Think I fried my 5800x?

[Dabba]

So last week I overheated my cpu while gaming. Being the dumb dumb I am I let this happen several times. It was always a bsod followed by a reboot with a bios screen telling me as such. Since then I've had intermittently had BSOD referencing AMDauth. (Not home can provide dump file later). This mostly occurred after closing a gaming/streaming session. But sometimes during.

This far I've:
- Updated chipset drives and bios
- Ran P95 for 13 hours no errors, but still BSOD.
- Have tried PBO disabled and enabled. Currently on stock bios settings aside from manually setting ram clock speed (3600 which is not OC, for some reason it automatically set to 2133mhz)
-It has not overheated since. Have had the case side off with a fan blowing in.
- Windows memtest negative
- Ran sfc - corrupted files repaired. Still bsod.

Plan:
- Will do p95 for 24 hours

Mobo is asus tuf gaming x570 with wifi. 64gig ddr4 ram at 3600. Gtx4080ti

Assuming the memtest and p95 are negative for 24 hours. Can it still be the cpu with this amdauth error? I mean it all started after the overheat. Not sure why it would crash gaming but not with p95. Also considering some voltage issuses? I remember when I first built it I believe I had to do something with the voltage but I don't remember.

Any more testing to rule out the cpu? I'd like to try to avoid spending 300 bucks to find out it's not the cpu.

That being said. If needed, replace with same cpu? 5800x3d? Something else? Cooling is being upgraded to an aio from air.

Thanks.


DMP file debug:

Debug

WHEA_UNCORRECTABLE_ERROR (124) A fatal hardware error has occurred. Parameter 1 identifies the type of error source that reported the error. Parameter 2 holds the address of the nt!_WHEA_ERROR_RECORD structure that describes the error condition. Try !errrec Address of the nt!_WHEA_ERROR_RECORD st...

docs.google.com

MODULE_NAME: AuthenticAMD

IMAGE_NAME: AuthenticAMD.sys

STACK_COMMAND: .cxr; .ecxr ; kb

FAILURE_BUCKET_ID: 0x124_0_AuthenticAMD_PROCESSOR_CACHE_IMAGE_AuthenticAMD.sys

OS_VERSION: 10.0.19041.1

BUILDLAB_STR: vb_release

OSPLATFORM_TYPE: x64

OSNAME: Windows 10

FAILURE_ID_HASH: {d6c4d869-cd29-1e48-0267-5766712959d2}

 

[EarthDog]

I'm not at all convinced your CPU is borked...yet. Chips have built in throttling and a hard cutoff point to prevent damage

What were your temps when these BSODs happened (and what BSOD says 'temps' were the cause?)?

What test(s) are you running in P95?

What power supply do you have?

64GB of RAM is alot... I'll assume you use more than 32GB doing something else? Otherwise, try running two sticks at a time.

Before you spend money on a new CPU (I'd get a 7800X3D and upgrade - if the CPU is bad, you may be able to RMA?), for giggles, reinstall your OS and see if that takes care of the problem.

 

[Dabba]

I'm not at all convinced your CPU is borked...yet. Chips have built in throttling and a hard cutoff point to prevent damage

What were your temps when these BSODs happened (and what BSOD says 'temps' were the cause?)?

What test(s) are you running in P95?

What power supply do you have?

64GB of RAM is alot... I'll assume you use more than 32GB doing something else? Otherwise, try running two sticks at a time.

Before you spend money on a new CPU (I'd get a 7800X3D and upgrade - if the CPU is bad, you may be able to RMA?), for giggles, reinstall your OS and see if that takes care of the problem.

MSI - MPG A850GF PCIE 5.0, 80 GOLD Full Modular Gaming PSU, Japanese Electrolytic capacitor, 850W ATX Power Supply - Newegg.com

Buy MSI - MPG A850GF PCIE 5.0, 80 GOLD Full Modular Gaming PSU, Japanese Electrolytic capacitor, 850W ATX Power Supply with fast shipping and top-rated customer service. Once you know, you Newegg!

www.newegg.com

That's the PSU.

64gig of ram just because. Never had an issue in 4 years and all sticks passed memtest86

Hard to say what it was when it crashed as far as temp. As soon as I rebooted and got the monitor up it was about 85c. Bsod didn't say overheat. Just whea error. However on reboot after the bios posted (the bios screen ) another bios screen came up saying cpu overheated specifically.

Reinstall OS as in complete format and clean or reinstall over what I have now?

Isn't the 7800x3d an am5 socket? I'd rather not replace the mobo yet.. unless we think it's that instead.

 

[JLK03F150]

Reinstall OS as in complete format and clean or reinstall over what I have now?

Yes, I would even delete partitions, create new & format. Then install. Overkill, I know, but it rules out any corruption that may be floating around in the OS.

The 5800X3D is a great CPU for a strictly gaming PC. I think E Dawg might be having an Uncle Leo moment with the 7800X3D. lol

What CPU cooling are you using? Have you checked the heatsink mount & fans?

 

[EarthDog]

Isn't the 7800x3d an am5 socket? I'd rather not replace the mobo yet.. unless we think it's that instead.

The 5800X3D is a great CPU for a strictly gaming PC. I think E Dawg might be having an Uncle Leo moment with the 7800X3D. lol

zOMG... ty. These moments are coming all too frequently these days. Ha!

Hard to say what it was when it crashed as far as temp. As soon as I rebooted and got the monitor up it was about 85c.

Yeah, I missed you said BIOS... I need to slow down or something. LOL! 5800X3D, yes.

For RAM, just remember unless you're using it, it's useless. If you upgraded today, I'd say get 32GB of DDR5... or maybe 48GB.

As far as the OS, yes as JLK said.

 

[Dabba]

I was considering the 5800x3d as well. Cooling is a massive air cooler. Just got a 240mm aio I intend to install.

Also forgot to answer regarding p95. I'm running a blended test currently. Of note the processor has been up to 91c during the testing so far and has not bsod. It doesn't seem to be throttling much either. Avg clocks about 4200. Max was 4800 and lowest so far was 3300 on one core only. The other cores min was a hair below 3800

 

[EarthDog]

My question now is............. why did it overheat in gaming which is typically A LOT less stressful than a stress test?

Try looping 3DMark while it's stress testing and see if it snafus. Maybe it is a power thing as you're not stressing the GPU when running P95.

 

[Dabba]

My assumption was hotter ambient temps plus maybe the paste is gone. I have seen it spike to 85-90 gaming but nor typically. But you'd think if it was temp related p95 would crash it. I don't know why it would bsod and cite cpu overheat but doesn't do it when sitting at 91c with p95. Unless it's not throttling appropriately and going even higher. But seems unlikely.

Whats 3dmark? GPU stress test? When I get home tonight it'll be about 24 hours of p95. I'll do it then. Thinking the psu may be crapping out and not delivering power at load?

 

[EarthDog]

I don't know why it would bsod and cite cpu overheat but doesn't do it when sitting at 91c with p95.

Offhand, I'm not sure what the shut off point is for that CPU. Maybe it got hotter than the stress test is what I infered above.

Yes, a GPU stress test (not named Furmark)... UL 3DMark.

The only way to see if it's the PSU is to put a load on it.

 

[Dabba]

Okay. I will do that tonight and report back. Ideally I'd like to try to definitively rule in or rule out the cpu before replacing it and/fresh windows install.

 

[JLK03F150]

FYI, the 3dMark Time Spy benchmark is available for free from Steam.

3DMark Time Spy benchmark on Steam

Time Spy is a DirectX 12 benchmark test for gaming PCs running Windows 10. With its pure DirectX 12 engine, Time Spy is the ideal test for benchmarking modern graphics cards.

store.steampowered.com

 

[Dabba]

So. I ran p95 for 38 hours without any errors or bsod. The last 10 of those I ran one of the 3dmark ones along side on a loop. (Just a high quality graphical cutscene?) No errors or bsod.

So. It seems I eliminated cpu, GPU, psu, and ram from the equation. I assume of the mobo had issues then they should have showed up in some of the testing as well. Guess that leaves wipe and reinstall windows as the only option left?

 

[Kenrou]

Could have been just a report error in the 1st place, software bugged out and reported a much higher temp. Happened to me on a couple of occasions

 

[Dabba]

Perhaps. But it happened multiple times. It seems either way to be the software at this point I guess. Either windows or something that deals with the hardware given the error type?

 

[Kenrou]

The 5800x3D had some interesting growing pains with earlier BIOS voltages, but even at stock with the latest, it can easily hit 80c+ while gaming. Have you tried undervolting it?

 

[Dabba]

The 5800x3D had some interesting growing pains with earlier BIOS voltages, but even at stock with the latest, it can easily hit 80c+ while gaming. Have you tried undervolting it?

I think at some point in the past I messed with voltages but not sure why. I can't remember and I don't have any posts discussing it. I also googled a lot regarding voltage for the 5800x. I saw some people do undervolt but that's after failing stress testing which this has not. I also saw some posts referencing crashes after returning to idle after being loaded and theories were the cpu undervolting itself but nothing concrete.

 

[Kenrou]

Undervolting is done to lower voltage/temperature while trying to maintain stock or slightly better performance, the 5800x3D is reportedly very good for this. CPU not returning from sleep/idle usually means too low voltage or Windows being Windows. One thing I remembered from the WHEA error was memory timings, check if your ram has 1 or 2 profiles, if by any chance one is giving off errors the other should be (more) stable, or you could jot down the settings to a piece of paper and manually insert them instead of using the profile, sometimes it bugs out.

Here is a good start for tweaking PBO, shaves off 5c-10c on average if stable: