When an Unreal Engine game crashes, it rarely behaves like a typical software failure. Instead of a polite “game encountered an error” dialog, you often get an instant desktop return, a black screen that freezes for several seconds before Windows kills the process, or a crash reporter window that dumps a wall of call-stack data most players never read. The symptom pattern matters. A crash during shader compilation on first launch points to a different root cause than a random freeze during heavy combat with Nanite geometry on screen. UE5’s architecture — with its deferred shader pipeline, Lumen lighting system, and heavy reliance on VRAM for virtual textures — creates a specific failure surface that generic “reinstall your drivers” advice barely touches.
- Likely Causes
- Fix 1: Verify Game File Integrity
- Fix 2: Clear the Derived Data Cache and Shader Cache
- Fix 3: Switch Between DX12 and DX11 via Launch Options
- Fix 4: Disable Overlays and Injection Tools
- Fix 5: Update or Clean-Install GPU Drivers
- Fix 6: Exclude Game Folder from Antivirus Real-Time Scanning
- Fix 7: Adjust Power Plan and PCIe Link State
- Fix 8: Lower or Disable Nanite and Lumen
- Fix 9: Perform a Clean Reinstall of the Game
- If Nothing Works
- FAQ
This guide focuses on the most common crash vectors unique to Unreal Engine 5 titles, with real file paths, error strings, and launch options you can verify on your own system.
Likely Causes
| Cause | How to confirm | Fix difficulty |
|---|---|---|
| Corrupted Derived Data Cache (DDC) or shader bytecode | Crash string contains “Shader compilation failed” or “Can’t load resource”; crash occurs on first or second launch after an update | Low |
| VRAM exhaustion from Nanite or Lumen workloads | Windows Event Viewer shows error code 0xC0000005 or “Video memory is full” in the UE log file | Low |
| GPU driver instability with DX12 path | Crash reports reference d3d12.dll or DXGI; switching to DX11 resolves it | Low |
| Mod or DLL injection conflict (UE4SS, custom ini) | Game runs fine after removing mod files; log shows “Fatal error” near a DLL load line | Medium |
| Antivirus or Windows Defender quarantining UE binaries | UnrealGame.exe or CrashReportClient.exe missing from game folder; Defender history shows recent quarantine | Medium |
| Corrupted game files from partial update | Steam “Verify Integrity” reports missing .pak files; crash occurs loading a specific level | Low |
| Overclocked VRAM or unstable power limits | Crashes only under sustained load; MSI Afterburner shows VRAM temperatures above 100°C | Medium |
| Windows power plan throttling or PCIe power state issues | Crash log shows “GPU hang detected” followed by TDR reset; only occurs after idle period | Medium |
Fix 1: Verify Game File Integrity
How to check
Open the game’s properties in Steam, select Installed Files, and click “Verify integrity of game files.” If any .pak or .ucas files in the game’s Content\Paks folder are missing or mismatched, Steam will re-download them. Alternatively, navigate to your game’s directory and look for a mismatch in file count between Content\Paks and the manifest.
What to do
Let the verification complete. If files are repaired, launch the game again. UE games are highly sensitive to pak file integrity because the engine maps these archives as virtual file systems at runtime — a single corrupted pak can trigger a fatal assertion during level load.
How to undo it
No undo needed. Verification only adds or replaces missing files.
Fix 2: Clear the Derived Data Cache and Shader Cache
How to check
Navigate to %LOCALAPPDATA%\[GameName]\Saved\ShaderPipelineCache and %LOCALAPPDATA%\[GameName]\DerivedDataCache. If these folders are unusually large (multi-gigabyte) or contain files dated before your most recent game update, stale entries are a common crash trigger.
What to do
Close the game completely. Delete the contents of the ShaderPipelineCache folder and the DerivedDataCache folder (not the folder itself — the engine recreates it). First launch after clearing will take significantly longer due to on-the-fly shader compilation, so be patient and avoid alt-tabbing.
How to undo it
No undo needed. The engine rebuilds caches automatically on next run.
Fix 3: Switch Between DX12 and DX11 via Launch Options
How to check
Open the last crash log in %LOCALAPPDATA%\[GameName]\Saved\Crashes (the most recent UECC-Windows-* folder). Look inside the .log file for references to d3d12.dll, DXGI_ERROR_DEVICE_REMOVED, or “DX12 is not supported.” If present, the DX12 path is your likely culprit.
What to do
In Steam, open Properties > Launch Options. Add -dx11 to force Direct3D 11, or -dx12 to explicitly set it if you are stuck on an older path. Some titles also expose this via their own launcher or a config file at GameUserSettings.ini where you can set GraphicsAPI=DX12 or GraphicsAPI=DX11.
How to undo it
Remove the launch option or edit GraphicsAPI back to its original value.
Fix 4: Disable Overlays and Injection Tools
How to check
Temporarily disable the Steam Overlay (Properties > General > uncheck “Enable Steam Overlay”), Epic Online Services overlay, Discord overlay, MSI Afterburner/RivaTuner, and any UE4SS or cheat-engine DLL injection. If the crash disappears, re-enable one at a time.
What to do
UE5’s render thread is sensitive to hooks placed by overlay injection, especially when using the DX12 path with async compute queues. Disabling all injections while testing isolates this variable.
How to undo it
Re-enable overlays individually to identify the conflict. You may only need to disable one.
Fix 5: Update or Clean-Install GPU Drivers
How to check
Check your GPU driver version. For NVIDIA, this is in the driver properties or by running nvidia-smi in Command Prompt. Compare against the latest release. For AMD, use Adrenalin or Radeon Software.
What to do
Use DDU (Display Driver Uninstaller) in Safe Mode to fully remove the driver, then install the latest stable release. UE5 crashes frequently after partial driver updates because the DX12 shader compiler (nvoglv64.dll or aticfx64.dll) can mismatch with the engine’s cached pipeline state. A clean install eliminates this. If you run an overclocked GPU, reset to stock clocks as a diagnostic step.
How to undo it
DDU-based installs are clean replacements. To revert, uninstall the new driver and install the previous known-good version.
Fix 6: Exclude Game Folder from Antivirus Real-Time Scanning
How to check
Open Windows Security > Virus & threat protection > Protection history. If you see recent blocks on UnrealGame.exe, UE4SS, or CrashReportClient.exe, your AV is the cause.
What to do
Add the game’s entire installation directory and its %LOCALAPPDATA%\[GameName] folder to your antivirus exclusion list. UE games generate and modify many small files rapidly (save games, temp shader files, crash dumps) which can trigger heuristic scanners into quarantining them mid-write.
How to undo it
Remove the exclusion entry from your antivirus settings.
Fix 7: Adjust Power Plan and PCIe Link State
How to check
Open Event Viewer > Windows Logs > System. Search for “Display” as the source. Repeated entries showing “The display driver nvlddmkm stopped responding and has successfully recovered” indicate a TDR (Timeout Detection and Recovery) reset, often tied to power state transitions.
What to do
Set Windows Power Plan to “High Performance” or “Ultimate Performance.” In NVIDIA Control Panel or AMD Radeon Settings, ensure Power Management Mode is set to “Prefer Maximum Performance.” You can also increase the TDR delay via a registry edit at HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers by setting TdrDelay (DWORD) to 10 or higher, but this is advanced.
How to undo it
Restore your original power plan. Delete or set TdrDelay to 0 to return to default Windows behavior.
Fix 8: Lower or Disable Nanite and Lumen
How to check
Open GameUserSettings.ini in %LOCALAPPDATA%\[GameName]\Saved\Config\Windows. Search for r.Nanite and r.DynamicGlobalIlluminationMethod. If the latter is set to 1 (Lumen), you are running the heaviest GI path.
What to do
Set r.DynamicGlobalIlluminationMethod=0 to disable Lumen and fall back to baked lighting (where available). Set r.Nanite=0 to force traditional rasterization. This dramatically reduces VRAM pressure and avoids known UE5 crash vectors on GPUs with 8 GB or less. As of 2026, most UE5 titles ship with these as configurable cvars even if the in-game menu does not expose them directly.
How to undo it
Delete the ini file or revert the values to their original numbers.
Fix 9: Perform a Clean Reinstall of the Game
How to check
Before doing this, confirm you have saved all fix attempts and logged results. If all previous steps failed and crash logs show repeated assertion failures pointing to engine internals, the install itself may be fundamentally corrupted.
What to do
Uninstall the game. Delete any remaining files in the installation directory. Delete %LOCALAPPDATA%\[GameName] in its entirety. Reinstall. This is the most invasive option because it removes all local configuration, save data (unless cloud-synced), and custom settings.
How to undo it
There is no undo. Back up your %LOCALAPPDATA%\[GameName]\Saved\SaveGames folder before deleting.
If Nothing Works
At this point, the crash is likely hardware-level. Stress-test your VRAM with a tool like OCCT or TestMem5 (GPU mode). Check your GPU’s PCIe power delivery — daisy-chained 8-pin cables are a known instability factor with high-draw cards running UE5 workloads. If your system has more than one GPU (integrated + discrete), ensure the game is actually using the discrete GPU via Windows Graphics Settings > assign > High performance. Finally, check whether the game has a known open crash bug in its community forums — some UE5 titles have per-title crash vectors that only a patch from the developer can resolve.
FAQ
What does “Fatal error: [File: ..\Runtime\Engine\Private\Streaming\StreamingManager.cpp]” mean?
This indicates the engine’s asset streaming system failed to load a resource within its allotted time or encountered a null pointer during level streaming. It most commonly means your DDC is corrupted, your drive is too slow (HDD instead of SSD), or the game is asking for more VRAM than your GPU has available for streamed assets.
My game crashes only in specific areas — what does that point to?
Area-specific crashes almost always mean a particular asset in that level has a corrupted shader, an oversized virtual texture, or a mesh with a topology issue that triggers a rare code path. Try lowering texture quality and disabling virtual shadow maps in the in-game settings. If it persists, clearing the DDC (Fix 2) forces recompile of that area’s shaders.
Where exactly are UE5 crash dumps stored?
In %LOCALAPPDATA%\[GameName]\Saved\Crashes\. Each crash creates a folder named UECC-Windows-[GUID]. Inside, you will find a .log file with the full call stack and a .dmp file for Windows debuggers. The .log file is the most immediately useful for identifying which module or asset caused the fault.
Does enablingResizable BAR help with UE5 crashes?
ReBAR can reduce VRAM-related stalls and stuttering in UE5 titles by letting the GPU address all system RAM in one aperture, but it does not directly prevent crashes. If you are experiencing “Out of Video Memory” assertions, enabling ReBAR may help marginally, but lowering graphics settings or reducing the number of active streaming textures (via ini console commands like sv.StreamingDurationScaleFactor) is more effective.
