Subject: Re: Q840 serial (was ethernet) performance?
To: Allen Briggs <briggs@puma.macbsd.com>
From: Henry B. Hotz <hotz@jpl.nasa.gov>
List: port-mac68k
Date: 11/04/1997 18:04:28
At 10:47 AM 11/4/97, Allen Briggs wrote:
>the slower machines...  A 33.6Kbps modem going full-throttle
>on one of the two ports is what?  1000s of interrupts/sec?

Theoretical max:
        1 interrupt/character
        8 data + 1 stop + 1 start = 10 bits/char -> 3,360 Kchar/sec
        V42bis compression = max 4X throughput -> 13,440 Kchar/sec

I'm pretty sure the 33.6 number includes the framing and other overhead of
how the data is coded on the line.  The Zilog chip has a 4 byte (3 usable)
fifo so you have a little leeway in responding to the interrupts.  I think
if you only responded to exactly every third interrupt it would still work
perfectly.  Anyone know if they deepened the fifo on the later replacements
like the AV?

What Bill S. has coaxed out of the serial driver is actually pretty impressive.

Signature failed Preliminary Design Review.
Feasibility of a new signature is currently being evaluated.
h.b.hotz@jpl.nasa.gov, or hbhotz@oxy.edu