At the end of October 2020, I built a computer with a 3900X, Gigabyte B550I Aorus Pro AX motherboard, 64 GB (2x32) G.Skill Ripjaws V RAM, and a Crucial P1 1 TB NVME SSD. I am running Ubuntu 20.04.1 on the machine. Up until last week everything worked great even under heavy load (compiling code and heavy numerical analysis), but then a Windows VM I use for development began to have constant BSODs. I also started having many pages crash in Firefox, even on pages with little to no code.
At first, I suspected that I had bad RAM, so I booted up Memtest86+ and ran four passes of all tests. After around 12 hours of test, the results were no errors. At this point, things had become so bad that the machine would not boot. It would hang with kernel panics, soft locks, and other random errors. In trying to find the problem, I did all of the following.
- Boot several different OSs (FreeBSD, Windows, and three different Linux distros). None of these would boot. I did find out that if I disabled ACPI in the OS, the system would boot. Unfortunately, without ACPI, you only get access to one core.
- Upgrade the BIOS to F10 and then F12 - no change.
- Test the NVME driver both with the BIOS long test and with the fsck command. Neither test found any errors or problems.
- Try each the RAM modules individually - still crashing.
- I then swapped in a second B550I motherboard, thinking the problem was there, but I continued to have the same errors.
It seems that the evidence strongly points to there being a CPU problem. Does anyone have any suggestions? Unfortunately, I do not have an extra board or CPU to swap these pieces out to try to find the cause of the problem.
