Looks like I might have identified (one of) my FPS degradation candidate
… the Pms50 GTN750 (or the underlying features of MSFS)
While exploring Iceland frame rates started to degrade after around 2,5 hours. Since there is basically no scenery in the most northern part, I ruled out the 3D landscape as the cause.
It was clearly a “main thread” issue …
… and since there have been many reports in the past that “glass cockpits” eat CPU I disabled FMS1 and FMS2 in my H145
… which does not only seem to disable the screens of the GTN750 units
… but it seems to disable the WASM sandbox (or whatever runs the GTN750 code)
… and … Bingo.
So now I recovered from 3 FPS back to around 20 FPS … and then I turned the FMS switches back on … and it stayed at around 20 FPS (which is normal for my DX12-4K-Ultra setup)
So I will watch this … but it somehow does makes sense, as I did recently move to a newer Pms50 GTN750 version … and so correlation might be causality after all.
While I do understand that memory leaks or inefficient code with potentially exponential complexity can start eating lots of unnecessary CPU cycles (stuff happens)
… I am still a little “puzzled” about this because it shows that even at this point MSFS (still) seems to make very poor use of multiple CPUs and that the dependency between certain aspects (sandboxes) of the sim seems way too tight.