Читать книгу Digital Universe - Peter B. Seel - Страница 16

In 1965, then-Fairchild Semiconductor executive Gordon Moore published a short article in the April issue of Electronics magazine entitled “Cramming more components onto integrated circuits.”⁹ In the article, Moore predicted that within a decade (by 1975), evolving silicon chip technology would permit the fabrication of integrated circuits (ICs) with 65,000 components (transistors) on a single chip. Given the state of IC manufacturing in 1965, his then-startling prediction suggested that the number of transistors on a chip would double each year in the decade between 1965 and 1975. Moore included a table (see Figure 2.3) that featured a logarithmic scale demonstrating this doubling of components on a chip from 1962 to 1965 and then extending this plot into the future. I have reversed the X and Y scales in the version below (with time on the Y scale) for the sake of clarity. Note that this calculation was based on just four confirmed data points (1962 to 1965), and was quite a bold prognostication given the predicted doubling of components at yearly intervals. Yet Moore’s prediction for this remarkable technological feat proved to be prescient, even if the doubling intervals were to be closer to 18 to 24 months and are now approximately 30 months.

Figure 2.3 Moore’s Law Re-plotted. Source: Modified by author after the original in Electronics, Vol. 38, Number 8.

Three years later, Moore left his position as director of the research and development laboratories at Fairchild to start a new company with partners Robert Noyce and Andrew Grove. Its name was short and memorable – the Intel Corporation. In 1975, Moore revised the time-frame for chip evolution from one year intervals to two years in a speech he gave to the Institute of Electrical and Electronics Engineers (IEEE).¹⁰ For several decades, Moore modestly declined the honor of having the law named after him and attributes the name to California Institute of Technology computer scientist Carver Mead.¹¹ It was to provide a prescient method of reference for the exponential growth in the power of integrated circuits over the following 60 years. It is important to note that the meaning of the law has evolved over time from a literal count of the transistors on a chip, to a contemporary interpretation related to the processing power and speed of the components on a multifunction chip.¹²

Computer users understand and appreciate the improvements in processing speed and reduced power demands in these chips, especially those developed in the last decade. Other uses of IC technology are less obvious. Automobiles, for example, have a number of computer chips that govern critical functions such as fuel injection, collision-avoidance features, and wireless sensors that automatically sync to a mobile phone for hands-free use that reduces driver distraction. The development of driverless vehicles by Tesla, Google, and other manufacturers is reliant on a host of sensors mounted around the car or truck to track barriers and obstacles and then relay this information instantaneously to AI-driven processors that control steering and braking. The sensor-processor network can respond to potential collisions much faster than a distracted human can react and take evasive action.