Computer Type: Desktop
APU: RYZEN 3400G
Motherboard: GIGABYTE B550M DS3H
BIOS Version: F13g
RAM: 2x8gb G-Skill F4-3200C14
PSU: Coolermaster 650W Gold
Operating System & Version: WINDOWS 10 PRO 20H2
GPU Drivers: Adrenalin Driver Version: 21.5.2
Description of Original Problem: Had a week of random BSOD's (with different errors each time) and boot loops after a crash. Eventually got through in Safe Mode, uninstalled every hardware driver, and reinstalled - everything was good as gravy for about a week.
When using anything that requires GPU utilisation (CAD, watching some YT videos, etc), I will get a random crash that keeps motherboard power on and the fan spinning, but nothing can be done - reset/power switch doesn't work, no video/audio output, etc. The only way to get the PC running again is to pull power for 10s and then it will boot. No errors are logged in Windows whenever it happens.
I have extensively tested CPU and RAM for stability and had no issues. This RAM/MB/NVME drive was used previously with a 2600 for 3 years with no issues. Problems arise when I do any sort of GPU tests, with the crashes immediately happening.
I tried manually setting GPU voltage to 1.25 and it gives me about 10s of stability before the crashes continue.
Manually setting clock speeds to -anything- in BIOS result in crashes as soon as it's left the BIOS image screen.
Watching HWInfo while initially loading a GPU test, some weird behaviour occurs. GPU clock speeds get pinned at 1500mhz, with VDDCR_CPU voltage rising to 1.55v. A second before the crash occurs, VDDCR_CPU voltage drops to 1.137 (with nothing else changing) and then everything dies.
Suggestions?