I thought so too, but the AX860i is one of the Corsair Link PSUs, so there is definitely a software component there. I took it to mean that there is a synchronization issue with the voltages. I'm not too sure. It still hasn't restarted again, so maybe that's it. I definitely think there's a heat problem too, but I just can't figure out just how yet. There appears to be a 20C gap in the readings from the BIOS and lm_sensors. I tend to believe that the BIOS is more accurate, but an FX 8350 clocked at 4.6GHz on a custom water cooling loop running at around 45-50C under load seems to make sense to me. I'm just really not sure right now.

I thought you were running an i3?
If this is an AMD chip then what, exactly are the 4 or more temperature sensors readings you see when running mrpime? If it's an AMD chip how do you know which temperature is the important one?

Sorry for the confusion. My desktop environment is i3wm. I have conky configured to display the output of the sensors command. I only have two temperature readings from that. One is listed as "amdgpu," so that one is obviously the RX 480. The other is "k10temp-pci-00c3." That's the one I'm going by. I know that AMD uses some sort of algorithm to display sensor data, so impossibly low idle temps should be ignored, but I thought load temps could be trusted.

sensors should give you something like two "temp1" values and a couple of others which may be relevant. I really don't know which ones are the ones you need to follow but take a look and when I boot my AMD machine I can post my values if need be.

I really only have the one. It seems to be the CPU, but I guess I can't be 100% sure. It also seems to align with the temps that are displayed in memtest.

UPDATE: I used an Ubuntu live CD to make sure that it wasn't a kernel issue, and I got the same results with only one temperature reading from lm_sensors.