PC restarts while gaming? Classify what actually restarted

Stop testing when there is a safety warning

Do not start another game if you smell burning, see smoke or sparks, hear electrical crackling, notice a swollen laptop battery, or find that the computer is abnormally hot even after it should have cooled down. Shut it down if that can be done safely, disconnect external power, protect your data, and arrange an appropriate hands-on assessment.

First decide which of five events happened

“It crashed” is not enough detail. Put the incident into the closest group:

• The game closed, but Windows stayed usable. This is an app-level crash. Use the game crash guide for launch errors, runtimes, overlays, and game-specific failures.
• A stop-code screen appeared, then Windows restarted. Save the code and any module name. Follow the blue-screen evidence path.
• The whole PC restarted without a visible stop code. Note whether the monitor lost signal, fans and lights stopped, and the machine booted again by itself.
• The PC froze until you held the power button. Record that forced shutdown. The later Event 41 describes the unclean shutdown, not the original reason for the freeze.
• Windows showed a normal restart or shutdown message. That may be an orderly restart requested by Windows, an app, an update, or the user.

If you are not sure, write down what you directly observed. “Screen went black, audio stopped, case lights went off for two seconds, then the logo appeared” is much more useful than “random reboot.”

Build an exact-time evidence card

Right after the PC returns, note the time as closely as you can. Include the game, what you were doing in it, how long the session had been running, and whether the same moment is repeatable.

• Did only the game close, or did Windows restart?
• Was there a stop code, black stop screen, frozen frame, or no message?
• Did audio continue after the picture disappeared?
• Did fans, lights, USB devices, or the monitor remain powered?
• Did the PC boot again by itself, or did you press a button?
• Was the computer unusually hot, loud, or otherwise different first?
• What changed recently: game update, Windows update, driver, firmware, peripheral, hardware, or performance setting?

Save important work and back up files before repeating a failure. If the machine restarts outside games or cannot stay stable long enough to preserve data, stop trying to reproduce it.

Read Event 41 as an outcome, not a verdict

Open Event Viewer and look in Windows Logs > System around the time you recorded. Event ID 41 from Kernel-Power means Windows detected on the next startup that the previous shutdown did not finish cleanly.

That event can include a bugcheck code, a power-button timestamp, or zero values. Zeros can appear when power was interrupted, when the PC was forced off, or when Windows could not write the crash details quickly enough. That is why “Event 41” and “bad power supply” are not the same statement.

Preserve the event details and the events immediately before it. Do not clear the log, delete dump files, or reset Windows before you know whether useful evidence exists.

Use the nearby shutdown events to classify the restart

Microsoft documents several events that help explain the sequence:

• Event 1074 records a restart or shutdown requested by a user or application.
• Event 6006 indicates that Windows completed an orderly shutdown.
• Event 6008 records that the previous shutdown was unexpected.
• Event 41 records that Windows started after an incomplete shutdown.

These events describe what Windows observed. They do not automatically identify the component, driver, or app that started the chain. A warning or error immediately before the restart may be relevant, but timing alone is not proof. Save the source, event ID, time, and message so you can compare another incident.

Compare one game with the rest of the system

If one game closes to the desktop while other games and Windows remain stable, stay on the app-level branch. Check that game’s update history, files, launch options, overlays, and known issues without turning a single app crash into a whole-PC hardware conclusion.

If several games trigger a full reboot, test your everyday non-gaming workload only if the PC is safe and stable enough. A restart during light use, sleep, startup, or file work changes the scope. Record that instead of continuing to push the machine under load.

Also compare the trigger. A restart at the same game scene, during shader compilation, when joining online play, or soon after a driver change can narrow the software branch. A restart at unrelated times across many workloads points away from one game.

Check the resolved KB5121003 gaming issue

Microsoft confirmed that some games could freeze, close without notice, show EXCEPTION_ACCESS_VIOLATION, or restart the PC after the August 11, 2026 Windows 11 update KB5121003. The affected cases involve certain peripherals or internal components with RGB-lighting support and drivers or software components with names similar to inpoutx64. Reported games include ARC Raiders, MARVEL Tōkon: Fighting Souls, and THE FINALS.

Microsoft marked the issue resolved on August 27. On consumer and business devices that are not managed by an IT department, Windows now applies a block that prevents the affected inpoutx64 driver from loading. Windows should show a message that the driver was disabled, and the game should then launch normally. Let Windows apply the mitigation, restart if prompted, and repeat one safe game-launch test. Do not delete the driver, uninstall the update, remove hardware, or disable a security control.

Enterprise-managed devices do not receive this mitigation automatically. If this is a work or school PC, stop and ask the IT administrator to follow Microsoft’s current Windows release-health workaround. Do not make the registry change yourself on a managed device.

If the timing, game, RGB-component evidence, or disabled-driver message does not match, keep following the broader restart classifier. Microsoft’s resolution covers this specific driver interaction, not every gaming restart.

Check recent changes before broad repairs

Make a short timeline. Include Windows updates, graphics or chipset drivers, game patches, firmware, new USB devices, new internal hardware, and performance-profile changes. Then reverse or isolate only a change that matches the timing and has a safe rollback path.

If a driver update clearly came first, use the driver rollback and boot-state guide. Do not uninstall every driver or use an unverified driver bundle. If no change lines up with the first incident, leave the system alone long enough to collect another clean comparison.

Record whether the PC is at its documented default settings

Overclocking, undervolting, custom memory profiles, third-party tuning tools, and altered power limits can change stability. Record which settings are active. If you knowingly changed them, use the motherboard, laptop, or component maker’s instructions to return to a documented default only when you understand the rollback and warranty implications.

Do not copy voltage or firmware settings from a forum post. If you did not configure the system yourself, ask the builder or vendor what the supported baseline should be.

Know when software evidence should hand off to hardware checks

A no-code whole-PC restart across several games, especially at documented default settings, can require power, cooling, memory, graphics, board, or other hands-on checks. Windows evidence can help describe the failure, but it cannot safely test a power supply or certify that temperatures and electrical protections are sound.

Hand the exact-time card, System events, visible stop details, game comparison, recent-change timeline, and current settings to the system or component vendor or a qualified technician. That evidence can reduce guesswork without pretending the event log already proved the faulty part.

Keep the next step tied to the evidence

• Game closes, Windows stays up: follow the game-specific crash path.
• Visible stop code: preserve the code and dump evidence, then use the BSOD path.
• Orderly restart with Event 1074: identify the requesting user or application.
• Freeze followed by a forced shutdown: investigate the freeze; Event 41 is the aftermath.
• Instant whole-PC restart with no code: preserve events and settings, stop unsafe repetition, and prepare for a system-level assessment.

The OmniMend help article library keeps app crashes, stop codes, driver changes, freezes, and performance problems separate so each clue leads to a focused next step.

Where OmniMend fits

OmniMend can help collect and organize diagnostic clues such as the restart time, event details, affected games, recent changes, and the difference between an app crash and a whole-system reboot. It cannot test a power supply, certify thermal or electrical safety, or prove a component failure from Event 41.

If you want to see how that evidence-first approach works, read how OmniMend approaches guided computer repair.