Fix "Low Level Fatal Error" in Unreal Engine Games

1. GPU Driver Timeout (TDR) — the most common cause

What's actually happening here is the GPU driver stops responding to Windows for more than 2 seconds. Windows then resets the driver, Unreal Engine loses its rendering context, and you get the LowLevelFatalError with a DXGI_ERROR_DEVICE_HUNG or DXGI_ERROR_DEVICE_REMOVED in the log.

This hits hardest in Unreal Engine 4 games built with DX11 or DX12 — think Fortnite (save the world mode), Squad, Hell Let Loose, or Ark: Survival Evolved. The trigger is usually a sudden scene change, loading a new map, or alt-tabbing.

The real fix: clean GPU driver install

1. Download Display Driver Uninstaller (DDU) from Guru3D — run it in Safe Mode. Don't skip Safe Mode, because Windows Update will reinstall garbage drivers while you're still in normal mode.
2. Select "Clean and restart" for your GPU vendor (NVIDIA/AMD/Intel).
3. After reboot, install the studio driver version from your GPU vendor's site, not the game-ready driver. Game-ready drivers often have bleeding-edge tweaks that break Unreal's rendering pipeline. Studio drivers are validated more thoroughly. Yes, you lose 2-3 FPS in some titles. Worth it for stability.
4. In the NVIDIA Control Panel or AMD Adrenalin, set the game's executable to "Prefer maximum performance" power management mode. Also disable any GPU overclocking — Unreal Engine hates unstable VRAM clocks.

If you still crash, you can lengthen the TDR timeout. This doesn't fix the root cause, but it stops Windows from killing the driver prematurely so you can diagnose further.

reg add "HKLM\SYSTEM\CurrentControlSet\Control\GraphicsDrivers" /v TdrDelay /t REG_DWORD /d 10 /f

This sets the timeout to 10 seconds instead of the default 2. Reboot required. If you still get TDR crashes after this, your GPU or VRAM is likely failing — see next section.

2. Corrupted or unstable GPU memory (VRAM)

The error message often buries a deeper clue: LowLevelFatalError [File:Unknown] [Line: 0] with no DXGI error. That usually means the GPU tried to allocate memory and got back garbage, or the shader compiler hit a corrupted instruction cache.

This one's hardware. I've seen it on RTX 3070s with micron VRAM that developed bad blocks after 2 years of heavy use. Also happens on overclocked cards where VRAM temps hit 95°C+.

Test and isolate

1. Run OCCT VRAM test (free) for 30 minutes. Any errors = VRAM failure. Replace the GPU.
2. If no errors, run MemTest86 overnight on system RAM. Bad system RAM can corrupt shader caches on the fly.
3. Lower your VRAM clock by 200 MHz in MSI Afterburner or AMD Adrenalin. If the crash stops, you've been running borderline unstable VRAM from factory.

Don't bother with "verify game files" for this — the files are fine, the RAM handling them isn't.

3. Broken shader cache or pipeline state objects

Unreal Engine precompiles shaders on first launch or after a driver update. If the shader cache gets corrupted—say from a crash during compilation, or a driver update that left half-baked cache files—the engine hits a LowLevelFatalError when it tries to execute a command list referencing a corrupted PSO.

The telltale sign: crash happens at the same spot in the game every time (same map, same lighting condition, same weapon effect).

Clear the shader cache and force recompilation

1. Delete the shader cache folder: %LOCALAPPDATA%\{GameName}\Saved\Shaders. For Unreal Engine 5 games, it's often %LOCALAPPDATA%\{GameName}\Saved\CachedShaderData. Check the game's wiki for exact path.
2. Open NVIDIA Control Panel → Manage 3D Settings → Program Settings → select the game → set "Shader Cache Size" to "Disabled" temporarily. Apply, then re-enable it. This clears NVIDIA's own cache too.
3. Launch the game. It will recompile all shaders — first launch after a driver update takes 5-15 minutes. Let it finish. Do not alt-tab during this.

If the crash persists, your shader compiler (part of the GPU driver) might be corrupted. A clean driver install from step 1 fixes that.

Quick-Reference Summary