PC restarts or shuts down while gaming: Kernel-Power 41

The game freezes or the screen goes black, and the PC restarts or switches off. Afterwards, Event Viewer shows a Critical entry from Kernel-Power, Event ID 41. That entry records that the shutdown went wrong; it is not a diagnosis. Below: what it tells you, where the rest of the evidence is, and what no software can see.

What Event 41 says, and what it does not

Microsoft's explanation: "If your computer shuts down unexpectedly, Windows logs Event ID 41 the next time that the computer starts" (Microsoft). The event's own message says the system "stopped responding, crashed, or lost power unexpectedly", and Microsoft adds: "By itself, Event ID 41 might not contain sufficient information to explicitly define what occurred."

The event appears whatever the cause, and sometimes not at all; the useful part is its event data.

Read the event

Press Win+R, run eventvwr.msc, and go to Event Viewer > Windows Logs > System. Open the Kernel-Power entry with Event ID 41 from the time of the restart and find two values in its event data: BugcheckCode and PowerButtonTimestamp. Microsoft separates three cases:

A Stop error code

Event 41 stores the code in decimal; Microsoft's bug check pages list codes in hexadecimal. Convert it in Calculator's Programmer mode (Microsoft's example: 159 is 0x0000009f) and look it up in Microsoft's bug check reference. Microsoft's pages for two codes name hardware causes directly:

• 0x124, WHEA_UNCORRECTABLE_ERROR (292 in decimal) "indicates that a fatal hardware error has occurred" and "can be heat related or a result of defective hardware, memory, or even a processor that's beginning to fail" (Microsoft). Microsoft's advice: disable any overclock and confirm that the cooling works.
• 0x116, VIDEO_TDR_FAILURE (278 in decimal) means "an attempt to reset the display driver and recover from a timeout has failed". The hardware causes Microsoft lists include over-clocked components, memory configuration and timings, insufficient cooling and insufficient power.

The power button

If you held it because the picture froze, Microsoft's page for 0x116 describes that scene: the system "appears completely frozen or hung" and "No screen updates occur". If freezes like this never recover, w0a's GPU crash recovery card on the Fix page restores Windows' GPU recovery settings to their defaults, in case something changed them (GPU crash recovery). It changes a setting; it repairs no hardware.

All zeros

Microsoft says the PC "might shut down too quickly to write the error codes or even detect an error", that this "might indicate a power supply problem", and lists "The power supply is underpowered or faulty" among the causes. First, though, look for Event ID 46 from volmgr, reporting that crash dump initialization failed. Microsoft says this can happen when the PC started without a configured dump file ("The default dump file is the pagefile"), and a Stop error could then restart it without its code reaching Event 41. If you find it, check the page file (The page file and games).

The entries around it

In the same log:

• 6008 says the previous shutdown "was unexpected" and gives its time. Microsoft suggests checking for "any suspicious event before the shutdown time" in both the Application and System logs.
• 1001 reads "The computer has rebooted from a bugcheck", with the Stop code in hexadecimal and the path of the dump.
• 1074 means an application or a user requested the restart; Microsoft notes that an unexpected reboot usually has none.
• WHEA-Logger entries are hardware errors, each with "a human readable message with the main details of the event". Windows tries to recover from non-fatal hardware errors "to avoid causing a bug check", so these entries can appear before anything crashes.

Why did my game crash? covers Event Viewer and Reliability Monitor (Win+R, perfmon /rel) in more depth.

Keep the evidence next time

Automatically restart is on by default, so after a Stop error the PC restarts by itself. If you saw a blue screen with a code but Event 41 has none, Microsoft's advice is to clear the option:

1. Select the taskbar search box, type Advanced system settings and press Enter.
2. On the Advanced tab, under Startup and Recovery, select Settings.
3. Clear Automatically restart.

The same window sets Write debugging information. The default, Automatic memory dump, is saved as %SystemRoot%\MEMORY.DMP; a Small memory dump goes to %SystemRoot%\Minidump and holds the Stop message, its parameters and a list of loaded drivers. Dumps need the paging file on the boot volume, so leave the page file on.

Power: the part software cannot see

Intel's ATX 3 power supply design guide explains why the standard changed: graphics cards had exceeded their power limits in brief excursions that older supplies "were never explicitly designed to withstand", and those excursions "could cause unexpected (and untraceable) card or system malfunctions, potentially triggering PSU overcurrent protection (OCP) or voltage droop" (Intel). Revision 2.1a, the one Intel publishes as of October 2026, sizes supplies for short peaks: for a supply over 450 W with a 12V-2x6 connector, its test table runs from 200% of rated wattage for 100 µs to 120% for 100 ms.

"Untraceable" is the honest limit. A supply that trips cuts the power, which is the zero-code case above, and the event cannot tell you more. What you can check:

• Rating. NVIDIA, for example, lists 750 W of required system power for the GeForce RTX 5070 Ti and 650 W for the RTX 5070, based on a PC with a Ryzen 9 9950X, and notes that "Power requirements can be different depending on system configuration" (NVIDIA). Compare your supply with your card maker's figure.
• The connector. In Intel's words, the 12V-2x6 connector "replaces the 12VHPWR connector introduced in PCIe CEM 5.0, to address reliability concerns". Investigating RTX 4090 connector reports in 2022, NVIDIA found connectors that were not fully plugged into the card, and advised plugging the power adapter into the card before installing the card, so that it is firmly and evenly seated (NVIDIA).
• A swap. Microsoft's advice for Stop errors is to "isolate the problematic components". For the supply, that means trying a known-good unit of adequate rating.

Heat

Intel says a processor that cannot hold a safe temperature by throttling "will automatically shut down to prevent permanent damage". Dell draws a useful line: a computer that shuts down randomly but "does not feel hot" is "Possibly a power or battery problem rather than heat" (Dell). Read your temperatures under the load that ends in the restart: CPU and GPU temperatures while gaming.

Overclocks and memory profiles

For zero-code restarts, Microsoft's first two steps are to disable overclocking and to check memory: "Verify that all memory chips run at the same speed and that every chip is configured correctly in the system." An XMP or EXPO profile is memory overclocking (Is your RAM at its rated speed?). On a 13th or 14th Gen Intel Core desktop processor, also read Intel 13th and 14th gen crashes in games.

The order to check

From free to costly:

1. Read Event 41: a code, the power button, or zeros.
2. With a code, look it up, then read Event 1001, the dump and any WHEA-Logger entries.
3. Return the CPU, graphics card and memory to their default settings.
4. Read temperatures under load.
5. Check the supply's rating, and that the power connector is fully seated.
6. Test the memory.
7. Still restarting at defaults, at normal temperatures and with clean memory tests? Isolate the supply with a known-good one.

What does not help

• Treating Event 41 as the fault. It records that a shutdown went wrong, not why.
• Hunting for a driver to blame when the codes are zero: Microsoft says those symptoms typically "indicate a hardware problem".
• Switching the page file off. By default, that is where Windows writes a crash dump.

Related

• Why did my game crash?
• CPU and GPU temperatures while gaming
• Is your RAM at its rated speed?
• Intel 13th and 14th gen crashes in games
• GPU crash recovery