperatures, and ensuring the PSU provides adequate wattage.
Can a faulty PSU cause GPU crashes?
Yes. An underpowered or failing PSU is a common cause of GPU crashes under load. When the GPU demands peak power during intense scenes, voltage drops trigger driver resets or shutdowns. Use a PSU wattage calculator to verify headroom. If the PSU is over 5 years old, consider replacing it regardless of rated wattage.
How do I know if my GPU is dying?
Signs of a dying GPU include: persistent artifacts even outside games, crashes that occur under lighter and lighter loads over time, GPU not detected in Device Manager, and Code 43 errors. Run FurMark — if artifacts appear at low load or the system crashes within minutes at stock clocks, the GPU hardware is failing and needs replacement.