Check your GPU and see if any other solutions work.
Check your GPU and see if any other solutions work.
According to the latest update (https://cultists.network/140/psu-tier-list/), the Leadex III Gold appears in the lower-priority high-end section. The Bronze falls within the middle range.
The only reason I doubt the "fix" versus "working" idea is that MSI says you’ll be billed to send back the part if it’s confirmed to function without problems. I didn’t get charged for returning the card, so I think they must have discovered an issue.
Well, assuming can lead to a whole lot of things . If we took a second and thought about it though, regardless of if they "fixed" your card or not, you still experienced the same error on your 2080. Could it be a coincidence, sure, but when chasing down intermittent issues like this, don't be quick to discount them. That old 2080 of yours uses 100 - 150 watts less power than the 3080, if it were a failing PSU, might explain why it didn't happen as often while you were waiting for your RMA.
Fair observation. The similarity to the 2080 case, albeit at a significantly reduced frequency, is what prompted this update instead of revisiting MSI directly. My logs are available if needed, but I suspect the power supply isn't the root cause—the crash occurs across different power consumption levels. The PSU appears to be failing, though it's only been around a year; perhaps a replacement or RMA would be more appropriate. The system has never fully crashed or restarted, with games simply closing to desktop due to an Nvidia driver error. The GPU is only three months old. I paused work briefly to run "sfc /scannow" as suggested, and it detected corrupt files—this is encouraging. Then I started AC: Valhalla quickly; it functioned normally for a short time before. I won’t have the chance this evening to conduct a full test to confirm closure or to evaluate further fixes. If the problem recurs, I’ll proceed with a clean install of Windows. Should that not resolve it, I’ll consider PSU replacement first, as you and @Gokul_P highlight it as the next likely weak point.
It seems the main focus is on why I’m concentrating on the GPU. In most situations, the issue with this "nvlddmkm cannot be found" problem turns out to be faulty RAM or memory on the GPU. I haven’t seen any tests confirming bad RAM, so the GPU appeared to be the most likely solution.
My thoughts drifted to RAM again; sometimes it passes a clean memtest, but real-world tests show otherwise. Do you have another unit available for evaluation? I usually assume most folks don’t have access to reliable, tested hardware to work with.
I don’t have a good alternative RAM set to try. Usually I keep a backup for most parts, like my old 9-year-old Corsair Gold 850W PSU, which is still functional. My last spare RAM was used by someone else, and upgrading to DDR5 would cost more than the performance boost would provide.