From the Fountainhead - April 1977

from the April 1977 issue of Interface Age magazine

by Adam Osborne

The first small signs of a long predicted shake out are beginning to appear among microprocessor manufacturers. The first two casualties are the Elecronic Arrays EA9002 and possibly the Rockwell PPS-8.

The EA9002 simply took too long to develop. The smallest EA9002 system has three chips: The EA9002 CPU with read/write memory, external read only memory and I/O ports. In mid-1976 that still stood a chance of making it; but now, for simple applications, the EA9002 has to compete with the one-chip F8 from Fairchild and Mostek— and the 8048 from Intel. In more complex applications the EA9002 must compete with the two-chip 8085 system from Intel and the two-chip 6500 system from MOS Technology. Mark the EA9002 as the first stillborn 8-bit microprocessor.

Rockwell has not withdrawn the PPS-8, but they are no longer pushing it.

The demise of the PPS-8 is a sad story. Here is a microprocessor that appeared on the market early enough to compete with the 8080 and 6800— and capable enough to have won the race. But no one at Rockwell ever understood what they had, nor how to market it; and now it is too late. The PPS-8 has more support devices, and had them sooner, than the 8080 or the 6800; Parallel I/O, Serial I/O, Direct Memory Access Control, Priority Interrupt Control, CRT Controllers, Floppy Disk Controllers and Keyboard Controllers— the PPS-8 had them all before anyone else. So why did it fail? Because it has a strange instruction set. The PPS-8 uses “memory pools”; these are dedicated areas of memory where you can store instruction object codes, immediate data (for immediate instructions) and addresses (for subroutine calls). You place a subset of your instruction set in the pools, then you can have 1-byte immediate instructions and 1-byte subroutine calls. Your instruction now reads:

“Execute immediate instruction #3”

or:

“Execute subroutine call #5”

and the PPS-8 goes to the pool area of program memory to find out what these instructions are.

Now PPS-8 pools are supposed to save program memory; in theory this is possible since the pools allow you to have 1-byte immediate instructions, which normally have to be two bytes long, and 1-byte subroutine Call instructions, which are usually three bytes long. But to achieve this economy of memory requires a level of programming effort that costs more than you could conceivably save by generating shorter programs; memory is inexpensive. And the confusion of pooling proved too much for many programmers who were having a hard time coping with assembly language programming in the first place.

But what Rockwell never figured out for themselves, and never advertised, was the fact that data pools could be used as a very powerful way of letting external logic control the microprocessor. By implementing the pooled area of program memory in bidirectional I/O ports, external logic could control immediate data, addresses, and even instruction object codes— thus generating very compact and extremely powerful microcomputer systems for industrial process control applications. And that represents a much larger part of the microprocessor market than most people realize.

So if you think microprocessor manufacturers know what they are doing, you are wrong; everything that has happened so far has been the result of chance and happenstance.

Now that Rockwell are discontinuing the PPS-8 and Electronic Arrays are discontinuing the EA9002 what will they do instead?

Electronic Arrays will probably do nothing. They are a small company who does not need a microprocessor in the product line.

Rumor has it that Rockwell will second source the MOS Technology 6500 series of products; Synertek never was a very viable second source. Having Rockwell as a second source for the 6500 may be just what the 6500 needs; MOS Technology has always been short of funds and incapable of providing the hardware and software support that 6500 users would like. Rockwell do not have cash flow problems.

There is also a rumor running around that Rockwell are producing their own Super Z-80.

Motorola will be staying essentially with the 6800 for the next year.

There is a 6802 which is a two-chip 6800 system. The instruction set is identical; all that has changed is logic distribution. A 6801 due out at the end of this year will be a one-chip 6800, again with the same 6800 instruction set. The 6802 is supposed to compete with the new 8085 from Intel while the 680I will compete with the 8048.

The only new upward compatible product which Motorola will introduce in 1978 is the 6809. This is going to be a 6800 with two or more Index registers and a somewhat improved instruction set. Details are not yet firm, but from what we hear, it certainly is a step in the right direction.

In my first column, I invited readers to telephone me if you have information that you think is interesting. My telephone number is (415) 548-2805. I extend the invitation again. So far the calls coming in indicate that there is need for some type of hobbyist “action line”. Most callers are upset because they feel that when they bought their hobby computer hardware or software they did not get what they paid for. At the moment, I do not plan to take on the mantle of the computer hobbyist consumer advocate because it would do no good. Most of the companies catering to the computer hobby market are marginal organizations; the reason they do not always deliver what they promise is because they would go broke and deliver nothing if they did. So before you buy hobby computer hardware just remember that you pay for what you get. If you want solid and reliable hardware built to the highest

engineering standards, go buy a minicomputer from Digital Equipment Corporation or Data General, and plan to spend 4 or 5 times as much money on it. Hobby computers are cheaper because the hobby computer manufacturers have cut corners. When you call the computer store with complaints, if you get no answer, it is not because they are trying to ignore you, rather it is because there is no one on the staff of the company whose job it is to listen to you; and when they start hiring the people to handle consumer complaints their payroll overheads will all increase.

Take your pick — do you want low cost or do you want service? You cannot have both.

Now as a hobbyist, if you wish to make sure that you get what you pay for, my only advice is that you buy products and not promises. Go down to your computer store and if you see it working in the store, buy it. If all you see is a brochure, don’t buy it; chances are they will cash your check and spend the money to build the products.

While on the subject of new products, watch out for the floppy disc system with the truly ugly name: “Micropolis”. Theirs will be the most sensibly-priced floppy on the market, and the guys who founded Micropolis are the same guys who founded Pertec. But for the ugly company name, they may have something; I wonder if the floppy will say “Kava Kava” every time it drops a bit.

Let us take stock of the software scene for the computer hobbyist.

Over the past year hobbyist magazines, including INTERFACE AGE, have done a superb job of printing source listings for a wide variety of imaginative and useful programs. In the beginning computer hobbyists distinguished themselves as thieves; they saw nothing wrong in copying programs for free, when they were legally supposed to pay a fee for the right to use programs. The result was that companies who planned to market software for hobbyists all went bankrupt in a hurry. But hobbyists have redeemed themselves by pitching in with large numbers of programs, all of which are donated free.

The only books of BASIC programs, up till now, have been ludicrously over priced and offer incompletely documented programs that frequently are less than advertised. I am planning to solve this problem during 1977 by publishing a series of low cost books describing general purpose programs, plus business application programs. Remember, a payroll program has more to it than calculating federal taxes based on gross income. There are a dozen different ways that people can be paid; but even that represents a small part of the problem. The really hard part of writing business application programs is coming up with complete operator dialogue schemes that second guess every silly thing an operator can do and give you easy methods of correcting erroneous data entry. File creation and maintenance is equally important. If your business application programs do not have adequate coverage in these areas, they are not adequate business application programs.