PC Randomly Restarting Lights Staying on - 5900x

[LouWii]

My Specs
CPU:AMD 5900X
GPU:3080TI
MOBO:MSI X570 UNIFY
RAM: Corsair Dominator Platinum 32GB 3200mhz
PSU: BeQuiet 850w Straight power
STORAGE: 3tb WD hdd, 500gb Samsung Evo 870

Errors -

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 8

A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 4

-Error setting traits on Provider {8444a4fb-d8d3-4f38-84f8-89960a1ef12f}. Error: 0xC0000001

My PC randomly reboots when playing games and once in browser (not sure if a game was open in the background ) seemingly randomly and about once a day, lights stay on no blue screen of death, started about a week ago, upgraded GPU to 3080ti around a month ago , had the cpu and mobo since the start of the year. I posted it under "CPUS" as the errors seem cpu related. I flashed my bios when I got the cpu and it's been on the same version this entire time A.70. All gpu and cpu drivers are up to date.Thanks for any help you guys can give me, let me know if you want anymore info.

 

[jigger]

Did the issue just start randomly out of the blue?

 

[LouWii]

It seems like it, the only thing I remember doing in the time before it was installing the gpu and a new exhaust fan, a little after that I also updated gpu driver twice. I've been told to upgrade PSU by a few people to around 1000w, although I would've thought 850w was plenty. I can run games stably for hours at times but other times just launching something will restart the pc

 

[jigger]

It seems like it, the only thing I remember doing in the time before it was installing the gpu and a new exhaust fan, a little after that I also updated gpu driver twice. I've been told to upgrade PSU by a few people to around 1000w, although I would've thought 850w was plenty. I can run games stably for hours at times but other times just launching something will restart the pc

It would point to the GPU and with Ampere the problem is probably heat. I had the same problem. Nvidia have pushed its RTX products way outside the ATX and Molex specs. Both in terms of heat and power use.

Take side panel off and point a desk fan at the side of the system if possible. This should give an idea if heat is the issue.

 

[LouWii]

Yeah I've ordered new thermal pads, I've got the founder's edition, although I'm still confused about the whea errors I received

 

[Tetras]

Just to rule it out I would turn off PBO and disable XMP/DOCP. It shouldn't matter if it was stable before, but the extra heat of the new GPU could contribute.

 

[jigger]

Thermal pads would improve the heat transfer and make any heat soak issue worse.

 

[LouWii]

So is there any real fix if the issue is heat? Get a beefier water cooler?

 

[jigger]

First I’d make sure that is the issue and then decide if you want to nuke the warranty. I’d personally be hesitant to pull cooler off early into its life as this is time when failures rates are highest.

 

[cliffy]

What cooling are you using, do any of the compnents seems excessively hot to the touch (RAM, Back of GPU, CPU socket Area)? Any apps running that are recording temperatures of components?

 

[LouWii]

I've got a kraken x62 cooling my cpu, stock fe on the gpu in a p500a drgb case, 3 intakes at the front, 1 exhaust out the back, the radiator is mounted at the top of the case with the fans on the inside blowing downwards, idle temps on the cpu are around 50 c with the limit being 90c, the gpu runs idle at around 50 c too, the gpu is hot to the touch and the ram is hot but not as hot to touch.

 

[Shac]

I have had the same messages when lowering the PBO voltages too far.
I would suggest running core recycler and if the cores are not stable it reports which one is at fault.
I believe the Apic ID is telling you which cores anyway.

 

[LouWii]

How do I run Core recycler?

 

[jigger]

I have had the same messages when lowering the PBO voltages too far.
I would suggest running core recycler and if the cores are not stable it reports which one is at fault.
I believe the Apic ID is telling you which cores anyway.

t’s super unlikely a CPU would randomly become unstable. Unless it’s been overclocked to within an inch of it existence.

 

[jigger]

How do I run Core recycler?

By taking the side off your case and pointing a desk fan at it.

 

[Tetras]

t’s super unlikely a CPU would randomly become unstable. Unless it’s been overclocked to within an inch of it existence.

It's unlikely, but it is possible. A 3080 Ti dumps a lot of heat in the case and with PBO on they're clocked pretty high already. Same with RAM, extra 3 - 5 degrees at high clocks can make it throw errors.

 

[LouWii]

Just to rule it out I would turn off PBO and disable XMP/DOCP. It shouldn't matter if it was stable before, but the extra heat of the new GPU could contribute.

turned off xmp and pbo, I'll wait and see if I get another restart

 

[LouWii]

Sorry If this is a stupid question but could it be voltage spikes that my 850w psu can't handle, was speaking to another guy with the same issue and after he upgraded to 1000w it seemed to resolve.

 

[Zeeflyboy]

Certainly possible, I read of many issues with transient spikes on 3090’s causing system resets until people went to beefier supplies (and even causing shutdowns on some seasonic 1000w supplies due to tripping OCP), and the 3080Ti is pretty much a 3090.

Amazingly your PSU has 4 separate 12v rails, so I’d say it’s actually pretty likely the problem, especially if you happen to be feeding the GPU from a single rail. You many want to check and re-jig to make sure you have one 12v from rail 3 and one from rail 4 as they have higher amperage than rails 1 and 2 (26A each vs 21A each).

… or just buy a decent PSU with single rail.

edit - so PCIE1 is 12V3/12V3, PCIE2 is 12V3/12V4 and PCIE3 is 12V4/12V4… about all you can do is try using PCIE 1 and 3 to evenly split the load across the connectors. Unfortunately with such a design you have 42A of 12v rail being wasted purely on CPU and accessories. You only actually have peak 26A on each of the 2x12V rails suppling the GPU, which while it should be enough in theory may be susceptible to spikes/drops with the power hungry GPU.

edit 2 - actually what I said above is recommended in the manual… so I’d double check you are using 1 and 3, if not then change over and see if it helps.

https://imgur.com/a/3S01J7h

 

[Grim5]

If the system is rebooting at idle then your cpu or ram is unstable. turn pbo off and run ram at default, not xmp.

If the system is booting under gaming load, then your gpu or psu is unstable. Undervolt gpu and try again or swap for entry level gpu or swap for better psu