Path: utzoo!attcan!uunet!lll-winken!ames!mailrus!cwjcc!ukma!gatech!rutgers!cbmvax!daveh
From: daveh@cbmvax.UUCP (Dave Haynie)
Newsgroups: comp.sys.mac
Subject: Re: New Macs
Message-ID: <6105@cbmvax.UUCP>
Date: 28 Feb 89 22:57:43 GMT
References: <975@wpi.wpi.edu>
Organization: Commodore Technology, West Chester, PA
Lines: 60

in article <975@wpi.wpi.edu>, tron@wpi.wpi.edu (Richard G Brewer) says:
>>In article <8400065@m.cs.uiuc.edu> gillies@m.cs.uiuc.edu writes:
>>>I think Apple has shot itself in the foot.

>>>People resist PS/2's & the microchannel bus partly because they must 
>>>throw away their AT/XT cards.  Now people can resist the SE/030 for the 
>>>same reason!

> That's not the case at all, as any hardware engineer could tell you. 

I'm a rather unbiased hardware engineer (I don't use Apples or IBMs) and
I'll tell you it's not the case, but for some additional reasons.

> The SE uses a 16 bit data bus - a limit od the 68000 16/32 bit architecture.
> The SE/30 is a true 32 bit machine. Putting a 16 bit slot in it would have 
> done 2 things :

> 1. slowed down the data bus speed 
> 2. MAJOR SYSTEM HEADACHES. 

Certainly #1 is true, and as far as I know #2 could be reasonable.  But there's
a more fundamental difference between SE slots and IBM slots.  Obviously IBM
could make the same arguments about Microchannel vs. AT buses; that the AT
bus is going to slow down their fast 16 bit or 32 bit machines, it being a
basic enhancement to a not that altogether clever 8 bit expansion bus.  But
the main point is, the AT bus is designed as a general purpose expansion bus
for multiple expansion cards.  So is Microchannel, NuBus, VME, Multibus,
and Futurebus.  Certainly the AT bus is _implemented_ as not much more than
an extension of the 8088/80286 bus, but that's just an implementation detail.
It serves the function of a general purpose bus.  That NuBus and some of the
others look less like direct CPU buses is just the realization that some of
the things done on CPU buses are hard to implement or lead to unwanted system
dependencies on general purpose expansion buses.

I think the difference between all of that and the Mac SE expansion connectors
is that the latter are not intended as general purpose expansion buses, but
merely a basic mechanism to bring out enough of the 68000 or 68030 lines to
allow devices to be added.  Each slot can directly accept one such device,
but if you really wanted multiple-device expansion, the best thing to do would
be to attach a NuBus backplane to your SE via an adaptor card that fits in
the expansion slot.  This slot was never designed to be a general purpose
expansion bus, and tends to be very specific to the machine it's in.  To
implement the SE slot on the SE/30 would require a considerable amount of
external electronics to convert 68030 bus signals to something 68000 
compatible, and of course back again.

On the Amiga machines I work on, and some modern PC compatibles as well, it
often makes sense to offer both kinds of slots.  The general purpose expansion
slot gives you access to a large number of types of cards that work with any
machine that supports the general expansion bus.  The machine specific expansion
slot allows direct access to the CPU bus, which for the Amiga lets us design
in the capability to add fast 68020 and 68030 cards to the 68000 machine, and
for some PC machines lets memory be added that is much faster than anything 
which could be added to the general purpose expansion bus.  

> Richard G. Brewer
-- 
Dave Haynie  "The 32 Bit Guy"     Commodore-Amiga  "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: D-DAVE H     BIX: hazy
              Amiga -- It's not just a job, it's an obsession