Martin Banks, Personal Computer World 12/82 - checked

December 1982

It is some time since I last wrote in any seriousness about chippery and things in this august magazine (Late again, Banks: this is the December issue - Ed), though it is a subject that is quite dear to my heart. So this month, given that something of interest has occurred in the semiconductor business, I am going to have a go.

In fact, I have in mind two distinct and separate subjects for this month and their only connection is the semiconductor industry, so I'm going to try to stick them together, with only a half-hearted attempt at finding a more cogent link between them.

The first is to do with marketing, and the application of the marketing ploys learned in the semiconductor industry to the personal computer business. The second is to do with the application of the technology of memories, in the form of firmware. No, this is not something that might prove of interest to Mary Whitehouse, but it is to do with the storage of software in hardware form, which to the gizmo-freak means ROMs, PROMs, and EPROMs.

The economies of semiconductor manufacture are those of many high technology, high capital cost products - it costs as much to produce one chip as it does a million of them. As there is more chance of selling a million - because a lower unit cost can be charged - than there is of selling one very expensive chip (the high price being to cover the high capital costs) then the semiconductor industry duly makes millions of them.

The drawback of this is that the chips themselves have to be of a type that users want to buy by the million, so, except in occasional circumstances, there is little profit in the semiconductor companies producing anything too far out of the ordinary.

Indeed, it is a general rule that uniqueness in chip design or manufacture is akin to the kiss of death. No self-respecting systems manufacturer is going to commit a new design to the production line when that product employs a type of chip that comes from just one manufacturer. Should that supplier go delinquent, which is a delightful American way of saying screw up on the supply of chips, then the whole production run, and the investment involved, can go down the Swanee.

So chip customers seek second sources, an alternate source of supply for the same part. The semiconductor companies now go to great lengths to ensure that such second sources are available to the customer, usually by licensing their 'rivals'.

Such 'second sourcing' has now come to the personal computer business, and it has come to Europe, not America. The particular deal in question specifically precludes our American friends from taking advantage, as will be revealed.

At the beginning of this year, the observant among you would have noticed the arrival of the Sirius 1 system from the drawing board of one Chuck Peddle, father of both the PET and the 6502 microprocessor. The Sirius company was funded in its efforts to develop the machine by a deal in the USA with Victor Business Machines. By way of recompense, Victor took the rights to the US market for the machine, which is manufactured for them by Sirius. Both companies are allowed to sell into the other markets around the world.

Victor declined to call their version of the machine the Sirius. Instead, they called it the Victor 9000. Well, you might ask, what's in a name? - and for European customer that would indeed be true, because the customers are getting real and legal second sourcing.

Yes, the recent introduction into this country of the Victor 9000 means that the customer has the opportunity of making a choice in the purchase of a system. And, if it happens to be the Sirius/Victor system being considered, to go out and shop from alternative sources.

From the user's point of view this can only be a good thing, for it will give greater scope to these all-important subsidiaries - things like service, support, and specialist knowledge, as well as price and delivery - that all dealers should find it important to offer. If a dealer in one type of system seems poor, the chances are there will be a dealer in the other system not too far away.

This could, of course cause a problem for the dealers themselves, and even for the two importing companies: ACT for the Sirius, and DRG for the Victor. The latter is hoping to have 100 dealers around the country and, according to them, some 60 percent of the dealers that have made an approach are existing Sirius dealers.

DRG has said that it has a working understanding with ACT, especially on such items as pricing, and is not against appointing dealers that already handle the ACT import. It will be interesting to see whether this understanding stands the test of business practice.

Having two sources for the same machine (the only differences are a slightly different keyboard and a re-sprayed box), especially when that machine looks like being extremely popular, can only be good for the customer.

And what is good for the customer forms the extremely tenuous link to the second theme for the month, which is all to do with firmware. Now this is the trick of loading commonly used software into a hardware form such as ROM or PROM. It is an idea that has both advantages and disadvantages, though its long-term potential as an aide to systems design, systems configuration, and user choice is considerable.

The reason I mention it this month is because of an announcement by Intel, the company that brought you the 8080 and the 8086 processors. As its 8088/8086 family looks like being the biggie in the 16-bit systems stakes, it being at the heart of the Sirius/Victor and the IBM PC among other machines, it is sensible for the company to look for other related areas where it can turn an honest dollar.

Now, as already mentioned, the real application of the firmware idea is with commonly used software. This is because of the nature of semiconductor manufacturing, where volume production is king. Only with software that is fully debugged and in great demand does firmware really make sense.

In the 16-bit marketplace, such a target for firmware is an operating system, and it is one of these that Intel has selected. The one it has gone for is CP/M-86 from Digital Research, for which Intel has just taken out a licence.

It is now in the throes of finalising a new chip that will incorporate the operating system in silicon. To be called the 80150, the chip is intended for use with the 8086, 8088 and the latest Intel processor offering, the 80186.

The chip will contain Version 1.1 of CP/M-86, held in 16 Kbytes of ROM, which will be accompanied on the chip by some important hardware functions needed by the operating system, such as timers and an interrupt controller.

The obvious market for such a chip is in the vast army of stand-alone systems currently being produced by the likes of Sirius/Victor and IBM. It makes a great deal of sense from everyone's point of view, for by having not only the operating system, but also some of the attendant chippery required for it, resident on a single chip, systems manufacturing costs will be brought down. It will also make the design of any new systems based on the Intel family that much easier, and therefore cheaper(?). Not surprisingly, it will lock systems manufacturers even more tightly into the grip of Intel (and almost certainly its official second source suppliers), which will no doubt please the company.

For the user, this should mean not only cheaper systems. It should also give (in theory at least) more reliable systems, as the chip should be relatively bug-free, and certainly free from being corrupted as it might be in disk form. It also makes a system more user-friendly as the O/S becomes an integral part of the hardware, with the booting-up procedure totally transparent to the user.

It is also one more step along the road to the appearance of more general software in a firmware form. This could well include some complete applications packages that are popular and 'standardised', such as the 'Calc' programs and the word processing packages. Games packages will be (damn it, are) another target market for firmware. Other applications areas, for example accounting, may well be too complex and variable for such overall treatment, though it is likely that 'tailored' programs may be constructed out of ranges of utility packages and sub-routines that can be linked together.

System configuration could then become easy. The user could actually do it himself. . . perhaps. A system would be a processor, memory, and spare sockets into which the user popped whatever ROMs held the software needed. The ROM'ed O/S would do the rest.

end