there's going to be a theoretical limit for each system based on CPU speed, bus speed, hdd speed etc, but on top of that i think software overhead is also going to play a part. i haven't done any network packet analysis to see if exactly why different FTP servers had such different max transfer and ramp up speeds. i suspect it could be a related to either error correction or perhaps different FTP server implementations use different timeouts and/or unsupported features that the older mac OS (or Fetch) just ignores causing some fallback implementation to occur... honestly though im just speculating atm.
Some day when i have the time i do intend to investigate further for curiosity sake to see eactly why however i've had to many other projects to get done so i haven't been in the position to setup a controlled environment for testing. also the thought of testing multible mac's along with different system versions, hdd's, netwokr cards etc is probably half the reason i havent got around to it yet. running basic speed tests is one thing but controlled testing is another thing all together..
perhaps once i finish my dual PDS slot adaptor PCB design i might get back into some speed testing along with trying to find out whats actually effecting the network speeds.
in the mean time if anyone has any insight or knows why id be really interested to hear.
Btw hows the screen sharing perform delay/latecy/overhead wise? i know about timbuktu but ive never actually tried it out in all these years