CPU dying? WHEA_UNCORRECTABLE_ERROR, AuthenticAMD.sys (dump, debugged)

[thecheeks]

WHEA_UNCORRECTABLE_ERROR (124)

Debugged dump file for quick reference:

https://pastebin.com/K1SWsMJB

BSODs went from once in a while, to once a day, to ~3 times a day, to 9+ and literally impossible to play FF14 anymore.
I've only managed to get ONE bsod to write a dump file; there's been 30+ BSODs that did NOT write a dump file (the other dumps in this zip are 2+ weeks old, maybe related not sure).

I ran the latest dump through the debugger, AuthenticAMD.sys is listed, running Arg2 shows "Common Platform Error Record".

 

Tried running Prime95 and Aida64, both CPU stress tests only hit about 90% on all cores, 3.4GHz instead of 3.8GHz. Aida64 cpu+gpu stress isn't causing the BSOD after 10 minutes. Running Final Fantasy tonight I was able to get it to crash within 5 minutes like 10 times in a row while testing stuff. I changed the system settings to NOT restart the computer on BSODs, so everytime I get the WHEA error it sits at 0% trying to 'collect' info and write the dump file.

PS: It's not just FF14 that crashes the system, it's any game that pushes the system. I was playing a game called GTFO tonight and I got BSODs 5 times in 2 hours (none of which wrote a dump file).

 

I've ran a quick 1 pass (30min) memtest86 on each stick individually installed into the system, no errors. Did not run an extended test though.

 

Windows 10 64bit
Gigabyte Aorus x399 (latest BIOS)
Threadripper 1900x (stock speeds 3.8GHz)
4x8GB (32gb total) RAM (xmp 3200)
GTX 1080ti
Samsung 970pro NVME (OS drive), some other basic Samsung nvme, 1tb samsung SATA SSD. SMART status on all drives are good.
Thermaltake 850W gold rated (Toughpower Grand series)

SysnativeFileCollectionApp.zip