Conference 7.286::macintosh

    this seems like the right note.

    I've got a new Performa 6400 with an internal modem.
    To the best that I can figure out, it has an internal
    Apple GeoPort 28.8.

    I'm wondering what this is, in light of the previous notes....is
    it the same as an Apple Teleport Adapter?  (only internal?)
    Is this what I see referred to as an 'express modem'?

    Out on Apple's SWUpdates site, there is a new GeoPort software
    update, but the read-me seems to imply that it only works
    for Telecom Adapters and not Express Modems.  I installed it and
    everything seems to still work, but nothing is faster (only maybe
    i'm not dialing faster modems?)

    My Performa came with Apple Telecom V3.0 pre-installed.  but
    no installation floppies.  I notice that Apple Telecom V2.4 or so
    is out on the archive.  Does someone have V3.0 they could upload
    (since I assume its public?)
    Thanks
    bob armstrong

Bob,
Congratulations for your new equipment! Although I have the same config, I can't
install the US software upgrade over the french version. My french default 
setup is very different (ccitt vs bell etc) and I'd loose the Minitel emulation 
mode if I used the US version. It's a pity because I have a problem with the
incoming calls detection which does not exist in the US version.
Jean-Pierre.

    Re .21
    
    Apple's GeoPort architecture is a split system.
    
    A conventional modem has hardware and soft/firmware to do its job, and
    it just takes serial data from the computer.
    
    GeoPort was intended to be a multipurpose subsystem based on the same
    kind of architecture the old-fashioned glass-house mainframe computers
    used to have.  Instead of distributing the computing tasks, GeoPort
    loads them onto the Mac.  It uses relatively inexpensive "pods"
    containing just the hardware; the soft/firmware part of the job is
    transferred to the computer itself.  This means that if you want to do
    4:1 data compression at 28800 bps, for example, you can't just let your
    computer spit data out under DMA control at 115200 bps and have the
    modem do the compression and buffering and flow control.  Instead, the
    computer is doing ALL of that stuff and then shipping the data to the
    pod.  The pod must have some buffering in it, of course, but the Mac is
    doing the lion's share of the work.
    
    What makes this even worse is that a GeoPort serial pod that connects
    you to a phone line, like a modem, isn't at all inexpensive; the
    current price in the catalogs is around $130.  You can get a SupraSonic
    voice/fax/data modem for $260 - this is twice the price, but it frees
    your processor of some work, and it's a dedicated system designed to to
    work right as a unit.  It also gives you some features you don't get
    with GeoPort, such as Caller ID and a very informative alphanumeric
    front-panel display.  The Supra is also flash-ROM upgradeable, and in
    my experience Supra (Diamond Multimedia) stays on top of the need for
    changes.  (Apple was REALLY REALLY SLOW in getting a 28800-bps GeoPort
    pod out there.)
    
    I consider GeoPort a bad attempt, and I heartily encourage you to get
    rid of it and use a real modem.  Your mileage may vary, of course.  :-)

    But what I'm really wondering (and I do appreciate your
    explaination) is what is the difference between
    the various things I see associated with GeoPort...

    GeoPort Telecom Adapter
    GeoPort internal modem
    Apple Express Modem
    PC Card Modem

    Is the AppleExpress Modem just a regular modem?
    Is the GeoPort Internal modem just an internal version
    of the GeoPort Telecom Adapter?
    bob

>     <<< Note 679.23 by SMURF::BINDER "Errabit quicquid errare potest." >>>
>    
>    I consider GeoPort a bad attempt, and I heartily encourage you to get
>    rid of it and use a real modem.  Your mileage may vary, of course.  :-)

Modems as most people know them are a hack and curse.
Integrating the data buffering, compression, and flow control
into the main processor allows for a much more tailored set of
operations on the data.
This is nicely (if verbosely) presented in the TidBITS articles on latency
that appeared two and three weeks ago.

My group has been integrating network option card function back into
the central processor of printers for better integration of data comm
and data interpretation.  Doing the same with modems and central (or 
near-central) processors is the right thing to do.

- tom]

Bob, as you know, I have the same config as you...

I downloaded and installed the latest Geoport / Express Modem software
updates and installed it just fine.  My understanding (it may not be fact)
is that our 6400's have Express Modems and these are not Geoports.  From what
I've read in the release notes etc. it is not clear whether the Express Modem 
can be / is upgradable to 33.6 from 28.8 via the install.  This is hard for me 
to test since I have no 33.6 sites which to dial into.  Digital has not 
upgraded the firmware in our Courier modems in the office....

The Telecomm software is the stuff which interacts with th modem such as
the FAX software and address book.


John 

    I think, but don't quote me...
    
    Express modems have a modem chip such as the Rockwell chip set, and use
    the Mac as the chip controller (implements the AT command set
    essentially).  The modem chip implements all the modem features and the
    FAX features.  The Mac just does the minimum to interface and transfer
    the data from the chip to memory or from memory to the chip.  The chip
    does the rest.
    
    The GeoPort is mostly the hareware to interface with the phone line and
    generate the modem signals.  The other modem functions compression,
    error correction, flow control, FAX implementation, etc, etc, etc...
    are handled by the Mac (a lot more responsibility).
    
    The advantage of giving more control in software is that it is possible
    to implement more features (Caller ID, Mac as an answering machine, Mac
    as the telephone to answer and place calls, Mac as the ability to
    answer the call and decide if it is a voice, FAX or modem incoming call
    and do the right thing.
    
    Now all of these advantages may not be implemented.
    
    The disavantages are that the more features implemented by the Mac, the
    more memory needed, and the more CPU overhead consumed while a modem
    connection is in use.

My understanding is that the Geoport adapter as it is included in the 6400 is
a "software" modem: There is the basic hardware, DSP and so on like any 
hardware modem, but the firmware is loaded from the Express Modem extension.
I have in the past installed the US Express Modem but then I had lost my french
modem default settings and Minitel emulation.
Using one of the ATI (ati5 maybe) you can identify the firmware.

After substitution of the Express Modem extension I restored the french modem 
characteristics. This is also how the 28.8 kbps rate can be upgraded to 
33.6 kbps.
Jean-Pierre

>My understanding is that the Geoport adapter as it is included in the 6400 is
>a "software" modem: There is the basic hardware, DSP and so on like any 
>hardware modem, but the firmware is loaded from the Express Modem extension.

    very interesting...
    in my friends 6400, after upgrading to 7.6, his express modem
    extension would no long load.  Said it would not work for this
    machine.  and the modem was dead
    bob

    Re .27
    
    > The advantage of giving more control in software is that it is possible
    > to implement more features (Caller ID, Mac as an answering machine, Mac
    > as the telephone to answer and place calls, Mac as the ability to
    > answer the call and decide if it is a voice, FAX or modem incoming call
    > and do the right thing.
    
    The SupraSonic modem implements all of these features, except that it
    has a microphone that plugs into the modem instead of into the Mac.
    
    > The disavantages are that the more features implemented by the Mac, the
    > more memory needed, and the more CPU overhead consumed while a modem
    > connection is in use.
     
    Bingo.  Anything that slows down the processor is a problem. 
    Obviously, I disagree with Tom (.25).  I was in this business when
    everything had to be done by the CPU, with stupid adapters hung off it,
    and we BEGGED for the ability to make our adapters smarter.  It was a
    landmark for the company I worked for back then, when I was allowed to
    design a Z-80 into a peripheral controller.
    
    It's fine to do tight integration on a single-purpose box like a
    printer.  On a CPU, which has to be a generalist, it's not a good idea
    IMHO - the more expertise you require from the CPU, the less of
    ANYTHING you leave for the other processes that it is running.

>     <<< Note 679.30 by SMURF::BINDER "Errabit quicquid errare potest." >>>
>     
>    Bingo.  Anything that slows down the processor is a problem. 

I think it is more appropriate to say that anything that slows
down THE WORK YOU NEED TO DO is a problem.  A faster processor may be a 
cheaper, better engineered solution than a smarter, special purpose "helper."

>    Obviously, I disagree with Tom (.25).  I was in this business when
>    everything had to be done by the CPU, with stupid adapters hung off it,
>    and we BEGGED for the ability to make our adapters smarter.  It was a
>    landmark for the company I worked for back then, when I was allowed to
>    design a Z-80 into a peripheral controller.

>    It's fine to do tight integration on a single-purpose box like a
>    printer.  

I will take (limited, impersonal) umbrage - a modern printer is NOT a 
single purpose box.  It does a lot of the same things that general 
purpose boxes-on-a-desk do, including handling data from many sources
(remote comm (in many flavors and protocols) and local operations), data 
interpretation and presentation, remote management, and complex computing.
We in printers have spent a fair amount of time defining interfaces
and implementations that address the point below.

> On a CPU, which has to be a generalist, it's not a good idea
>    IMHO - the more expertise you require from the CPU, the less of
>    ANYTHING you leave for the other processes that it is running.


Mildly arcane but VERY important point of engineering pedantry:

    Interfaces, whether between hardware components or people in an 
    organization, need to be selected and defined at points in their 
    respective processes where the interactions between the entities 
    on the two sides of the interface
        1) meet the bandwidth constraints available at the interface, and
        2) where the job breakdown allows appropriate decomposition
           or reconstitution of the workflow.

When modems were just MOdulator-DEModulators, point 2 was not violated.
When they adopted the burden of additional data compression, packetizing, 
fax encoding and handling, error detection/correction, and such, 
point 2 was overstepped.  The imposition of latency to allow for packetized
data comnpression is the example we were talking about in earlier replies.
Re-centralizing the work flow and/or defining new interface controls 
and breakdown becomes necessary.  In the absence of new standard of
task breakdown, re-centralization becomes the necessary approach.

- tom]