14900K with Asus Maximus Z790 Hero ever increasing instability

[Cuscaden]

Bought an Intel 14900K, an Asus Maximum Z790 Hero and Corsair Vengeance DDR5 2x32GB 6600 RAM. Built the system on April 25th, all went smoothly, transferred over my 4090 and two WD850+ SSDs. System powered up and booted up on the 1st attempt.

I updated the bios to version 2202 and applied the Intel defaults. System was rock solid until Saturday morning when it started crashing. Read up and manually applied 253w/125w/308A power limits as recommended by other people seeking stability. Seemed fine for a few days, then Discord would crash every time I tried to open it, reinstalled it kept on crashing. Removed the windows store version and downloaded it from the Discord website, and it seemed fine. The Ubisoft Connect app would start crashing. MPV started crashing with app crashes and exception in the Windows Application Event Viewer, OTOH I could run something like Far Cry 6 for hours without any problem.

This Saturday I decided maybe I needed to have a fresh windows installation, so set about doing that, created the windows install USB on another PC, started install and I was getting a mixture of different stop exception errors, 0xc000021a repeatedly and sometimes a system service exception and a few other random blue screens.

Turned off XMP, ran mem test and got a clear bill of health on the memory.

Reseated memory, re-did thermal paste on CPU (H150i XT Pro with the 1700 screws), but try what I may I cannot get this system to install windows, boot off the USB stick, it copies data over, reboots, and then 0xc000021a in the same place each time.

I replaced the SSD with an older version which was my OS drive a few months ago, boots up fine. Grabbed a different SSD that I was happy to format, went into Windows install, removed all partitions to allow Windows to do what it wanted to the drive and started getting errors saying it could not format the drive. Rebooted, this time it moved forward, and then on booting the system failed with a "system thread exception not handled". A fresh install OS is impossible on two different SSDs. The system is "stable" in the bios and in memtest. It would seem reading data is fine, but as soon as you go for writing data errors begin to crop in.

Just contacted Amazon and have started a return for the CPU since that is relatively easy to return and swap out. 

Any ideas what could be going wrong? Thanks.

[RamyerM_Intel]

Hello Cuscaden, 

Thank you for posting in the communities. Let us first try to check your system configuration to determine the root cause of the instability. As such, please share with us your SSU logs. I also noticed you mention you have started a return for your CPU. Were you able to do a swap test? How did it go? I will be waiting for your reply. 

Ramyer M. 

Intel Customer Support Technician 

[Cuscaden]

Hi Ramyer,

At the point I sent this email I could not get an OS to install so I would not have been able to run the SSU tool you were requesting. 

I pulled out the system GPU and went with using the iGPU on the CPU. Did another full round of memtest with the parallel CPU test to ensure all the cores on the CPU were being used. Whilst this was running there was graphic artefacting, after that trying to reboot the system led to it not even posting and the Motherboard was reporting a VGA error. 

The replacement CPU arrived shortly afterwards and the behaviour was exactly the same, which led me to suspect the motherboard. Sent the motherboard back to Amazon and ordered a replacement. 

Installed the new motherboard this last Monday, was able to install windows and so far the system is running smoothly. This time I have not upgraded the BIOS on the motherboard, but I am happy with thermals and stable behaviour so for now will leave it as-is with an older bios and a system which, so far, is running smoothly.

In this case the problem was the motherboard and the likelihood is that the first returned CPU was fine..... Very frustrating not to able to run a bootable diagnostic which would pin one over the other, had to go whack-a-mole between the CPU and the mobo and the first call of swapping the CPU was not the right one in the end.