Path: utzoo!utgpu!water!watmath!clyde!att!pacbell!ames!pasteur!ucbvax!decwrl!nsc!pyramid!octopus!pete
From: pete@octopus.UUCP (Pete Holzmann)
Newsgroups: comp.sys.ibm.pc
Subject: Re: GATHER and say NO to MCA!
Summary: AT bus is fine, as long as you extend for 32 bit memory!
Message-ID: <257@octopus.UUCP>
Date: 26 Jun 88 14:21:21 GMT
References: <42900016@uicsrd.csrd.uiuc.edu>
Reply-To: pete@octopus.UUCP (Pete Holzmann)
Organization: Octopus Enterprises, Cupertino CA
Lines: 81

In article <42900016@uicsrd.csrd.uiuc.edu> leein@uicsrd.csrd.uiuc.edu writes:
>Dear PC clone makers;
>...Why do you not say no to IBM's
>proprietary Micro Channel Bus Architecture?  The possible answer I can
>expect from such companies as Ta??? and De?? is that they do not have
>the technology to develop their own version of OS/2 for an advanced
>32-bit bus they might choose instead of MCA.  I think this is the
>primary reason the other PC clone makers are still stuck to AT
>bus which is out of dated for the present 386 age.
>...
>   So listen to me and gather as OSF members did, and choose a
>new advanced bus like NuBus (which is in the public domain) or 
>Multibus II (which is supposedly better (? well...) for the Intel chips) and
>make your own version of OS/2 to this new advanced bus standard.

*If* the AT bus were really 'out of date' for '386 machines, then it might
make sense to switch to a completely new architecture. If that were true,
though, they'd all jump to IBM's lead and use MCA. IBM, unfortunately, *does*
set the standards in the eyes of the big corporations, and that's what counts.
We may not like it, but its the truth.

Fortunately, the big corporations have accepted another standard for PC
bus architecture (the AT bus), and everybody is using it. Is it *really*
out of date? Too slow? I think not! Sure, it doesn't incorporate all the
latest in bus architecture knowledge, but then it didn't do that even when
first used!

Consider IBM's claims for MCA, and the facts regarding the AT bus, and decide
for yourself:

1) MCA is cleaner, newer, nicer, etc etc:
	All true, all irrelevant. On a bus, what works is what counts! The
	AT bus was supposedly clean, new and nice when implemented, but all
	the hardware people in the world surely knew that it wasn't. Didn't
	make any difference at all. 

2) MCA is faster. AT bus is too slow.
	Various manufacturers have SCSI controllers that can pump 4-5 MB/sec
	through the 'slow' AT bus. There isn't a MAINFRAME disk drive that
	can use up that bandwidth, as far as I know. Let alone a wimpy
	Winchester drive! 

	I've never seen a situation yet where the AT bus was the limiting
	factor for I/O. Usually it is the rather slow DMA chip on the AT 
	motherboard.  When faster throughput is needed, board makers put their
	own, faster, DMA controller on their board.

	The AT bus can become a bottleneck for memory access. But board
	makers seem to be doing a good job of getting around this. A 32 bit
	extension standard would help a lot in allowing standardized 32 bit
	RAM cards for 386 machines.

3) MCA handles multiple CPU's.
	So does the AT bus. A kludge, true, but you can get a 68020, 32032,
	etc coprocessor board for your AT. Even for a PC for that matter,
	but then the 8 bit bus really starts to cause trouble! This brings
	up the real limitation of the 'AT' bus: we need a standardized 32
	bit extension! The MCA can handle >2 processors, so someday and
	in some high end applications, the AT bus may be a liability for
	this.

My conclusion: the clone-makers need to pick a 32-bit AT bus extension
	standard. There is little engineering reason (right now) to go to the
	MCA.

Pete

PS: If you're inclined to believe IBM's marketing hype regarding how wonderful
	and unique the MCA is, I suggest you think about their OS/2
	announcement at COMDEX last fall. They were proclaiming that with
	OS/2, you could *finally* get multitasking on a PC, while, in another
	booth, Quarterdeck [Desqview] was merrily DEMONSTRATING multitasking
	on various flavors of PC's, using lots of existing software, etc etc.
	It was a sad commentary on the power of marketing.
-- 
  OOO   __| ___      Peter Holzmann, Octopus Enterprises
 OOOOOOO___/ _______ USPS: 19611 La Mar Court, Cupertino, CA 95014
  OOOOO \___/        UUCP: {hpda,pyramid}!octopus!pete
___| \_____          Phone: 408/996-7746