Mysterious hard crashes with newly built rig

[bjolac]

I recently built myself a new PC, mainly to be able to play Elite Dangerous again after my old Macbook didn’t cut it anymore. It started crashing a few weeks in, when I installed a second RAM module. I tried and failed to find the reason ever since.

The Facts

Hardware
Mainboard ASRock AB350M-HDV R3.0
CPU AMD Ryzen 5 2400G
RAM 1x, later 2x8GB G.Skill Aegis DDR4-3000 DIMM CL16-18-18-38
PSU 520 Watt Seasonic M12II-520 Evo Modular 80+ Bronze

Symptoms:
- Monitor loses signal, Computer freezes completely, even reset button has no effect
- happens in Win 10 and Ubuntu 18.10
- exclusively while playing games that use the GPU
- but not necessarily intensively (seen in Elite Dangerous, Cities Skylines, and Crusader Kings II

Known Issues
- Had to rotate CPU Cooler while building the rig. Heat sink paste may thus be distributed unevenly.
- Bought the RAM modules separately, not as a kit.

Suspects
- RAM: Fewer crashes with only one stick installed, or with higher voltage/lower frequency/slower timing. BUT no memtest errors, and crashes happen sooner or later no matter what.
- CPU: Maybe the integrated GPU is insufficiently cooled due to the heat paste thing? BUT max temperature reading I ever got was about 65 Cels. under heavy load, and stress tests do not reproduce the crashes.
- Drivers: Lots of people seem to report driver issues with the 2400 G BUT the crashes happen in two different OSes, so that seems improbable, maybe?
- Mainboard, PSU? I have no idea how to test those short of putting new ones in. Also, both are brand new, and the PSU should be sufficient by far, AFAIK.

Question
Any ideas how to find the reason for the crashes? I don’t want to buy the while rig a second time just to find the bad part, and that’s the only idea I have right now.

I admit, I’m kind of out of my depth here. So, before I go and order new RAM, a new mainboard, PSU and possibly a new CPU in the end I wanted to ask If I’m missing anything obvious or if anyone has an idea what else I could do.

Any help is appreciated. Many thanks in advance!

 

[bjolac]

Okay, so here's what happened:

Short Version:
The 1.7 BIOS Update to the ASRock Mainboard seemed to solve the problem.

Long Story
While searching for a solution, I found someone somewhere on the internet saying that the 4.18 Linux kernel didn't support the 2400G CPU out of the box. So I gave it a try, installed the brand new kernel 4.20, and... boom! Everything worked like a charm. In Linux, that is.

Having the problem thus reduced to something obviously driver related (I thought), I reinstalled Windows, all the ASRock drivers and the AMD graphics driver, and... NO CHANGE!

So I thought, maybe I grabbed the wrong drivers somehow and tried again with the ASRock driver DVD - something I usually never do, because drivers on disks are old and if you click the wrong button you'll get all kinds of BS you don't want on your computer. But this time, it actually was the solution. ASRocks auto-update utility found the 1.7 BIOS update, installed it, and now, Windows runs fine, too.

The update was there since December 24th or so, I just hadn't bothered to look, as I had already done the BIOS update from 1.1 to 1.3 and figured that no one would release something like this just before christmas.