Economy of scale and specialization—revisited
For many years conventional wisdom stated that computers possess large economies of scale. This notion was formalized by a relation stating that a computer’s power is proportional to the square of its cost (8, 10). More than a decade ago the universality of this relation (1) was questioned and succeeding events have further limited its applicability. Substantial economies have appeared in small specialized systems frequently overshadowing the economies associated with large centralized systems (5). It is hard to believe that not many years ago many computer center directors insisted that all computing within their institution be carried out on one large central machine. The success of minicomputers has forced the realization on even the computation center faithful that large systems can frequently be outperformed by smaller special purpose systems.
KeywordsFast Fourier Transform System Cost Economical Computing Total System Cost Core Memory
Unable to display preview. Download preview PDF.
- 1.Adams, C. W. Grosch’s law repealed. Datamation 8(7): 38–39, 1962.Google Scholar
- 2.Bell, C. G., and J. Grason. The register transfer module design concept. Computer Design 10(5): 87–94, 1971.Google Scholar
- 3.Clark, W. A., and C. E. Molnar. Macromodular computer system. In: Computers in Biomedical Research, edited by R. W. Stacy and B. Waxman. New York: Academic, 1974, vol. 4.Google Scholar
- 5.Cox, Jr., J. R. Economy of scale and specialization. Presented at Future Goals of Engineering in Biology and Medicine, Washington, D. C., 1967. Reprinted in Computer Design 7(11): 77–80, 1968.Google Scholar
- 7.G-AE Subcommittee on Measurement Concepts. What is the fast Fourier transform? IEEE Trans. A-E AU-15: 45-55, 1967.Google Scholar
- 8.Knight, K. E. Changes in computer performance. Datamation 12(9): 40–54, 1966.Google Scholar
- 9.Signetics memory systems. Economic Advantages of Microprogramming. January, 1971.Google Scholar