I’m not into AMD CPU’s but at first glance your 5800X3D and 4080 don’t match at their potential performance.
In my case I’ve built a system with an i5 13600k, 4070ti, 32GB/5200 ( which is in fact a 5600 but limited by motherboard)
This combo runs MSFS pretty smooth with most settings on high and some even on ultra.
Balance is key.
Have to add that I’m not that interested in FPS counters.
No flight simulator will ever beat the real life world.
I respectfuly wish you luck on asking for a refund on your investment.
Happy 
Thanks for your comment. 5800X3D and 4080 are a well matched and very common pairing. There is no reason not be into a certain brand of CPU, a CPU is a CPU. Right now some of AMDs are offerings are particularly well suited to MSFS thanks to their proprietary 3D V-Cache.
This bug does not always occur. If all is functioning correctly I can run smooth 60 FPS locked with Ultra/High settings at 150% upscale 1440p.
However when the bug occurs no amount of lowering settings makes any difference, which suggests it is not dependent on hardware performance. This is further suggested as neither the CPU nor GPU are fully utilised when this bug occurs. (Yes, that is accounting for CPU single thread performance cap.)
This used to be a far more prevalent issue, and Asobo did make some efforts to resolve it, however they have been unable to fully eliminate it, and it still pops up often enough to ruin some of the few times a year I have available for long haul flights.
The issue is that corners were cut during the development of MSFS by using legacy FSX code, which keep rearing their ugly heads, so much so that now we are being forced to move to a “new” platform (2024), which also will still contain legacy code from FSX. Would be unsurprised if there is some other un-resolvable issue caused by that which are then again solved by selling everyone a “new” product again, rather than delivering a clean sheet fully-functioning solution from the beginning.
I would be curious to see a graph of per process memory use over time.