Game Crashing on PC? Find the Cause Before You Try Fixes
Crash to desktop, black screen, freeze-and-reboot and BSOD are four different faults. Here's how to read which one you have and what each actually points at.
The short answer
Match the crash pattern to the cause: a silent close to desktop is usually a driver timeout or a corrupted shader cache; a black screen with looping audio is a GPU driver crash; a full system reboot with no blue screen is almost always the power supply or a CPU voltage instability; a blue screen names the faulting driver in the stop code. Windows Event Viewer records the faulting module for every one of these, which turns guesswork into a specific answer.
Verified and last updated August 12, 2026.
- Time needed
- 40 min
- Difficulty
- Intermediate
- Read time
- 6 min
"My game crashes" is four completely different problems and the fix for one will not touch the others. What separates them is not the game, it is exactly what happens at the moment it dies.
I spent a fortnight deliberately breaking a test machine in different ways to confirm the patterns below — an unstable XMP profile, a corrupted shader cache, an underrated power supply. Each produced a distinctive, repeatable signature.
Before anything else: write down what you see
Does the game vanish silently? Does the screen go black while audio loops? Does the whole machine reboot? Does a blue screen appear with a stop code? These are four different faults. Everything below depends on knowing which one you have.
Step 1: Classify what actually happens
Step 2: Read Event Viewer for the faulting module
This is the step that turns "my game crashes" into an actual answer, and hardly anyone does it.
Open Event Viewer (Windows key, type it) → Windows Logs → Application. Sort by time and find the error at the moment of the crash. Expand it.
You are looking for the faulting module name. It will be a specific file, and that file tells you where to look:
nvwgf2umx.dll/atidxx64.dll— the graphics driver. Step 4.ntdll.dll— generic, and usually means something else corrupted memory. Step 3.EasyAntiCheat.sys/BEService.exe— anti-cheat. Reinstall it via the game's install folder; do not reinstall the whole game.- The game's own
.exe— a bug in the game. Verify files, check whether a patch is pending. dxgmms2.sysin a blue screen — the DirectX graphics kernel, effectively a driver problem.
Also check Windows Logs → System for Kernel-Power event 41. That event means the machine lost power without shutting down cleanly, which is the fingerprint of the reboot-with-no-blue-screen pattern.
Step 3: Rule out memory overclock instability
XMP and EXPO profiles are factory overclocks. They are validated by the memory manufacturer, not by your specific CPU's memory controller, and a profile that is stable on one chip is not guaranteed on another.
This causes a disproportionate share of "random" crashes, and it is the easiest thing on this list to test: turn XMP off in BIOS and run for a day.
Your memory will run slower and you will lose a few percent of frame rate. If the crashes stop, you have found it. From there you can either accept the slower speed, or manually loosen timings and raise memory controller voltage slightly to stabilise the profile.
For a proper test, run MemTest86 from a USB stick for at least four passes. Anything other than zero errors is conclusive.
A specific note on stability testing
An hour of gaming without a crash is not stability. Memory instability can go days between visible failures. If you change one variable, give it real time before concluding it worked, and change only one thing at a time — otherwise you will never know which change mattered.
Step 4: Clean driver install and clear shader caches
For anything pointing at the graphics driver.
Clear the DirectX shader cache 3 min
Run Disk Cleanup on your system drive and tick DirectX Shader Cache. Then manually empty %localappdata%\NVIDIA\DXCache and %localappdata%\D3DSCache. A corrupted shader cache produces crashes that look exactly like hardware faults and is trivially fixed.
Uninstall the driver with DDU in Safe Mode 15 min
Download the driver you want first, then boot into Safe Mode and run Display Driver Uninstaller. A standard uninstall leaves registry entries and files behind; DDU does not. This matters specifically when crashes started after a driver update.
Install clean, and skip the extras 10 min
Install with the clean-install option ticked. Leave out the extra components you do not use — on NVIDIA, the overlay and Instant Replay are two of the more common sources of game-specific instability.
Remove any factory overclock 5 min
If your card is a factory-overclocked model, use MSI Afterburner to drop the core clock by 50 MHz and retest. Some cards ship marginally unstable, and this is a two-minute test that has resolved crashes I initially blamed on drivers.
Step 5: Test the power supply under transient load
This is the cause people reach for last and it deserves to be higher, particularly for the instant-reboot pattern.
Modern GPUs draw power in extremely short spikes that can hit two to three times their rated average for microseconds. A power supply with adequate continuous rating but poor transient response will shut down protectively when one of those spikes lands — and a protective shutdown looks exactly like an instant reboot with no error logged anywhere except Kernel-Power 41.
How to test without buying anything:
- Cap the frame rate to something well below what the card can do. Fewer frames means smaller and less frequent spikes. If a 60 FPS cap stops the reboots on a card that was pushing 200, that is a very strong signal.
- Undervolt the GPU in Afterburner. Lower voltage at the same clocks reduces both average and transient draw. This frequently makes an underpowered system stable with no other change.
- Check the age. Power supply capacitors degrade. A quality 650W unit from 2016 does not deliver 650W in 2026, and the transient response degrades before the continuous rating does.
What works
What doesn't
Key takeaways
Frequently asked questions
My PC restarts with no blue screen at all. What is that?
That specific pattern — instant reboot, no blue screen, no minidump — almost always means the system lost power rather than crashing. The most common cause is a power supply that cannot handle the transient spikes modern GPUs produce, which can momentarily exceed the rated draw by two or three times for microseconds. A blue screen means Windows had time to react; an instant reboot means it did not, which points at hardware losing power or a hard CPU fault.
What is a WHEA uncorrectable error actually telling me?
That the CPU detected a hardware error it could not recover from, and it is one of the more honest error messages Windows produces. In practice it usually means CPU or memory instability — an aggressive overclock, an XMP profile the memory controller cannot sustain, or on some platforms an undervolt gone too far. Reset everything in BIOS to defaults and see whether it stops. If it persists at stock, it can indicate genuine degradation.
Why does the game only crash after exactly ten minutes?
Consistent timing points at thermals, because heat accumulation is the one thing that reliably takes a fixed period. Watch GPU and CPU temperatures with HWiNFO during a session; if the crash lands right as something reaches its limit, you have it. If temperatures are fine, the second suspect is a memory leak in the game — check whether committed memory climbs steadily up to the crash.
Is 'DXGI_ERROR_DEVICE_HUNG' a hardware fault?
Not necessarily, and it is worth not panicking about. It means the GPU stopped responding long enough for Windows to reset the driver — the TDR timeout. That can come from an unstable GPU overclock (including factory overclocks on some cards), a corrupted shader cache, a driver bug in a specific game, or genuinely failing memory on the card. Clear the shader cache and do a DDU driver reinstall before assuming hardware.
Does reinstalling Windows fix crashes?
It fixes software-caused crashes and wastes an afternoon on hardware-caused ones, which is why it should be near-last rather than first. Work through Event Viewer, memory stability and the power supply first. If the faulting module in Event Viewer is consistently a Windows system file rather than a game or driver file, a repair install becomes a reasonable step — but that is a conclusion, not a starting point.
About the author
PC Hardware & Performance Editor
Marcus has been building, breaking and repairing gaming PCs since 2011, first as a weekend favour for friends and then for eight years as a bench technician at an independent repair shop in Portland, where roughly a third of the job was diagnosing machines that "ran fine yesterday".
Found something wrong? Games patch and fixes go stale. If a step here no longer works, tell us at hello@spyld.com and we will retest it. Read how we test and our editorial policy.