I have a slightly undervolted ZOTAC RTX 4090 Trinity OC card running at 975MHz @2730MHz, with a +1000MHz memory OC. I notice that in at least one game (Black Myth: Wukong), sometimes, when the game is paused... MSI Afterburner records a temp limit AND power limit spike for a short moment, before returning to normal. I've read sometimes this happens to some RTX 3xxx/4xxx people when exiting/entering a game's or benchmark's menu/screen.
It is quite strange to me, since the GPU and memory temperatures do NOT spike when this happens. In fact, in that moment, GPU usage in that spike goes down to 0% before picking right back up to normal, and temperatures spike downward before picking back up to expected temperatures. No crashes, no reduction in performance when the spikes don't happen. My personal GPU temps for a game like that do not go over 60C, and none of this happens during any kind of gameplay.
I've ONLY seen this happen in that "pausing the game" moment, and the GPU continues like nothing ever happened. Researching forums have brought up confusing conclusions saying that it's a weird thing with modern GPUs when undervolted in anyway, to a bug in the GPU drivers themselves past 457.xx with sensor reading software, to happens when there's a sudden load change post 457.xx...
Is this something I should worry about? Is this just a known bug? Or is, for a very very short moment, something in the GPU is overheating? I can bring in charts if needed.