Path: utzoo!attcan!uunet!bu.edu!purdue!haven!mimsy!midway!oddjob!rfl
From: rfl@oddjob.uchicago.edu (Bob Loewenstein)
Newsgroups: comp.sys.mac.comm
Subject: Re: Very High Speed Modems and Macs
Message-ID: <1990Nov12.153526.1239@midway.uchicago.edu>
Date: 12 Nov 90 15:35:26 GMT
References: <85816@lll-winken.LLNL.GOV>
Sender: news@midway.uchicago.edu (News Administrator)
Reply-To: rfl@oddjob.uchicago.edu (Bob Loewenstein)
Organization: U of Chicago - Astronomy & Astrophysics
Lines: 46


> I'm interested in high speed modems, and I've been reading about new standards
> proposals for very high speed operation over standard phone lines. The only
> trouble with all this is finding serial port devices that will go that fast
> and writing system software to support them.Tthis naturally leads me to wonder
> just how fast the serial ports in Macs will go. I've seen measurements
> indicating that Macs have trouble keeping up with 19,200 bps data rates, but
> I believe this is due to software, not hardware. I know that my Mac IIcs can't
> keep up with my Telebit T2500 in either of its two fastest modes (PEP and
> V.42bis) with color turned on, and that it does much better with color turned
> off. Anybody out there in netland know how fast Apples modem port can operate,
> and how fast the system software can send and receive data? Are there any
> plans to make them faster? Will the nuBus serial cards that can be used with
> the Comm Tool Box support any higher speeds? I'm thinking of the proposed
> ultimate modem standards, which have numbers like 56,000 bps, 78,000 bps, and
> even 100,000+ bps being bandied about in them. if these things come about, it
> sure would be nice to be able to use them. And for that matter, what of the
> possibility of an ISDN card that can offer all of the initial phase speed of
> 144,000 bpc (to say nothing of the ultimate planned T1 speeds of 1.544 Mbps)?

I have been using high speed modems for over a year with a mac II. The modems
are both Microcom V.32c and V.42bis with MNP9 and MNP10 protocols, respectively.
The macII talks to the modem via the modem port at 38400 baud with NO
handshaking. Most of the data is input to the mac, with small control strings
going out of the mac. The data returning is binary, sometimes as big as
1.4MBytes. The only problem I have had has come from having AppleTalk ON...and
the more Appletalk related things running (such as TOPS). I asked about this to
MACDTS and received this reply in part:

>   There are no guarantees about data acquisition whenever AppleTalk is being
> used, or even turned on.  While the Mac is receiving an AppleTalk packet,
> interrupts are disabled for the entire length of the packet.  This means that
> if the largest possible AppleTalk packet was sent, it would take about 26
> milliseconds to receive the packet.  Asynchronous serial data transferred at
> 9600 bps would take about 1 millisecond to transmit.  Simple math says about 26
> characters  could potentially be lost if a 603 byte packet was received.  This
> is very realistic especially when there are routers on the network which send
> out large routing table maintenance packets (RTMPUs).

So, If I turn AppleTalk OFF, I have had no problem (with or without
handshaking!). Once AppleTalk is turned on, I get an error=-89, with 1 to 3
bytes lost.

I also am hoping to use a Nubus board that will alleviate this problem and go
faster. I am beta testing a DigiChannel multiserial board that might do the
trick (and also communicate >56000baud). I'll post my results.