Hard Resetting - Kernel Power

[mrtherm]

Alright folks i need some help trying to identify a very annoying problem.

 

The Problem:

- Hard PC reset with no warnings or BSOD

 

Occurs:

- Heavy games - Palworld & Red Dead Redemption 2

- Does not do it on Destiny 2, BeamNG Drive

- Has done it randomly on idle

- Hard to manually trigger the Problem.

 

Troubleshooting I've Done

- turned off Automatic Restart to try see info

- Memtest & Windows Memory Check Passed

- Brand new PSU

- Stress test GPU & CPU

- Hard drive checks

- Updated & Reset BIOS

- Fresh Windows Install

- I've unplugged an RGB strip and turned off 1 of 4 Noctua case fans (in case there's some sort of power limit reach?)

- Stock clocks on all hardware

- Thermals all seem fine. CPU max temps (celsius) ~80 / GPU max ~70

 

Event Viewer

- Critical Kernel Power error, Incorrectly shut down etc.

 

PC Specs

Ryzen 5900X

Gigabyte X570 AORUS PRO WIFI

MSI RTX 2080 Super

Noctua NH-U12S

2x Corsair Vengeance (4x16GB) 3200MHz CL16 DDR4
CORSAIR RM850x

Focusrite Scarlett 2i4 USB Interface

 

My Thoughts

- at first i thought a dodgy capacitor on the PSU under load, brand new PSU this weekend did the same thing.

- Could it be dodgy VRAM? How can i run a test for that?

- will reseat cpu cooler & refresh thermal paste (my understanding is a thermal shudown, would shutdown, not restart)

- Would a dying usb audio interface trigger a hard reset?

 

I'm open to ideas on how to identify this.

[Bjoolz]

Open Event Viewer → Windows Logs → System. On the right hand side, select "Filter Current Log". In the new window that pops up find the Event Sources dropdown menu and select "WHEA-logger". Click Ok to apply the filter. If you have any WHEA events, highlight them, right click and save. Upload the .evtx file to the forum.

 

If you don't have any WHEA events, the main suspect would be power. 

[191x7]

PSU test in OCCT? Maybe a combined CPU load + GPU load like Core Cycler or Prime95 combined with Furmark?

Doesn't crash under Furmark? How about 3Dmark Stability test?

Normal 3DMark scores?

Any undervolts or overclocks applied?

Triggering VRM protections can cause both shutdowns or hard resets without a BSOD.

Latest SSD firmware? SSD temperatures?

[mrtherm]

8 hours ago, Bjoolz said:

Open Event Viewer → Windows Logs → System. On the right hand side, select "Filter Current Log". In the new window that pops up find the Event Sources dropdown menu and select "WHEA-logger". Click Ok to apply the filter. If you have any WHEA events, highlight them, right click and save. Upload the .evtx file to the forum.

 

If you don't have any WHEA events, the main suspect would be power. 

Here we go,

 

A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 7 (and a 6 some of them)

mrtherm-filtered-whea.evtx

[Bjoolz]

1 hour ago, mrtherm said:

Here we go,

 

A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 7 (and a 6 some of them)

mrtherm-filtered-whea.evtx 68 kB · 0 downloads

The info in the general tab is what reported the error, not what caused the error. To see what caused the error you have to decode the error packet in the RawData fields using the UEFI documentation.

 

The CPU is ordering the shutdowns. These show up as "Previous Error", meaning it's not from the actual error, just a notification that it shut down the PC from the previous error. We can't see the original error with these issues. 90% of the time it's the CPU, but it can also be a PCIe device or even the motherboard is the CPU/PCIe section is faulty.

 

We have seen a lot of voltage issues on 5000 series so we could try setting a static voltage which has fixed a lot of 5900x and 5950x CPUs that have had lots of different CPU related crashes. In the BIOS, set a static voltage to the CPU of 1.3v and to the SoC set a static voltage of 1.1v (CPU Vcore and VCORE SOC on this motherboard). 

[mrtherm]

1 hour ago, Bjoolz said:

The info in the general tab is what reported the error, not what caused the error. To see what caused the error you have to decode the error packet in the RawData fields using the UEFI documentation.

 

The CPU is ordering the shutdowns. These show up as "Previous Error", meaning it's not from the actual error, just a notification that it shut down the PC from the previous error. We can't see the original error with these issues. 90% of the time it's the CPU, but it can also be a PCIe device or even the motherboard is the CPU/PCIe section is faulty.

 

We have seen a lot of voltage issues on 5000 series so we could try setting a static voltage which has fixed a lot of 5900x and 5950x CPUs that have had lots of different CPU related crashes. In the BIOS, set a static voltage to the CPU of 1.3v and to the SoC set a static voltage of 1.1v (CPU Vcore and VCORE SOC on this motherboard). 

Tried those values and failed on boot. Had to load optimized defaults.

[Bjoolz]

16 minutes ago, mrtherm said:

Tried those values and failed on boot. Had to load optimized defaults.

I don't have any suggestions then. I'm not into overclocking or anything so messing with voltages is a bit outside my wheelhouse.

 

I did see that I forgot telling you to make sure that Precision Boost Overdrive (PBO) is set as Disabled in the BIOS. You could also try the voltage tweak again after disabling it if it was enabled. 

[mrtherm]

1 minute ago, Bjoolz said:

I don't have any suggestions then. I'm not into overclocking or anything so messing with voltages is a bit outside my wheelhouse.

 

I did see that I forgot telling you to make sure that Precision Boost Overdrive (PBO) is set as Disabled in the BIOS. You could also try the voltage tweak again after disabling it if it was enabled. 

yeah will do, thanks anyway.

It seems like a very annoying issue and it's very random.

[Bjoolz]

4 hours ago, mrtherm said:

yeah will do, thanks anyway.

It seems like a very annoying issue and it's very random.

And just to clarify, the main suspect from the get-go was a faulty CPU so this was just for trying to bring it back into stability. So if nothing works, the CPU is likely faulty. Though it could be a PCIe device (And remember that most components on the motherboard connect over PCIe internally. Audio, network, storage controllers, etc).

[emothxughts]

22 hours ago, mrtherm said:

Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error

I once had this kind of self restart error too, I blamed my then new Ryzen 3600 to the point of RMA-ing it. It still happened with the replacement CPU.

 

The ultimate culprit turned out to be my GPU at the time, an RX 5600 XT bought used. See if you or a friend has a GPU to spare, swap out your current card, see if it still restarts.

[mrtherm]

On 6/12/2025 at 6:26 PM, emothxughts said:

I once had this kind of self restart error too, I blamed my then new Ryzen 3600 to the point of RMA-ing it. It still happened with the replacement CPU.

 

The ultimate culprit turned out to be my GPU at the time, an RX 5600 XT bought used. See if you or a friend has a GPU to spare, swap out your current card, see if it still restarts.

I had a Ryzen 2600X that i'm currently testing with.

Backup GPU is ready as well.

 

Will pretty much swap out as many hardware i have backup to accurately figure it out.

So far hasn't crashed. Will keep testing/running games.

 

On a side note, still getting this with the changed CPU??

"The device driver for the Trusted Platform Module (TPM) encountered a non-recoverable error in the TPM hardware, which prevents TPM services (such as data encryption) from being used. For further help, please contact the computer manufacturer."

[mrtherm]

Update.

 

Been running an old 2600X without any issues thus far.