Hey — a desktop computer that needs frequent restart can mean a few very different things, and the “right next step” depends on which pattern you’re seeing.

First, I want to separate two categories (because they’re diagnosed differently):
1) It restarts by itself (unexpected reboot, sometimes no warning).
2) It doesn’t restart by itself, but it becomes unstable until you restart (apps hang, internet drops, everything slows, then reboot “fixes it”).

Quick questions — answer whichever you can:
• Is it an automatic restart or a “needs reboot to behave” situation?
• Any blue screen / stop code? Or does it just go black and come back?
• When does it happen most? gaming, browsing, idle, waking from sleep, or right after Windows updates?

Also, what Windows version (10/11) and roughly how old is the desktop?

Sudden reboot with no blue screen usually points toward one of these buckets:

• Power/PSU or power delivery instability (can look like “random restarts,” especially under load).
• Overheating / thermal protection (CPU or GPU spikes, then the system reboots to protect itself).
• Driver crash / system error that Windows auto-restarts from quickly (sometimes you don’t notice the stop screen).
• Windows update / firmware change that lined up with the start of the issue.

People often search this as: “PC keeps restarting randomly”, “Windows 11 desktop restarts without warning”, “Kernel-Power 41”, “reboot loop but boots fine”.

Safe check #1: do you know if Windows is set to Automatically restart on system failure? If that’s enabled, a blue screen can flash and reboot so fast it looks like a power cut. (We’re not changing it yet — just noting it.)

Safe check #2 (purely diagnostic): can you open Reliability Monitor (View reliability history) and tell me what it shows around the reboot times — “Windows was not properly shut down,” “Hardware error,” or any repeated “critical event”?

That’s actually a great signal. “Windows was not properly shut down” typically means Windows didn’t get to close normally — it can be power loss, thermal shutdown, or a crash that didn’t cleanly log a stop code.

Let’s do safe narrowing without assuming it’s “definitely PSU” or “definitely Windows.” A few targeted questions:
• Any recent changes? New GPU, new RAM, new power strip/UPS, new drivers, Windows update, or BIOS update in the last 1–2 weeks?
• Does it reboot more under load? For example: within 5–20 minutes of a game starting, or only after hours?
• Any signs of overheating? Fans suddenly ramping, the case feels hot, or performance drops right before the reboot?

Also: when it reboots, does it go completely black instantly like someone pulled the plug, or does it freeze first, then restart?

Instant black-screen reboot is the classic “looks like power” symptom — but I don’t want to jump to conclusions. Windows updates can still be involved indirectly (driver changes), and heat can mimic power loss too.

Here are the safest “quick help” checks that don’t change your configuration in a big way:
1) Note the exact time of the next restart (down to the minute). Then we match it to logs, instead of guessing.
2) Event Viewer clue (read-only): after a reboot, check Windows Logs → System around that time for entries like Kernel-Power or a “BugCheck” event. You don’t have to interpret everything — just whether Windows logged a stop code vs “unexpected shutdown.”
3) Temperature pattern: does it happen after the PC has warmed up (gaming) and does it happen faster if the room is warm? (Overheating tends to be time-and-load sensitive.)

Quick risk check: is your desktop overclocked in any way? (CPU/GPU overclock, or RAM XMP/EXPO profile enabled). Even “default XMP” can be a stability clue when random restarts begin.

That’s a smart caution. We can get meaningful answers before touching BIOS. XMP is just a data point right now — not “the cause,” but a stability factor to keep in mind if logs point toward memory-related crashes.

A few more safe questions that help separate categories:
• Do you ever see a blue screen for a split second? Even once? If yes, that pushes us toward driver/stop code analysis.
• Any pattern with sleep/wake? Some systems restart when waking due to driver/power-state issues.
• Is the PC plugged directly into the wall or through a power strip/UPS? (Not blaming it — just checking for power delivery variables.)

Related search terms that match this stage: “Windows 11 random reboot no BSOD”, “Kernel-Power Event ID 41”, “PC restarts under load”, “desktop rebooting while gaming”.

If you find a “BugCheck” entry or any stop code text, even just the number, that’s extremely useful — it tells us whether Windows detected a crash vs a power cut.

That lines up with the “unexpected shutdown / Kernel-Power” story. Important nuance: Kernel-Power is usually a result (Windows noticed the shutdown was unclean), not a precise cause by itself. So we use it to confirm the type of event, then we look for what happened right before it.

The safest next diagnostic step is: in that same System log, look a minute or two before the Kernel-Power entry. If you see driver warnings, disk/controller errors, display driver resets, or a “BugCheck,” that shifts the diagnosis away from pure power and toward a specific subsystem.

Two very low-risk tests you can do (optional) that don’t “change everything”:
• Wall test: if possible, plug the desktop directly into a wall outlet for a short test window. If restarts stop during that window, it’s a useful clue about power delivery (still not a final verdict).
• Light-load vs heavy-load timing: note whether it can run for hours on light use but fails within minutes of a game. That pattern is strongly diagnostic for thermal or power headroom issues.

And because you mentioned XMP: if logs hint at memory instability, a built-in memory diagnostic can be used as a signal (pass/fail trend) — not a guaranteed answer, but it helps decide what to investigate next.

Perfect approach. If you capture these, you’ll have enough evidence to choose a safe next step (instead of chasing random “fixes”):

1) Exact restart times (write the clock time when it rebooted).
2) Reliability Monitor details for the critical events (especially anything labeled “Hardware error” or repeated critical failures).
3) Event Viewer System log around the restart: the Kernel-Power entry plus the 5–10 events right before it (look for “BugCheck,” display driver messages, disk/controller errors, or update/driver install notes).
4) Trigger context: what was happening (game name, video playback, idle), and roughly how long the PC had been running.
5) Environment clues: room temperature, fan noise, and whether it happened sooner under heavy load.

With that, you can safely decide whether this looks like a stop code/driver path, a thermal pattern, a power delivery pattern, or a memory stability signal — without immediately changing BIOS settings or reinstalling Windows.