Assuming it’s true, what’s the difference to basic people like me?
A CPU has a bunch of functional units, so one for integer maths, one for floating point maths, sort of thing. When a piece of code needs to do some integer maths, it doesn't use the floating point unit so that sits there twiddling its silicon thumbs. Where multithreading comes in is you can run two pieces of code sort of in parallel. That is, when one piece of code needs to do integer maths, and one needs to do floating point maths, both can work independently. However, when they're done and both need to do some memory accessing, they have to wait in turn.
As such, more threads enables you to get more utilisation from your CPU. The overhead is fairly low, so it's worth it. Something like 30% better overall processing performance peak for Intel, with sustained benefits somewhere below that. Googlage threw up
this first benchmark that actually shows HT slowing down a game (FFXV)! More
workload benchmarks - raytracing sped up 20% with SMT. For more than two threads, though, certainly other architectures wouldn't benefit much. It depends on what functional units you have per core and how the threads can use them. If the cores went fatter with more resources, you could get more, resulting in more like half-cores than just threads. Otherwise, I doubt 3 threads could bring much more than what 2 threads can. It'd also not be especially good for raytracing unless extra FPUs were added, but even then you'd be bottlenecked by memory accesses. Really, if you want CPU based tracing, you want lots of small CPU cores.
Hypothetical, if Microsoft uses Zen-3, then a 4 physical core setup (16 logical threads), would be more ideal than an 8 core (16 logical threads) Zen 2 setup, on achieving higher clocks and saving die-space in favor for a larger GPU. Could be possible....
You'd have way less overall processing power though. If Zen 3 could get 50% better utilisation from 4 threads, you'd have the equivalent of 6 cores work from the CPU. An 8 core setup with 20% better utilisation through 2 threads per core would be 9.6 cores worth of processing. And more importantly, the eight core solution would be more consistent, dropping to a worst case of 8 cores versus 4 cores worst case for the Zen 3. If you can clock 4 cores higher, enable switching off 4 cores on the 8 core for boost modes.