On the "Motorola vs. Intel" flames
Ian Merritt
With regard to the so-called Motorol vs. Intel flames that have been
traversing the news lately, Motorola vs. Intel doesn't quite adaquately
describe the argument. Rather, it is a question largely of the new,
albeit somewhat ignorant generation vs. the older, perhaps wiser
generation. By that I mean that Intel represents a generation of
microprocessor which, starting with the 8008/4004 stuff has
systematically ignored (initially for good reason) virtually all the
history of the computer industry which preceeded it. Motorola, however,
started with the basic design of the PDP-11/VAX style of architectures
which were based on a somewhat longer history of experience dating back
perhaps 20 years or more.
While Intel and what had by then become the microcomputer establishment
continued to produce small model/small-minded systems long after the
initial reasons for this (i.e. the I.C. technology of the time) had
passed, Motorola, having introduced the 6800 which was slow and perhaps
not all that useful produced the 6809 as the follow-on to the 6800, and
the 68000, their entry into what was to become the next major revolution
in processor technoloy: the supermicroprocessor; the micro that the
serious computer community, who had initially laughed at the
microprocessor, saying it would never be useful, would eventually
embrace.
The 6809 was really too late to play a serious role in the 8-bit market
which was by then dominated largely by the Z-80, though it was quite
nice for what it was, and even at its relatively slow 2 MHz processor
clock speed, showed better throughput figures than the 4 MHz Z80s
available at the time.
The 68000, however, significant of Motorola's realization at that early
date, that the 8-bit world was limited and its domination of the
microprocessor world was to be replaced by 16-bit processors and 32-bit
was not far behind, was the beginning of the maturation of the
microprocessor world. Though it had its problems, as would any new
technology, it was clearly the right direction to take, and had Motorola
done as good a job on their marketing and support as they had done in
the design of their processor family, I suspect it would have been the
dominant force in the microprocessor market, notwithstanding Motorola's
disinterest in selling out to Big Blue.
Meanwhile Intel, having introduced the 8086 in what year was it? 1977 or
so?, was willing to sell out, and because of that, and their ability to
produce millions of the little buggers, managed to make their way into
the IBM-PC and thereby into being the de-facto standard small-computer.
This coupled with the unfortunate popularity of the CP/M program,
sometimes laughably described as an operating system gave birth to the
MS-DOS system which is largely responsible for the generally stunted
state of public awareness of computer technology. Granted MS-DOS has
grown and now does much more than CP/M could ever have, but it is still
far from providing the level of standardization and support necessary in
a field of literally millions of installed systems.
Intel continued to elaborate that small architecture, making small
incremental speed improvements and and adding little gadgets here and
there, eventually evolving it into a slightly faster and massively more
complex small architecture. What we see today in the 386 is an
outgrowth of a history that started in 1977; an overgrown 8086 with some
of the architectural restrictions lifted and lots of special little
gadgets on chip which, if you can figure out how to use them, and you
agree with Intel's pre-determined model of operation, can actually speed
things up, though perhaps at a substantial cost of programmer time, code
size and complexity. It would be only fair to mention, however, that
the 186 did make a very nice communication controller for simple
applications where the processor didn't have to get too intimately
involved with the traffic itself. The on-chip DMAs worked reasonably
well, and for this purpose a 186 would have been the chip of choice.
On the other hand, Motorola, Dec, National and others continue to follow
the path which is a logical outgrowth of what has been learned over the
past perhaps 30 years of technological development, producing true
general purpose architectures which are simple to use and don't restrict
designers to the pre-determined ideas of one manufacturer. Powerful in
their simplicity, the 680xx, the 320xx and the VAX lines all exhibit an
orthogonal architecture, wherein the straightforward way of doing
something is usually also the fastest, and it is not necessary to
constantly examine the documentation to determine whether or not a
particular instruction is willing to operate on the combination of
registers which currently hold the data, or how to get around the
restrictions pre-determined task-switching model of some other special
on-chip 'feature'. This makes the lives of all who have to program them
much easier, and providing for much faster growth from a given level of
performance to the next, with upward compatibly which doesn't depend on
an emulation mode that sacrifices most of the advantage of the next
generation of chip, not at the cost of execution speed, however, as the
real benchmarks, those done by disinterested 3rd parties; not the
manufacturers, have shown. I have flamed enough, but I just HAD to
break in and say my part here, having seen all the flames from the Intel
people. The argument is ridiculous.
--I
Ps: I am not a "Motorola worshiper", nor do I have some sort of vendetta
against Intel. These are observations of what I have been watching
over the years of my participation in this industry. The issues
that have, for the past 10 years or so, dominated the industry, seem
to have been more political and less technical or forward thinking.
This is a very costly trend to our industry, and must change if we
are to survive as a driving force in the world.