Guest Commentary

Is RAM Memory A Status Symbol?

Barry Miles

Many expensive technological items are bought as status symbols. Are all those Hewlett Packard HP 41c's really used to their fullest extent, for long programs and the use of ROM libraries of fancy programs, or are they merely left on the executive's desk to say "I'm so important that I can justify a purchase of the state-of-the-art programmable calculator"?

The advent of really large RAM sizes means that we should rethink the relationship between RAM and disk storage. We have for a long time lived with the idea that we should use RAM sparingly. This probably stems from the need to conserve RAM usage in a mainframe environment, so that as many users as possible may access the machine at once and so that the queuing problem is reduced to a minimum. Programmers are likely to continue to think in this way, even when the need has evaporated.

Perhaps an example should be taken from the approach used in managerial economics. In budgeting for the future, businessmen seek to identify the Principal Budget Factor - that factor which prevents the business from expanding to infinity. They then seek to make the very best use of that scarce resource, so as to maximize profits. They usually make strenuous efforts to remove the bottleneck which that resource represents, by increasing the amount of it which is available: if you are short of skilled labor, you seek to take on more people, for instance. The successful businessmen are the ones who first remove the constraint which is holding them back, then correctly identify the new constraint and seek to remove it, and so on.

What I am saying is that once RAM ceases to be a scarce resource, we should cease trying to economize in its use, especially as it becomes progressively cheaper, and particularly when it becomes cheaper than similar amounts of secondary storage (such as disks or tapes).

A potential buyer of the Sinus computer has an interesting choice before him; with a limited budget, he will need to decide between various amounts of RAM, and whether to go for double-sided disks to increase secondary storage capacity. He may choose the largest amount of RAM, out of habit, without really considering whether he will make effective use of the extra memory.

More Is Less

Again, economics may come to our aid. The Diminishing Marginal Utility theory says in this context that every extra 1K of RAM is less important to us than the previous one, to the point where more is really of no interest.

Surely we must examine whether what we are doing now will become easier, faster, or more efficient if we have more RAM, and whether there are other things which we could do with more RAM but which are impossible at present, and finally whether we should adopt a whole new approach. There is a danger of misleading ourselves or of being misled by salesmen into thinking that more RAM must be a good idea, without thinking out why. There is even a danger of rationalizing in order to justify what is really only wish-fulfillment.

We might compare this to buying a fast car. Some say that you're much safer in a fast car than in a slower car, regardless of the speed at which you are traveling. The braking system and suspension of such a car have been designed to cope with the effects of traveling quickly, and these systems therefore work very much within their capacity, and very efficiently at slower speeds. A similar argument can be made for extremely powerful hi-fi systems: distortion is less if you do not have to turn up the volume very far to get the loudness you require.

Do these arguments carry over to microcomputer memories? Probably not. The trouble is that you merely get more of the same. If you do not use it, then it just lies idle. Are you really going to write massive BASIC or machine code programs? Are you really going to handle vast amounts of data? Most likely not, at least not unless you change your way of doing things to optimize the use of your principal technological factor.

New Freedoms

What I am suggesting is that disks came about because of limited RAM. Now that RAM limitations can be of increasing greater size, we should explore new freedoms. What follows may seem a little far-fetched, but may also be just around the corner.

First, we may take it that a one megabyte RAM is not likely to be filled with a BASIC or machine code program of anything near that length. The debugging alone would take too long! This leaves us with other possibilities.

We could fill a lot of the RAM with a wide range of programs, and call up any of the whole suite, instantaneously, from a special menu program.

We could have as many programming aids in our machine as we could conceivably wish for, and barely scratch the surface of our new-found capacity.

We could have a vast range of help screens available for instantaneous recall when in trouble.

We could call in a whole succession of high resolution pictures, which are usually slow to load from disk, so rapidly that even animation would be possible.

We could have split processing in one machine. After all, it is common for two processors to be in one machine, so why not a schizoid machine with each part operating independently?

We could have a really enormous amount of text in our word processor at any one time, and have many different text areas. Our word processor could perhaps interact with our accounting and data base programs in RAM.

Accounting suites of programs could be truly integrated, so that final accounts are updated after every transaction.

Our data bases could be loaded from disk into RAM first thing in the morning, and all updating could take place in RAM, so as to be almost instantaneous. All the disk activity would have to do is merely dump RAM contents, for safety's sake, at convenient time intervals. Battery backup could protect contents from voltage spikes and power failures.

It might be that disks of all types will become a thing of the past, with programs and data being loaded and dumped over the telephone by a modem, with suitable passwords and protections, into your friendly local overnight datastore. (There are problems in this, in that the use of telephone lines is subject to error, but presumably this will improve and is not an insurmountable obstacle.)

In any case, if the function of the disk unit changes from continual random access to infrequent loading and dumping, disk operating systems could be simplified at the very least. Perhaps the very small diameter disks which the major companies are now developing will become the norm; and disk units will come down in price to become a trivial expense. That, too, is an intriguing prospect.

This would all require greater addressability than even the current 16 bit machines offer, but the megabyte chip is probably just around the corner.