🚀 go-pugleaf

In article <kY4pH=y00YU4I6pXh1@andrew.cmu.edu> rl1b+@ANDREW.CMU.EDU ("Robert A. Locke") writes:
>DO NOT USE THE RAINBOW AT 19.2Kbaud.  It won't work past 9600 baud.
>In either send or receive, the Rainbow cannot handle the speed (though
>it is documented that it can).  It supposedly has to do with the
>communications chip that they are using (can't remember the number).

Actually, I have used the Rainbow (100A and 100B) at 19200.  The chip is
a NEC 7201-C.  At this speed, you have roughly 2500 clock cycles between
received chars--on an 8088, about 425 average machine instructions, so a large
interrupt handler or polling loop will not be able to handle 19200.
This was done for a DataComm class--a small (4 computer) token ring network.  
Included in one of our test setups was an IBM XT.  Our original intention was
to demo the thing (to the prof.) at 2400bps, but with time to play with it
we decided to try 19200, and had no problems whatsoever.  By the way, the
whole thing was written in Turbo Pascal--interrupt handler and all.
					Tom Colley