Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!nike!ucbcad!ucbvax!ernie.Berkeley.EDU!shebanow
From: shebanow@ernie.Berkeley.EDU (Mike Shebanow)
Newsgroups: net.micro.mac,net.micro.68k
Subject: Re: What's Nu with VME for Mac?
Message-ID: <16299@ucbvax.BERKELEY.EDU>
Date: Sat, 1-Nov-86 17:37:53 EST
Article-I.D.: ucbvax.16299
Posted: Sat Nov  1 17:37:53 1986
Date-Received: Mon, 3-Nov-86 23:40:11 EST
References: <842@gould9.UUCP> <1240@hoptoad.uucp> <203@druil.UUCP>
Sender: usenet@ucbvax.BERKELEY.EDU
Reply-To: shebanow@ernie.Berkeley.EDU.UUCP (Andrew Shebanow)
Organization: University of California, Berkeley
Lines: 44
Keywords: NuBus, VME, slotted Mac
Xref: mnetor net.micro.mac:7778 net.micro.68k:1416

(sacrificial lamb line)

    While I can't say one way or another what kind of bus the new Mac
will use, I will put in my own two cents about VMEbus:

    I have been writing software for VMEbus based 68000/68020
systems for about three years now. All of this software has been
related in some form or another to image processing. After using
the VMEbus for such an extended period of time, I have two points
to make in favor of NuBus.

	A: VMEbus is still too slow for many image processing
	and other IO and/or memory intensive applications. Every
	application I have worked on has been VMEbus limited.
	On a 68020, adding memory on a VMEbus board exacts
	a severe performance penalty (4 or more wait states).
	Even using a fast memory bus like VMX32 doesn't give
	you access to 0 wait state memory. The NuBus, which
	is about 3x faster than VME, will make a large difference.

	B: Existing VMEBus cards will not be cheaper than the
	new cards that Apple and new third party developers
	will bring out, regardless of which bus Apple uses.
	Even simple IO cards for VMEbus cost at least $1000
	currently (by simple I mean a 4x8 parallel board with
	no onboard intelligence). Smart VMEbus boards like disk
	controllers typically cost $2000 or more. For a personal
	computer like Apple's, these prices are way out of line.
	Although I would expect the introduction of a mass
	market VMEbus computer to do wonders for VMEbus pricing,
	I don't expect that the small companies who make VMEbus
	boards today could compete with companies like Apple
	and AST when it comes to cutting manufacturing costs.

I think the second point is the more important one: no matter
which "high performance" bus Apple chooses, they will be forced
to do everything from scratch. So why not pick the bus which
has the better bandwidth, so that the machine will be as good
as it can possibly be?

I vote for NuBus (or FutureBus).

Andrew Shebanow
bus wierdo at large.