@Kocicak
I noticed you have a (now locked) thread on a 3800x's boost behavior. All I can really say is: imagine taking the "good" CCD from a 3950x and restricting the max boost to 4.5 GHz instead of 4.7 GHz. Viola! Now you have a 3800x. They're a very high bin, and they're likely capable of boosting on individual cores above 4.5 GHz were their boost maps capable of such.
As for all-core OC, the only way you will know is to try. Don't assume that all-core OC = higher temps (except in stuff like Prime95 SmallFFTs and similar, where you will want to be cautious). Try CBR20 or Blender Benchmark, and use Ryzen Master to set your voltage to something conservative like 1.2v which should be "safe" even with Wraith Prism. See how far up it will clock before it crashes, and monitor temps. You can squeeze out some extra stability by increasing VRM switching frequency in the UEFI. Not sure what you will want to do with LLC settings; on older AGESA versions, you had to get aggressive with LLC on some boards to increase stability at a given actual voltage, but in 1.0.0.3ABBA, lower LLC seems to be slightly better (not sure why). Use CPU-z to monitor your voltage since HWiNFO64 can give some wonky numbers (not sure about HWMonitor, but I'm not a big fan of it for Matisse systems). If you set 1.2v in Ryzen Master, you may get much lower "actual" voltage thanks to vdroop, which means you'll have to set it higher to get the desired number. So test with some low-ish clocks (4 GHz) just to check your vdroop and tune that voltage until you get exactly 1.2v (or really close to it) under load. Idle voltage is largely meaningless.