Question [Solved] Constant BSOD on new pre-built PC - AMD Northbridge Fatal Hardware Error ?

[AskenLurom]

Hello to everyone! After a couple years out, I'm coming again to request from all help that anyone familiar with AMD can provide. The case is a bit difficult, and the hints contradictory.

As a summary:

1. I'm an Intel user, and all my life I've been working with Intel. I'm completely new in AMD world.
2. HP 705 G3 SFF PC with AMD Pro A10-8770 CPU, video AMD Radeon R7 Series (Bristol Bridge) - Source
3. Preinstalled Windows 11 by the seller. Never worked, never could go beyond user initial setup.
4. SSD formatted and installed Windows 10 64-bits. Keeps crashing either by force reboot or BSoD not leaving a minidump in the next 5-10 minutes after turning on, message: WHEA_UNCORRECTABLE_ERROR. Only trace is System Events throwing 2 errors written by WHEA-Logger: - Screenshots (in Spanish)
   1. Fatal Hardware error: AMD Northbridge / Machine Check Exception / Error Type: 25 / Processor: 0 (Id 20)
   2. Fatal Hardware error: Processor Core / Machiche Check Exception / Cache Hierarchy Error / Processor: 3 (Id 18)
5. PC working perfectly in Safe Mode, stress tests with CPU-Z and other tests completed succesfully - Screenshots
6. At the moment of this post, Windows Memory Diagnostic Tool running and says it found problems - Screenshot (in Spanish)

So, to add further details, my impressions and all the information I've collected so far, the Fatal Hardware errors were maybe by an overheating or electrical problem in the CPU. I was scared of this possibility, and contacted the seller, who basically replied that "this has never happened before, check this and that but idk". I don't have thaaaaat much IT knowledge to know what is going on, to be honest.

My first contradictory hint was that, if the problem is electrical, overheating or something like that it should've crashed even in Safe Mode when CPU-Z was doing the stress test. The CPU fan was going at enough speed to rival speed of a Boeing 747 flying ( 🤣 ) but it never crashed, in fact, the stress test lasted half hour and survived succesfully. As I mentioned before, in normal Windows running it crashed between 5 and 10 minutes after turning on. Can't run HWiNFO 64 because I am (was?) in Safe Mode.

Also, I do have (unrelated to this issue, but it's a sort-of knowledge precedent) a problem with my Thinkcentre M91p MT + Nvidia 1050 Ti (LLP) that if I do install Nvidia latest drivers it gets epilectic and the system crashes with BSoD and a lot things until I manage to unistall it. Soooo I'm very familiar with Graphics Driver issues, then my first thought is that there is a driver problem, and considering Safe. However, looking through internet reports a lot people that changes drivers and that solves nothing, so I'm unsure that installing correct AMD drivers can solve the issue.

And finally, with the memory problems, it's something that caught me off by surprise, genuinely. In fact, I'm very confused, because something I tried (following advice from someone else case in internet) was extracting the 4 RAM cards and testing them one by one in the slot the computer demands me using (yes, this is something that never happened to me: The computer has 4 RAM slots, but if there is no memory installed in 1 of them, it beeps over and over and never starts. I can even fill the other three, but no, the computer demands there has to be a RAM card installed in a specific slot or it never turns on. It beeps and throws red lights and all that) and in all scenarios it just kept failing... BUT in Safe Mode it doesn't fail. As I said, it ran for an hour without crashing, so... what is going on there?

So, with all my case explained, if there is someone that can help me to find how in the world I can solve this issue, I would be very thankful...

As for more information of how this HP G3 is built, I'm leaving some screenshots of the reports given by AIDA64 Extreme and CPU-Z:

AIDA64: Pictures (Spanish)
CPU-Z: Pictures (English)

Thanks a lot to everyone in advance.

 

[Lutfij]

SSD formatted and installed Windows 10 64-bits.
Where did you source the installer for the OS? Did you install the OS in offline mode? Once your OS was installed, did you manually install all relevant drivers for your prebuilt in an elevated command, i.e, Right click installer>Run as Administrator.

 

[AskenLurom]

SSD formatted and installed Windows 10 64-bits.
Where did you source the installer for the OS? Did you install the OS in offline mode? Once your OS was installed, did you manually install all relevant drivers for your prebuilt in an elevated command, i.e, Right click installer>Run as Administrator.

I did it with a bootable flash created with the official tool for it, it's in fact the same flash I use since a year ago. I... guess it was offline? Isn't linked to any Microsoft account yet, despite the computer already smelled internet with the TP_Link WiFi Adapter I got from the seller (edit: also, just in case I temporarily disconnected that one and connected my own currently-using adapter, as first with the W11 pre-installed OS the first BSoD and force-rebooting moments happened exactly after giving internet to the computer, so I first thought was fault of the WiFi adapter).

Haven't had the chance to install any driver aside from those that Windows gives. As I mentioned, the computer crashes within 5-10 minutes after turning on (haven't tried installing in Safe Mode, tho.)

 

[CodeConnoisseur]

I would look more into the memory issue with MemTest86 especially given that the error is related to the cache hierarchy. Test each DIMM one at a time in the slot that your motherboard expects.

It's not clear to me in your post if you did end up testing each individually with the Windows Memory Diagnostic, but if an error is detected with every single stick then I would be concerned about an issue with the CPU cache, motherboard, or memory controller.

 

[AskenLurom]

I would look more into the memory issue with MemTest86 especially given that the error is related to the cache hierarchy. Test each DIMM one at a time in the slot that your motherboard expects.

It's not clear to me in your post if you did end up testing each individually with the Windows Memory Diagnostic, but if an error is detected with every single stick then I would be concerned about an issue with the CPU cache, motherboard, or memory controller.

I'll run that test with MemTest86in the morning. Right now (5 hours have passed) the Windows Memory Diagnostic Tool keeps going, last time I checked it was in Phase 2/2 and around 60%. However (don't take me wrong, it's just a personal question) if there is a physical problem with the RAM memory, why the Safe Mode didn't crash in the whole hour it kept running?
(Side note I didn't mention before, but maybe can be appreciated in the ton screenshots I share) is that I just got to see in Safe Mode that around just around 45% of total installed RAM is usable... maybe that's a red flag I'm not aware of?

Oh, uhm... what I did was extracting all 4 memory cards (sorry if I'm not nailing the correct name), numbering them 1-2-3-4 and testing one by one in the slot the computer demands being occupied. No matter which one I tried in said slot, or even if attempting connecting in pairs combining them (I remember reading somewhere about memory cards work better if paired), the BSoD randomly appearing as well as sudden force rebooting kept happening.

Edit: I just went to see before sleep, and it's in 94%. I think it's worth mentioning test started 7 hours and half ago.

Spoiler

 

[CodeConnoisseur]

I'll run that test with MemTest86in the morning. Right now (5 hours have passed) the Windows Memory Diagnostic Tool keeps going, last time I checked it was in Phase 2/2 and around 60%. However (don't take me wrong, it's just a personal question) if there is a physical problem with the RAM memory, why the Safe Mode didn't crash in the whole hour it kept running?
(Side note I didn't mention before, but maybe can be appreciated in the ton screenshots I share) is that I just got to see in Safe Mode that around just around 45% of total installed RAM is usable... maybe that's a red flag I'm not aware of?

Oh, uhm... what I did was extracting all 4 memory cards (sorry if I'm not nailing the correct name), numbering them 1-2-3-4 and testing one by one in the slot the computer demands being occupied. No matter which one I tried in said slot, or even if attempting connecting in pairs combining them (I remember reading somewhere about memory cards work better if paired), the BSoD randomly appearing as well as sudden force rebooting kept happening.

I didn't forget about the fact that it seems to be working fine in safe mode, but yes that's a good question. The fact that it's only showing 45% usable RAM is certainly not normal, however, and likely does point to deeper (see: hardware) issues. However, that's another reason that I recommended MemTest. It's OS-agnostic, so it will help determine if the problem lies in software (drivers) or hardware.

 

[AskenLurom]

I didn't forget about the fact that it seems to be working fine in safe mode, but yes that's a good question. The fact that it's only showing 45% usable RAM is certainly not normal, however, and likely does point to deeper (see: hardware) issues. However, that's another reason that I recommended MemTest. It's OS-agnostic, so it will help determine if the problem lies in software (drivers) or hardware.

I managed to find a picture I took in the hour it was running in Safe Mode. As it says, from 16 GB RAM installed, a bit less than 8 GB are usable.

Spoiler

According to AIDA64 report, all the rest is "Virtual Memory", I don't know what does that mean to be honest, and who does control that. In case being fault of the BIOS, I haven't touched things there.

Spoiler

 

[AskenLurom]

SSD formatted and installed Windows 10 64-bits.
Where did you source the installer for the OS? Did you install the OS in offline mode? Once your OS was installed, did you manually install all relevant drivers for your prebuilt in an elevated command, i.e, Right click installer>Run as Administrator.

I would look more into the memory issue with MemTest86 especially given that the error is related to the cache hierarchy. Test each DIMM one at a time in the slot that your motherboard expects.

It's not clear to me in your post if you did end up testing each individually with the Windows Memory Diagnostic, but if an error is detected with every single stick then I would be concerned about an issue with the CPU cache, motherboard, or memory controller.

Hello again to my two IT angels, and thanks again for all the help so far you've been providing to me.

Sorry for disappearing for a bit, sadly work was a madness for me so couldn't return to the computer until today.

Have to mention I requested help from AMD as I was also thinking in my theory of a problem with the display drivers. They gave me assistance in that matter to cover at least said possibility.

Spoiler

Now, the computer survives for around 1 hour without failing, but then, BSoD again or force rebooting, and Event Viewer keeps showing the same WHEA errors of before. So... apparently here I had more than 1 problem.

I then returned to test with MemTest86, and it started minutes ago after creating the flash drive. Immediately something is burning my eyes:

If you remember, Windows days there are 16 GB installed but only 7.43 GB are usable. Memtest86 tells a bit more but still in the margin of 7 GB.

Would be then fair to think 2 memory cards are damaged, or... what's happening on there? I'll post again when Memtest is done.

 

[AskenLurom]

I didn't forget about the fact that it seems to be working fine in safe mode, but yes that's a good question. The fact that it's only showing 45% usable RAM is certainly not normal, however, and likely does point to deeper (see: hardware) issues. However, that's another reason that I recommended MemTest. It's OS-agnostic, so it will help determine if the problem lies in software (drivers) or hardware.

Finally finished MemTest86. Something that has me absolutely confused is the fact it analyzed a bit less than 8 GB as I mentioned above, but it manages to find the 4 memory cards. I don't know what's going on there.

Spoiler

However, in the space it analyzed, 7.618 MB, it found 1 error and threw a big FAIL screen in the end.

I uploaded some screenshots, as well as the results file in GDrive:
MemTest86 Results May 11 2024

 

[CodeConnoisseur]

Finally finished MemTest86. Something that has me absolutely confused is the fact it analyzed a bit less than 8 GB as I mentioned above, but it manages to find the 4 memory cards. I don't know what's going on there.

Spoiler

However, in the space it analyzed, 7.618 MB, it found 1 error and threw a big FAIL screen in the end.

I uploaded some screenshots, as well as the results file in GDrive:
MemTest86 Results May 11 2024

Run MemTest with each stick individually in the one slot that your motherboard demands. If that all passes, then there is no issue with that slot or any of the sticks themselves. You can then have one stick in that one slot then try one at a time in each of the other slots. It's highly unlikely that you have multiple failed DIMMs, unless they shipped defective from the factory or there was ESD or other damage during installation or maintenance.

Additionally, if every module does fail in the first slot, then the issue is likely with that slot, the motherboard, or the CPU.

 

[AskenLurom]

Run MemTest with each stick individually in the one slot that your motherboard demands. If that all passes, then there is no issue with that slot or any of the sticks themselves. You can then have one stick in that one slot then try one at a time in each of the other slots. It's highly unlikely that you have multiple failed DIMMs, unless they shipped defective from the factory.

Additionally, if every module does fail in the first slot, then the issue is likely with that slot, the motherboard, or the CPU.

Run this tests yesterday night, and 2 memory cards ended with "FAIL" result. At the end I discarded those two, and ended using the two that passed correctly MemTest. I'm happy to see that, at least apparently, the issue is finally solved with discarding those two memory cards. I even installed Microsoft Flight (2012) and played the game for over an hour with no issues so far, no BSoD, nothing.

Oh, and the ones that are yet installed (yes, this means I downgraded from 16 to 8 GB RAM) are in Slots 1 and 3, based in the fact they're dual channel.

I don't know if considering this topic as solved, or if there would be something else you consider I should do?