This is not a support request or anything, more of an captain’s log for any poor sod who might have something similar in the future. If I helped someone: You’re welcome.
I’ve had quite a weird, although interesting case of display artifacts which in any other corner of the internet would probably be handwaved away as a GPU issue. Let me explain:
Last week I got a message at boot by my overclocked i7 5820K (remember - it’s X99, so no iGPU) of four or five years I believe: CPU overvoltage something something" and an automatic reset of BIOS settings on my Asus X99-A USB3.1 (mouthful, I know). This, at that point, seemed like a normal overclock failure and it went back to factory settings. I hadn’t played with any settings in quite a while, so it came as a surprise considering it was quite stable for 4.5GHz around 1.285V I think for two years.
Boot completes and I’m in Linux. Steam came up filled with giant green flickering squares.
Your typical failing GPU anyone would think at that point, especially considering my GTX 770 is getting old. Though getting a replacement GPU in these times is a ridiculous task. There was an interesting part of the artifacts however: Only certain applications with hardware acceleration was affected. Firefox? Green squares. Chromium? Green squares and corrupted shapes. Steam? Green squares. The rest however seemed fine.
So I decided to boot my Linux work VM which has my GTX 1080. No issues at the get-go until I opened up a YouTube video. Artifacts all over the video, fading squares. Certain icons were scrambled. At this point I started to think both my GPUs were on life support. Absolute hell. Booted my Windows VM and had the exact same artifacts on YouTube videos, icons and so forth.
A hardware failure at this point was guaranteed, most likely the GPU like everyone on the internet says, and they’re always right… Right?
So I decided to test both my GPUs in a friend’s computer. To our surprise, they work flawlessly, no artifacts, no issues. So what is it then? RAM? The 650W PSU? After testing everything from upgrading the BIOS, dmesg, running single GPUs both in Linux and Windows bare metal, testing each memory module, running Memtest86, checking connections and more, I decided to bite the dust and buy a new PSU in hopes that it might be power to the GPU that is failing.
How wrong I was.
After plugging in the new PSU, it still failed. Same green squares and corrupted geometric shapes. Defeating.
At this point it had to be the motherboard failing, or the CPU. And after more testing I decided to turn off every single core in my 5820K except one and slowly enable them one by one.
Core #3 is the culprit.
If it was disabled, the green squares and shapes disappeared and YouTube videos stopped having fading artifacts in the VMs and the host. Why? I have no idea, I assume the core is faulty and are causing the display artifacts for whatever reason. I’ve never heard of artifacts like these being the CPU, especially considering the X99 CPUs does not have an iGPU.
So the problem has been solved. The CPU itself might be failing on me slowly but surely I’d assume, but it’s working fine with one core cut off. I guess a Ryzen will be in the pipeline soon enough. I’m just happy I didn’t instantly try to buy a new GPU in these trying times.
Hopefully that helped someone out there. Everyone might be telling you that the GPU is the problem when dealing with display artifacts, and most of the time it is, but sometimes it might not be as cut and dry as you’d like to think.