ihm@minnie.UUCP (Ian Merritt) (10/06/86)
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.