Moore's law

The mankind always tried to glance in the future. Its best minds to the extent of the forces tried to solve this problem, using various techniques and tools.

Hundreds years and even thousand years, practically, all predictions were grounded on a narcotic trance of numerous priests, foretellers and prophets. The contribution to future prediction has brought whole and army of the astrologists who have created, perhaps, millions of various horoscopes.

In our more educated time for these purposes rather difficult scientific techniques of an estimation of the researched objects providing construction of various mathematical models use. As a basis such resources as time-series analysis, regressionnyj the analysis, spectral analysis, etc. are applied, for example . Construction of adequate models, as a rule, becomes extreme a challenge, but their usage and interpretation of received results is not less difficult. Cases when the model represents very bulky system from tens are frequent, and is frequent also hundreds the nonlinear equations badly giving in to simplification and fairly loading during calculations modern computers. And the more difficult researched objects, the turn out models more difficult.

However in the difficult world of numerous resources of the analysis and forecasts there are the tools confirming the well-known thesis: all ingenious — is simple. Such tool of the forecast of development of electronic and computer industries already almost four ten years is Moore's law. This law excites many years the public and does not allow to sleep easy to the legislators of the computer industry presented by experts and businessmen, substantially defining its forward development.

More three tens years ago Gordon Moore (Gordon Moore, Chairman Emeritus of Intel Corporation) has predicted regular doubling of number of transistors in difficult electronic arrangements. Now it is named Moore's as law (Moore's Law).

Fig. 1. Gordon Moore (Gordon Moore, Chairman Emeritus of Intel Corporation)

However what events preceded and how there was an opening of the given surprising law?

The answer to this question Gordon Moore in the short video film shown during performance of the vice-president of corporation of Intel of Patrick Gelsingera (Patrick Gelsinger, Vice President, Chief Technology Officer has given, Intel Corporation) on IDF Spring 2002, the past in San Francisco, California, the USA, in the end of February, 2002.

Here it is necessary to remind, that forum IDF (IDF — Intel Developer Forum) is the largest action for implementators of hardware and the software and is spent some times in a year. The forum gathers leading representatives of a science and engineering for consideration of the various questions linked to the advanced computer processing techniques and production of electronic branches,

As to Moore's law its opening has taken place extremely simply. The log on electronics in 1965 has asked to predict Gordona Moore development of the semiconducting industry on the following of 10 years. And as Gordon Moore, at that time has marked the chips accessible in the market, were to modern measures rather simple. The most difficult contained all three ten units on the chip. However in laboratory there was already experimental chip consisting of 60 components. Gordon Moore has analyzed possibilities of processing techniques existing at that time and rates of complicating of semiconductor chips. Further it has fulfilled extrapolation for 10 years and has received the legitimacy formulated above predicting appearance of very difficult chips from several tens thousand of transistors. Results of the given analysis have been presented in appropriate article: Gordon E. Moore. Cramming more components onto integrated circuits. Electronics, Volume 38, Number 8, April 19, 1965. (Gordon E. Moore, Director, Research and Development Laboratories, Fairchild Semiconductor division of Fairchild Camera and Instrument Corp.). Resulted below a drawing from specified article well illustrate an estimation and the forecast of development of the semiconducting industry. By the way, it is necessary to remember, that this operation has been done in 1965 in the conditions of rather small number of data.

Fig. 2. Change of value of a metric the price/component depending on number of units in the chip (from Gordona Moore's published in 1965 ) article 

Fig. 3. The forecast of growth of number of units in chips (from Gordona Moore's published in 1965 ) article 

As it has appeared, accuracy of a prediction of rate of complicating of the architecture of electronic chips has appeared much more above, than someone expected. According to the forecast doubling of number of components was foretold every year, that precisely enough corresponded to evolution of chips for the period prior to the beginning of the seventieth years. Further with correspondence with development of the semiconducting industry and accumulation of data Moore's law has been modified: doubling of number of transistors in processors should occur each one and a half-two year.

For good illustration of justice of this law the schedule showing growth of number of transistors, making a basis of the architecture of processors of the corporation of Intel is.

Fig. 4. Growth of number of transistors in processors

Year of release and quantity of transistors in Intel processors

Predictably, in 2007 there will be processors with number of transistors one billion. And as Krejg Barrett in the report on IDF Spring 2002 has marked, the next fifteen years development of solid-state technologies will allow implementators of processors to realise following characteristics:

• 2 billion transistors,
• Clock rate of processors will reach 30 GHz,
• 1 billion instructions a second,
• The size of transistors — 10 nanometers (0,01 microns),
• There is possible usage of wafers 18 "(transition with 8" on 12 "slices now is carried out).

Article continuation>>

To pass to razrelu Processors