Random crashes to black screen after replacing GPU

After replacing the GPU to a newer model, I started experiencing seemingly random PC crashes to black screen. I've never experienced an issue like this before and I have no idea what part might be at fault nor how to diagnose/troubleshoot this.

Last configuration that wouldn't crash was:

 - CPU: AMD Ryzen 7 1700 (8c/16t), stock cooler, no oc - GPU: MSI GeForce GTX 960 2GB (6 pin), no oc (TDP: 120W) - mobo: ASUS PRIME B350M-A - RAM: 4 x 8GB 2666 - PSU: Corsair VS650 650W - disk: 1x SSD (OS), 2x HDD 5400RPM - cooling: 6x 120mm case-mounted fans - screens: 1x 32", 2x 24", plugged to DisplayPort - couple USB peripherials like BT, Logitech Unifying, other receivers - OS: Windows 10

During that time, the computer would run just fine. Worth noting is that before that, early 2021, I used to have MSI GeForce RTX 2080 Ventus 8GB (2x8 pin), no oc (TDP: 215W) which would run without any issues as well.

Then, after replacing the GPU again sometime in March, PC started randomly crashing. The upgrades went like this:

1. Replaced GTX 960 for Gigabyte GeForce RTX 3060 Aorus 12GB (8 pin), no oc (TDP: 170W), used DDU to purge drivers and do fresh driver install, PC started crashing.
2. Returned the card and got used Gainward GeForce RTX 3060 Ti Ghost 8GB (8 pin), no oc (TDP: 200W) (I am sure it wasn't used for mining or heavily overclocked), used DDU to purge drivers and do fresh driver install, PC was still crashing.
3. Did a fresh install of Windows 11, PC was still crashing.
4. I noticed my PSU is often listed as "replace immediately", so I went for brand new Gigabyte P650B 650W power supply, PC is still crashing.

Most of the PC parts are about 6 to 7 years old at this point, except for:

• RTX 2080 was about 3 years old when sold in 2021
• brand new Gigabyte PSU
• one HDD is about 3 years old
• two RAM sticks are about 3 years old
• 3060 Aorus was brand new
• 3060 Ti Ghost is about half a year old

The crashes go like this:

They occur seemingly at random, while using 3D intensive apps like Blender or video games, or while the PC is idle, or immediately after booting into OS. One surefire way to induce the crash though is to unpack a certain set of heavily compressed files from one hard drive to other hard drive.

The symptoms are:

• screens going black
• BT devices disconnecting
• on 3060 Aorus, the fans would up to 100%
• on 3060 Ti Ghost the fans stay as they were, instead the RGB lighting turns white and starts blinking continuously

Other things I've tried:

• Reseating the GPU, along with dusting the PCI slot with canned air
• Reseating all the other cabling
• Reseating RAM sticks
• Ran Furmark for extended period of time, GPU has fans kick in and temps stay around 70 Centigrade (158 Fahrenheit)

Other things I've noticed:

• One of the drives (source drive for unpacking the compressed files mentioned before) has yellow Caution status in SMART, some reallocated sectors and a few unrecoverable sectors. This is a non-OS drive.
• All drives are encrypted using Bitlocker since fresh Windows 11 install. They were encrypted before using Veracrypt since I moved onto GTX 960, which worked flawlessly.

1 Answer

The issue at hand was BIOS firmware.

Latest stable version 5609 for ASUS B350M-A proved to be unstable in this configuration. Installing 6026 beta firmware resolved the issue.

Currently PC is running stable, no amount of heavy load on CPU, GPU and/or disk is able to crash the system, which is also stable during extended periods of idling (36 hours of continuous idling without a system crash).

Both firmware files were obtained from ASUS' website. As a sidenote, it's commendable that a motherboard from 2017 is still receiving firmware in 2022, five years later.