Question PC shuts down during games after a few minutes – PSU replaced, no change

[deus_lexx]

Hi everyone,
I’m struggling with a persistent issue and would really appreciate any advice.

Build Info:

• CPU: AMD Ryzen 5 5600X
• GPU: Gigabyte Radeon RX 6650 XT
• Motherboard: MSI MAG B550 Mortar Max WiFi
• RAM: 2x8GB DDR4-3200 Patriot Memory
• SSD: WD SN570 250GB (boot) + 1TB (storage)
• PSU (original): Chieftec Task TPS-700S (700W Bronze)
• PSU (current): Thermaltake GF 750W Gold
• Case: Ginzzu CL220
• CPU Cooler: ID-Cooling SE-224-XTS ARGB WHITE
• OS: Originally Windows 11 → downgraded to Windows 10 (no change)

The issue:

Ever since I built this PC two years ago, it has occasionally shut down instantly during games, like someone pulled the power cable. No BSOD, no restart, just a sudden power loss.

In most modern games (like Baldur's Gate 3 or Space Marine 2), shutdowns happen irregularly — sometimes after 10 minutes, sometimes after 2 hours. I managed by quicksaving often and pushing through.

However, I recently started Clair Obscur: Expedition 33, and now the system shuts down reliably within 5 minutes of gameplay, making it unplayable.

Once, it also shut down while generating an image with Stable Diffusion (using GPU acceleration), which was the only non-gaming-related crash I’ve had so far.

What I’ve tried so far:

• Replaced PSU: from Chieftec to Thermaltake Toughpower GF 750W – no effect.
• Stress testing: Prime95, FurMark, OCCT, and AIDA64 all pass with no issues.
• Thermals OK: CPU stays below 89°C, GPU below 55°C under load.
• Power logging with HWiNFO64: +12V holds steady at 11.95V, +5V at 5.09V during shutdowns.
• MSI Afterburner: set GPU power limit to -6% – no effect.
• Updated BIOS to the latest version – no improvement.
• Windows downgrade (11 → 10) – didn’t help.

Symptoms:

• Instant shutdown (no warning, no BSOD)
• Event Viewer only shows generic “unexpected shutdown” (Event ID 6008)
• GPU fans ramp up audibly before the crash and also you can hear a little tsk-tsk sound.
• Only occurs under GPU-intensive load (games or Stable Diffusion)
• Never crashes during CPU-only workloads or stress tests

Any help or ideas would be greatly appreciated — I’m ready to troubleshoot anything at this point.

 

[logainofhades]

Remove that 2400 ram. Ram should always be in matched sets, never mixed like you have.

 

[deus_lexx]

Remove that 2400 ram. Ram should always be in matched sets, never mixed like you have.

Already did, forgot to edit that part. The issue persists unfortunately

 

[logainofhades]

Have you done a CMOS reset?

 

[deus_lexx]

Have you done a CMOS reset?

No, not yet. Should I try it?

 

[logainofhades]

No, not yet. Should I try it?

Yes, especially since you updated the bios.

 

[Misgar]

RAM: 2x8GB DDR4-3200 Patriot Memory

Have you tried testing your RAM at 3200MT/s (overclocked) and also at 2133/2400MT/s (no overclock) with MemTest86. Bad system RAM can cause sudden crashes. If you get any errors after performing a full run (several hours), you need to take action.
https://www.memtest86.com/

MSI Afterburner: set GPU power limit to -6% – no effect.

This fixed a problem I had with Topaz Video AI until they sorted out the bugs in their software. I set GPU max power to 95% in Afterburner and the system didn't crash (7950X, RTX3060).

GPU: Gigabyte Radeon RX 6650 XT

Are you running the latest graphics driver dated 8-5-2025?
https://www.amd.com/en/support/down...eon-rx-6000-series/amd-radeon-rx-6650-xt.html

Since the problem seems related to the GPU, it might be worth running DDU and then reinstalling the driver, but by trying Windows 10, you've probably got this covered.
https://www.guru3d.com/download/display-driver-uninstaller-download/

I don't suppose you could try a different GPU card?

 

[deus_lexx]

Yes, especially since you updated the bios.

Just did the reset (removed the battery and all) – unfortunately, the shutdowns are still happening. Thanks for the suggestion though

 

[deus_lexx]

Have you tried testing your RAM at 3200MT/s (overclocked) and also at 2133/2400MT/s (no overclock) with MemTest86. Bad system RAM can cause sudden crashes. If you get any errors after performing a full run (several hours), you need to take action.
https://www.memtest86.com/

Gonna try it, thanks

 

[deus_lexx]

I don't suppose you could try a different GPU card?

Don't have one at hand unfortunately, but I'll ask around

 

[DaleH]

Hi everyone,
I’m struggling with a persistent issue and would really appreciate any advice.

Build Info:

• CPU: AMD Ryzen 5 5600X
• GPU: Gigabyte Radeon RX 6650 XT
• Motherboard: MSI MAG B550 Mortar Max WiFi
• RAM: 2x8GB DDR4-3200 Patriot Memory
• SSD: WD SN570 250GB (boot) + 1TB (storage)
• PSU (original): Chieftec Task TPS-700S (700W Bronze)
• PSU (current): Thermaltake GF 750W Gold
• Case: Ginzzu CL220
• CPU Cooler: ID-Cooling SE-224-XTS ARGB WHITE
• OS: Originally Windows 11 → downgraded to Windows 10 (no change)

The issue:

Ever since I built this PC two years ago, it has occasionally shut down instantly during games, like someone pulled the power cable. No BSOD, no restart, just a sudden power loss.

In most modern games (like Baldur's Gate 3 or Space Marine 2), shutdowns happen irregularly — sometimes after 10 minutes, sometimes after 2 hours. I managed by quicksaving often and pushing through.

However, I recently started Clair Obscur: Expedition 33, and now the system shuts down reliably within 5 minutes of gameplay, making it unplayable.

Once, it also shut down while generating an image with Stable Diffusion (using GPU acceleration), which was the only non-gaming-related crash I’ve had so far.

What I’ve tried so far:

• Replaced PSU: from Chieftec to Thermaltake Toughpower GF 750W – no effect.
• Stress testing: Prime95, FurMark, OCCT, and AIDA64 all pass with no issues.
• Thermals OK: CPU stays below 89°C, GPU below 55°C under load.
• Power logging with HWiNFO64: +12V holds steady at 11.95V, +5V at 5.09V during shutdowns.
• MSI Afterburner: set GPU power limit to -6% – no effect.
• Updated BIOS to the latest version – no improvement.
• Windows downgrade (11 → 10) – didn’t help.

Symptoms:

• Instant shutdown (no warning, no BSOD)
• Event Viewer only shows generic “unexpected shutdown” (Event ID 6008)
• GPU fans ramp up audibly before the crash and also you can hear a little tsk-tsk sound.
• Only occurs under GPU-intensive load (games or Stable Diffusion)
• Never crashes during CPU-only workloads or stress tests

Any help or ideas would be greatly appreciated — I’m ready to troubleshoot anything at this point.

I'd still be concerned about the temps. Try to reproduce the problem by opening the case and having an external fan blowing on the motherboard. Let us know the results.

 

[hotaru251]

I'd still be concerned about the temps.

temps are perfectly fine.The ryzen 5x00 and newer are all rated to be able to run at 95c 24/7 before issues.(thoguh obviously you dont want to run that hot but its within spec)


I'd try to test and find out exactly what is causing issue.

Lower system ram to default spec (as in out of box not the "oc" that it says on box) and see if it happens. (if it does then your rams not issue and put it back to normal)

Lower GPU clock, power limit, and memory 1 at a time and test each and see if it is issue then if it still is revert to normal and change other.

 

[DaleH]

temps are perfectly fine.The ryzen 5x00 and newer are all rated to be able to run at 95c 24/7 before issues.(thoguh obviously you dont want to run that hot but its within spec)


I'd try to test and find out exactly what is causing issue.

Lower system ram to default spec (as in out of box not the "oc" that it says on box) and see if it happens. (if it does then your rams not issue and put it back to normal)

Lower GPU clock, power limit, and memory 1 at a time and test each and see if it is issue then if it still is revert to normal and change other.

You may be right about the temperatures, but I would want proof other than what is printed on the box it came in. Parts degrade with time and temperature and even new parts don't always meet the advertised specifications. Thus, if I wanted to eliminate temperature as a problem, an easy test would be to just cool it more to see if it affected your problem.

 

[hotaru251]

Thus, if I wanted to eliminate temperature as a problem, an easy test would be to just cool it more to see if it affected your problem.

modern cpu/gpu have failsafes (unless flaw in specific batch) to if they got too hot they will downclock to prevent any damage. if ti was thermal issue they'd get worse performance not just straight shutdowns.

 

[DaleH]

modern cpu/gpu have failsafes (unless flaw in specific batch) to if they got too hot they will downclock to prevent any damage. if ti was thermal issue they'd get worse performance not just straight shutdowns.

You may be right, but how hard is it to prove or disprove that heat is a problem? Are you saying that you're sure that the symptoms of overheating would never be shutdowns? When presented with a chance to prove or disprove a theory, I opt for the easiest test. It's possible that some component other than the cpu is heat sensitive. Cooling it could help prove or disprove it. It's probably taken longer for this discussion than the test would take.