PC Randomly turns off while gaming

[Azureblood2]

OS: Windows 11 Pro 23H2

MB: B650M AORUS ELITE AX (rev. 1.3)
BIOS: F30 / F32C

Build: https://pcpartpicker.com/list/TWV9xH

Software running: Steam Big Picture / Final Fantasy XIV

 

PC will randomly hard powers off while gaming, no error messages shown or logged in Event Viewer (that I can find. Just a click, fans stop spinning, and USB audio makes a continuous tone until I hold the power button for 5-10 seconds. 

 

I've had it happen three times in short succession, but if it happens, it is usually once or twice a night. Some notable times that It has happened is when playing with the FSR slider in game and when pressing the PS button to bring up steam big picture in game. At first I thought it might be a transient GPU power spike tripping the PSU, since I am cutting it close...  but while the GPU LEDs turn off, all of the fan LEDs remain on.

I have a WigiDash on my desk, and the GPU is usually pulling 347-398W (depending on power slider) and the CPU is around 60W with occasional spikes to ~85W.

Things I've tried to resolve issue:

• GPU Power slider from 115% > +100%
• Changing GPU Memory timing from 'Fast' to 'Default'
• Setting Max GPU clock speed to 2860
• Updating Graphics drivers to (Latest)
• Updating BIOS from F30 > F32C (Latest)

I'm thinking it has something to do with the GPU.  At one point the LEDs on the GPU didn't come back on and my FPS  was at 15 instead of 110-130.  Restarting the system a couple of times the LEDs and FPS returned to normal.  I'm at a loss for what could be causing this, but I figured I'd ask here before pulling out the parts cannon.

Can the PSU trip OCP but keep the motherboard / LEDs powered?

[SieteGL]

29 minutes ago, Azureblood2 said:

OS: Windows 11 Pro 23H2

MB: B650M AORUS ELITE AX (rev. 1.3)
BIOS: F30 / F32C

Build: https://pcpartpicker.com/list/TWV9xH

Software running: Steam Big Picture / Final Fantasy XIV

 

PC will randomly hard powers off while gaming, no error messages shown or logged in Event Viewer (that I can find. Just a click, fans stop spinning, and USB audio makes a continuous tone until I hold the power button for 5-10 seconds. 

 

I've had it happen three times in short succession, but if it happens, it is usually once or twice a night. Some notable times that It has happened is when playing with the FSR slider in game and when pressing the PS button to bring up steam big picture in game. At first I thought it might be a transient GPU power spike tripping the PSU, since I am cutting it close...  but while the GPU LEDs turn off, all of the fan LEDs remain on.

I have a WigiDash on my desk, and the GPU is usually pulling 347-398W (depending on power slider) and the CPU is around 60W with occasional spikes to ~85W.

Things I've tried to resolve issue:

• GPU Power slider from 115% > +100%
• Changing GPU Memory timing from 'Fast' to 'Default'
• Setting Max GPU clock speed to 2860
• Updating Graphics drivers to (Latest)
• Updating BIOS from F30 > F32C (Latest)

I'm thinking it has something to do with the GPU.  At one point the LEDs on the GPU didn't come back on and my FPS  was at 15 instead of 110-130.  Restarting the system a couple of times the LEDs and FPS returned to normal.  I'm at a loss for what could be causing this, but I figured I'd ask here before pulling out the parts cannon.

Can the PSU trip OCP but keep the motherboard / LEDs powered?

First, you could start by checking the PSU, making sure it's properly connected to the motherboard and the GPU. Maybe something is loose. Another thing you could do is test the RAM, one module at a time, to see if one is failing under heavy load. I’ve experienced something similar, where it worked fine under light loads but would fail and shut everything down when I started playing games.
 

[johndms]

1 hour ago, Azureblood2 said:

Build: https://pcpartpicker.com/list/TWV9xH

My troubleshooting suggestion below is based on my own experience with the 750 watt version of your PSU. I'll save you from all the details, many of which go far over my head. SeaSonic denies our particular model is one of those affected by a "noise in the 12V sense wire from the GPU" shutdown problem. Supposedly the PRIME versions made in 2017 had this issue, but not the FOCUS. If your PSU was manufactured after 2017, then we likely have different problems. I purchased mine in January 2018.

 

I am able to recreate a scenario that is guaranteed to shut my PC down, as if I'd flipped the switch on the PSU, by running a "Small FFT's on all cores" CPU stress test with Prime95 while simultaneously running a Furmark GPU stress test and pressing the Space bar every few seconds causing the donut to disappear and reappear. (Space bar pauses the test, pressing it again resumes.. it'll make sense if you do it). As soon as I resume the Furmark test, instant crash.

 

What worked for me may not work for you, but it can't hurt to try. Simply swap which "CPU/PCI-E" port your GPU power cable is plugged into on the PSU. Mine has five ports to choose from, one of them I've noted as bad. After swapping, I've had no crashes. Hopefully something I've said may point you in the right direction. Good luck.

[HoneyBadger84]

One way to check right away if it's a psu/gpu issue, after you make sure all your cables are properly seated, use Afterburner or your app of choice & set your GPU power limit down to 80 or 75%. If the issue doesn't happen with it set like that, it's either the GPU or the PSU. Only way to determine for sure which it is would be either replacing the PSU or trying to run A GPU stress test loop with nothing else stressed on the system & see if the problem persists. 

From the way you describe the system acting when the problem happens, I'm inclined to think it's power supply not handling power spikes. Running something that's high fps with high power draw for a stresser should do the trick. 

Your PSU is quality (Seasonic make great units), is it possible it might've been damaged by a surge during a storm or something, or do you have it on battery backup/protection? 

[Mark Kaine]

3 hours ago, johndms said:

not the FOCUS

focus was notorious for having issues with 3000 series gpus at least...

[johndms]

1 hour ago, Mark Kaine said:

focus was notorious for having issues with 3000 series gpus at least...

That's what I tried telling SeaSonic support, but they disagree saying the focus has no such issues. Now that I'm able to reproduce the shutdown using the Prime95/Furmark method they recommended, I could probably successfully apply for RMA. I'm procrastinating now that it's working properly using a different port.

[Azureblood2]

Thank you for the replies!

So I tried the test as outlined by @johndms, and I wasn't able to induce a crash.  I turned up the power slider to +15% (Max) and still wasn't able to induce a crash. My power supply is definitely much newer than 2018, but I was using the 'bad' PCIE connector for my CPU, the GPU is using the three lower ones.

I also tried  @HoneyBadger84 's suggestion and was unable to induce the failure with the power at -10% (Min), but it is difficult to prove a negative.

I did pull all 4 PCIE connectors and re-seat them, making sure not to put the CPU power in the 'Bad' connector... so I guess we just wait and see if that resolves issue.  If it does, then I may just pull the trigger on a 1000W PSU, but either way I'll keep this thread updated

[HoneyBadger84]

6 hours ago, Azureblood2 said:

Thank you for the replies!

So I tried the test as outlined by @johndms, and I wasn't able to induce a crash.  I turned up the power slider to +15% (Max) and still wasn't able to induce a crash. My power supply is definitely much newer than 2018, but I was using the 'bad' PCIE connector for my CPU, the GPU is using the three lower ones.

I also tried  @HoneyBadger84 's suggestion and was unable to induce the failure with the power at -10% (Min), but it is difficult to prove a negative.

I did pull all 4 PCIE connectors and re-seat them, making sure not to put the CPU power in the 'Bad' connector... so I guess we just wait and see if that resolves issue.  If it does, then I may just pull the trigger on a 1000W PSU, but either way I'll keep this thread updated

Yeah unfortunately when it's a PSU issue it's really hard to know for sure, you pretty much have to just use the system as normal & see if the problem persists, hopefully the wire plug swaps fixed it.
Good luck, hope it's resolved.

[BetteBalterZen]

The crashes you describe OP can also be RAM related. I know this from my own system. 

If I had your PC, I would use some sort of RAM tester, let it run for at least 2 hours, see if any errors pop up. 

Personally I use MemTest86 and HCI MemTest Pro (HCI MemTest can be used in Windows).