Suse 10 Review

[John Macdonald]

On Tue, Dec 13, 2005 at 10:45:46AM -0500, Lennart Sorensen wrote:
> On Tue, Dec 13, 2005 at 09:38:11AM +0200, Peter wrote:
> > The tty on a 33MHz machine is slow, period. I still have 33MHz machines 
> > (two) that work and they are slow slow. That's the way it is. 33MHz 
> > machines have problems keeping up with 115kBaud serial streams while not 
> > doing anything else, for example (of course no deep FIFOs).
> 
> I would think a 16550A with the 16byte FIFO would take care of that.
> Always did in the past as far as I recall.  Of course the 8250's with no
> FIFO could barely run 19200 or 38400 on a machine that speed.  I
> remember my Amiga500 could handle 19200 just fine, although I think I
> had it configured to use a 512byte buffer on the serial port (gotta love
> hardware flexible enough to allocate a chunk of memory for the serial
> ports own use).

I recall the days of running 7 and 8 developers on terminals
(at 19200) all sharing one 16 MHz 386 running Interactive Unix.
The hardware only had 16450 serial interfaces, so the buffering
was very limited.  However, that just meant that output to the
terminal didn't always run at the full speed of 19200 when
everybody was busy displaying stuff.  The modem ran at 2400
(later 9600) without problem.  ASCII terminals don't often run
into buffer overrun issues, even on slow (by today's standards)
hardware because the input is limited by typing speed and
you'd have to type extremely fast to have problems.  (Even the
high speed touch typists did not suffer from character loss on
that system.)  We would only have problems when we tried to put
long strings into the programmable function keys of a terminal
- four characters would occasionally have one dropped, and if
the string was six characters long or longer, it would rarely
get sent correctly.

-- 
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml