If the VRAM budget was exceeded, the game itself could certainly crash under DX12 (since it does not protect against that in the way that DX11 does), but it wouldn’t crash the OS, which can happily survive the display driver faulting and being re-loaded, since it no longer has any components running at Ring 0. You’d get a traditional crash to desktop. Also, I had this happen while just setting keyboard controls, where VRAM use would be minimal, and I have a 4090 with 24GB of VRAM. Seems unlikely as a culprit to me.
This crash takes down the whole OS, and begins by blanking the displays on which DX12 windows are being drawn - though not necessarily all of them, as I’ve seen an Air Manager instrument remain visible after the crash process begins - and then somehow causes the fans to run at maximum, which could be as simple as pinning enough cores at 100% to make the system respond. Running tasks continue to run - I managed to continue on a Zoom call for a good couple of minutes after a crash started, and I could still hear sim audio, indicating that the sim process itself was still running. Input is blocked, but this could be because the threads responsible for processing hardware I/O like keyboard and mouse are deadlocked.
Eventually, the system blue-screens - I wasn’t in the room to see if an actual BSOD was displayed - and reboots. As I mentioned earlier in the thread, the driver exception that caused the BSOD in this case was in tcpip.sys, which indicates a networking issue, although that isn’t necessarily the actual root cause.
It would be useful to gather system data from those who have these specific crashes. Are we all using the same Intel network adapter chipset, for example? Do we all have the same CPU / CPU family? The same motherboard chipset? Are we all on Windows 11? Which version? And so on.
While it’s clearly the sim that is triggering this somehow, it’s not the sim itself that is crashing. It’s Windows. The solution will either be isolating the Windows problem (most likely a driver) or Asobo identifying and replicating this crash and working out how MSFS triggers it, and fixing that.