This is definitely a weird intermittent issue. I had a similar situation yavorsky. I was gaming one day and noticed in HWiNFO that there were a bunch of errors (212 or something). I kept playing, left the computer running for hours, got up to over 1000 errors. It was every 30 seconds or so. The idea I had before with memory timings may be not at all related. There was a GPU driver update in the middle of the session, but I think errors started before and continued after. I didn't do a clean install or anything and didn't reboot after.
One thing I've been tweaking and testing is slowly lowering my CPU voltage to see how low I can run it and have everything work correctly. I think I had stopped at 1.380V with a +50 offset and -25% droop. So the next time I booted up, I changed the core voltage to 1.390V in E-LEET (not BIOS...haven't made any adjustments there). This time, I gamed for a few hours, then left the PC running over night and halfway into the next day, HWiNFO running the whole time: 0 errors. Does that mean the voltage tweak fixed it? I have no idea. I think it might just be a fluke with boots. Sometimes it is happy, other times not so much. Is it the first boot after I've switched off power to the PSU that has errors, but reboots do not? I don't know. I definitely have too many variables to be sure one thing or another "fixes" it, but there's definitely something odd that happens intermittently that causes it and I couldn't tell you for sure if it is OC, memory timings, voltages, or just a random error from the GPU that only happens sometimes. I don't think I had this issue before I put a 3080 ti in, but I'm 100% positive on this (only about 90% sure).
Both boots I was gaming, had my GPU at +75 on the core and +600 on the memory. It can go higher, but I only do that while running benchmarks.