Testing to the maximum performance limit at the hardware level

Windows 10 Home 64-bit
Intel Core i7 7700 @ 3.60GHz 4 cores 8 threads 4090.0 MHz Kaby Lake
ASUSTeK COMPUTER INC. H110-PLUS
4095MB NVIDIA GeForce GTX 1080 (NVIDIA)
111GB Western Digital WDC WDS120G1G0A-00SS50 (SATA (SSD)) SATA III 6.0Gb/s
memory XPC 16384 MBytes

I want the maximum performance the games physics engine can deliver which is said to be based solely on the CPU vs physical block limits.

?1. I have read that SE is coded to use up to 4 threads only . is this true or not?
?2. taking core count and clock speed into consideration who could possibly perform better?
the latest AMD or latest intel?

previously testing appears that it does seem to do some computation after 4 threads but not much of a noticeable difference for the amount of affinity applied on those threads or cores.

considering a LAN server, could splitting the server from the client pc gain any benefit or would the network lag be introduced as an issue at that point?
Does the server compute the physics and record the result then stream to the client OR does the client compute the physics and record that result to the server?

I have tested the game on an i7 8700k with 32GBrm and 1080GTX and gain a plausible 20% - 25% increase in sim performance in a set test environment with server residing on the client pc. at this point is it in the code of the game that the devs need to tweak or or can I squeeze any more performance at the hardware level without altering my game settings, block limits and test procedure?

i7 7700 3.6ghz 8 threads = 98-% - 110% resource monitor max cpu usage. cpu process setting realtime
I7 8700k 4.0ghz 12 threads = 87% - 92% resource monitor max cpu usage cpu process setting normal