Unexplainable reboots on extreme high end system

[aurrorax]

Hi there,

 

I recently build a machine for a good friend of mine, top of the line system.

He has been having an issue with the system from the start where, without warning the PC crashes to black screen and reboots itself, temps all fine.

This crash occurs randomly, idle at the desktop, whilst gaming, under heavy graphical work, can't hard reproduce it.

 

I have reseated all the power cables and made sure the BIOS is up to date.

I'm leaning towards a faulty power supply due to above symptom, however since its a new Seasonic model its hard to find similair stories about this issue.

 

Anyone encountered something similair?

 

Specs : 

Intel Core i9 13900KS

Gigabyte Z790 AORUS XTREME

Gigabyte RTX 4090 Aorus Master (connected with 12VHPWR connector)

2x 48GB Corsair Dominator titanium 6600 Mhz CL32

3x Samsung 990 Pro 4TB

EKWB EK-Nucleus AIO CR360

Seasonic Prime PX-1600 ATX 3.0

3x Noctua NF-F12 PWM

4x Noctua NF-A14 PWM

Fractal Design Torrent (relevant because we are using built in fan controller)

 

Love to hear if anyone else has encountered something similair

 

 

 

 

[Catzzye]

Can you check Event Logger for errors? Might give us more info to work with

[aurrorax]

7 minutes ago, Catzzye said:

Can you check Event Logger for errors? Might give us more info to work with

 

Basic kernel power error

[Catzzye]

Yeah, honestly hard to gather a lot of info from that.

I can't tell you much more, other than that the power supply should be pretty high quality and should be able to sustain those workloads, unless it's faulty as you say.

You can try contacting Seasonic for replacement

[overbuilt_gaming]

How much is that side panel pressing up against the GPU connector?  From what I can tell in the pics of the torrent and the GPU you got (very wide) I would say it's likely.  This might be shorting out the power connector on the GPU, and this is how the fires from last year all started.  This would cause your unexpected reboots.  Nvidia said the customer didn't connect it right and everybody is like, oh okay.  Well, I argue the side panel presses up against the connector causing it to come dislodged or connected unevenly. 

 

If it's even touching it a little have your friend take the side panel off and make darn sure that connector is all the way in.   Maybe try using the nvidia connector and use the 4 eps connectors to the psu to test.

[aurrorax]

11 hours ago, overbuilt_gaming said:

How much is that side panel pressing up against the GPU connector?  From what I can tell in the pics of the torrent and the GPU you got (very wide) I would say it's likely.  This might be shorting out the power connector on the GPU, and this is how the fires from last year all started.  This would cause your unexpected reboots.  Nvidia said the customer didn't connect it right and everybody is like, oh okay.  Well, I argue the side panel presses up against the connector causing it to come dislodged or connected unevenly. 

 

If it's even touching it a little have your friend take the side panel off and make darn sure that connector is all the way in.   Maybe try using the nvidia connector and use the 4 eps connectors to the psu to test.

Not touching at all.

[Whatisthis]

12 hours ago, aurrorax said:

 

Basic kernel power error

Filter the log only for error and critical and repost. The warnings are just spam.

[aurrorax]

27 minutes ago, Whatisthis said:

Filter the log only for error and critical and repost. The warnings are just spam.

[Whatisthis]

1 hour ago, aurrorax said:

Ok, not pointing to anything specific. Seems like a hardware fault.

 

can you run hwinfo64 in logging mode and post a log of the crash?  If it is voltage or temperature related hwinfo will catch it.

 

edit - before you do the hwinfo, Go to C:\Windows\Minidump. See if there is a dump file there. If so, post it here so someone can look thru it and find the reason for the dump.

[aurrorax]

1 hour ago, Whatisthis said:

Ok, not pointing to anything specific. Seems like a hardware fault.

 

can you run hwinfo64 in logging mode and post a log of the crash?  If it is voltage or temperature related hwinfo will catch it.

 

edit - before you do the hwinfo, Go to C:\Windows\Minidump. See if there is a dump file there. If so, post it here so someone can look thru it and find the reason for the dump.

No dump file present, see CSV for HWinfo logs

Broken.CSV

[Whatisthis]

4 hours ago, aurrorax said:

No dump file present, see CSV for HWinfo logs

Broken.CSV 159.1 kB · 2 downloads

No throttling, no temp issues, solid power rails. The only one thing I see is that you have a decent bit of vdroop (core voltage dips with load) of about 0.05v. It may or may not be the problem, but it’s easy to fix to see.

 

Go into bios and in the cpu voltage section there should be a load line calibration option. Put that on the 3rd or 2nd highest option. It will prevent the voltage drop. See if that fixes things.

 

Edit - scratch that. If you get the issue at idle too, vdroop ain’t it. 
 

This one is going to be hardware roulette…one of the components is causing the psu to power cycle.

[aurrorax]

13 hours ago, Whatisthis said:

No throttling, no temp issues, solid power rails. The only one thing I see is that you have a decent bit of vdroop (core voltage dips with load) of about 0.05v. It may or may not be the problem, but it’s easy to fix to see.

 

Go into bios and in the cpu voltage section there should be a load line calibration option. Put that on the 3rd or 2nd highest option. It will prevent the voltage drop. See if that fixes things.

 

Edit - scratch that. If you get the issue at idle too, vdroop ain’t it. 
 

This one is going to be hardware roulette…one of the components is causing the psu to power cycle.

Yep, I was fearing this too.

 

I've spent the better part of 4 hours on it with my friend yesterday.

Can't hard reproduce the issue, it seems total at random, small sidenote, SEEMS to happen more often under combined CPU/GPU load however this may be coincidence...

 

From my knowledge, if its PSU, it should turn off and not turn on automatically again.

If its memory, this could be the issue, however we tried different RAM.

 

GPU performs really well under load where I would expect a crash, so does CPU.

 

 

SO we are both leaning towards the darn 1000 euro motherboard being bad.

 

This one is going to be RMA first...

 

 

 

[aurrorax]

On 11/19/2023 at 8:59 PM, Whatisthis said:

No throttling, no temp issues, solid power rails. The only one thing I see is that you have a decent bit of vdroop (core voltage dips with load) of about 0.05v. It may or may not be the problem, but it’s easy to fix to see.

 

Go into bios and in the cpu voltage section there should be a load line calibration option. Put that on the 3rd or 2nd highest option. It will prevent the voltage drop. See if that fixes things.

 

Edit - scratch that. If you get the issue at idle too, vdroop ain’t it. 
 

This one is going to be hardware roulette…one of the components is causing the psu to power cycle.

So as a try we did the load line calibration adjustment running the LINPACK test it basically consistently crashes within a minute.

Reviewing OCCT personal test also shows CPU error.

 

 

 

 

[Whatisthis]

5 hours ago, aurrorax said:

So as a try we did the load line calibration adjustment running the LINPACK test it basically consistently crashes within a minute.

Reviewing OCCT personal test also shows CPU error.

 

 

 

 

It’s all the same core and logical processor. Looks like possibly a bad cpu. 

[aurrorax]

12 hours ago, Whatisthis said:

It’s all the same core and logical processor. Looks like possibly a bad cpu. 

Yep, its in the RMA mail now, I'll update when I know more.

 

Thanks so far for the help.

[aurrorax]

On 11/21/2023 at 2:51 AM, Whatisthis said:

It’s all the same core and logical processor. Looks like possibly a bad cpu. 

Got a brand new CPU, same issue, the computer seems possessed at this point.

[aurrorax]

On 11/21/2023 at 2:51 AM, Whatisthis said:

It’s all the same core and logical processor. Looks like possibly a bad cpu. 

Update,

 

GPU , not the issue, tested with IGPU, still crashed

CPU, has been swapped after RMA, still crashed

PSU, got an RMA swapped one, still crashed 

RAM , is incoming in the mail to be swapped. has been swapped, issue persists..

Mobo,  ordered a more "reasonable motherboard the Z790 Aorus master and the XTREME will go back, as this is the last part it must be the issue.

 

FUCK ME this has been a haunted system if I have ever seen one

[aurrorax]

On 11/21/2023 at 2:51 AM, Whatisthis said:

It’s all the same core and logical processor. Looks like possibly a bad cpu. 

Swapped the board, was able to run furmark and OCCT CPU TEST for 30m without issues.

 

haunted motherboard it was!

[Whatisthis]

7 minutes ago, aurrorax said:

Swapped the board, was able to run furmark and OCCT CPU TEST for 30m without issues.

 

haunted motherboard it was!

Glad you got it fixed. Happy gaming!