Two reports for marketing executives have been published byElectronic Trend Publications, of San Jose, Calif. Their titlesare "Board-Level Embedded Computer Markets and Trends" and "TheWorldwide IC Packaging Market."
Board-level computers have taken over functions previously doneeither mechanically or with analog systems, or perhaps not doneat all because of cost. Most of them are embedded in largersystems, some to the extent that the user is totally unaware ofthem (deeply embedded), others where their presence is obviousbut where they are still essential to the system (highlyembedded). These definitions are loose; the report on thissubject discusses how the computer industry tries to tightenthem.
Standardization has assured, and in the future will continue toassure, that successful older applictions have survived and willcontinue to survive -- although their implementations may shrinkas advances in technology allow more and more transistors to besqueezed onto a single chip. The report lists well-establishedstandards and others that remain open to a certain extent, anddescribes how these standards came to exist.
Shipments of such computers are expected to grow from about 1billion in 1993 to 3 billion in 1999. This growth consists inpart of sales of VME products (1 billion to 2 billion during thesame years), and in part of sales of PC-based hardware ($500million to $1.4 billion); sales of systems based on otherstandards will be more nearly flat and may possibly decline. Thereport discusses these trends in more detail.
Long-range forecasts (beyond five years) are unrealistic, thereport says. Nevertheless, some generalized trends are evident.
For one thing, manufacturers who are well-established with oneparticular standard are reluctant to change, as long as thestandard permits products to be updated. Such changes requirenew software, new mechanical designs, and re-qualification -- allof which translates into new money. This reluctance tends toperpetuate the use of current standards, updated as technologyrequires.
Other generalizations: The size of manufacturers (measured indollar volume) will increase and their number decrease, throughattrition, mergers, and acquisitions. Processing requirementswill require increased use of multiprocessors; single-processorsystems will tend to fall by the wayside. New implementations ofsoftware, protocol, and logic are likely to arise.
"In the final analysis," the report says, "technology is likelyto move standards more slowly than many people would like tobelieve. . . . In the meantime, board-level technology willcontinue unabated at least through the first decade of the 21stcentury."
The other report, "The Worldwide IC Packaging Market" discussesthe outlook for various kinds of IC packages, classified by thenumber of pins (external connections). Highest volumes can beexpected for the smallest packages, of no more than 18 pins, butthe largest compound annual growth rates are anticipated for thegiant packages of 200 or more pins -- rates that are actuallyunrealistic because they are based on small numbers.
They are examples of a statistical fallacy. If the murder ratelast year in a certain village is reported to be 500 per 1,000residents, that is indeed horrifying, until one realizes that atthe beginning of the year the total population of the village was2, before one of the residents killed the other; the villagesuffered only one murder. (The county seat of one county inwestern Texas, which was the county seat because there were noother towns in that county, did indeed have a population of 2just a few years ago; whether it still does, or what its murderrate is, is uncertain. But East Palo Alto, Calif., was recentlyin the news as having one of the highest murder rates in the U.S.-- rather meaningless, considering the size of East Palo Alto.) Likewise, if 1,000 ICs with 300 pins were sold last year, and2,000 are expected to be sold in the year 2000, that is indeed a100% simple growth rate (14.9% compounded annually), but it isnot as impressive as an increase in 8-pin ICs from 20 billion to30 billion in those years.
Classifying shipments by other categories is also interesting. Dual in-line packages, in the early 1970s practically the onlykind of package in general use, are declining at about 2% peryear, whereas all other common configurations are increasing --some spectacularly so, but again look out for the statisticalfallacy.
Much growth is expected in the smaller packages with closelyspaced pins, for use in such things as notebook computers andcellular telephones. Meanwhile, surface-mount packages arerapidly replacing packages whose pins protrude through themounting board.
A disadvantage in both reports is their tendency to use"alphabet-soup" acronyms, in many cases without spelling out thedesignations on first use. Many readers will recognize theseabbreviations, but others more familiar with the almighty dollarthan with the details of technology may be puzzled by them.
Complete copies of the full reports are priced at $1,795 for the"Board-Level" report (320 pages) and $1,995 for the "Packaging"report (also 320 pages). Both may be ordered from ElectronicTrend Publications. Electronic Trend Publications
1975 Hamilton Avenue, Suite 6
San Jose, Calif. 95125
Return to Headlines