On April 19, 1965, at number thirty-fifth anniversary of the magazine Electronics, Gordon Moore, then director of research and development company Fairchild, published an article predicting that the number of components in integrated circuits would double every 12 months. A decade later, Moore corrected the number, and was as we know it today: the component density will double every two years. Tomorrow, then, Moore’s Law will meet half century.
Fifty Years in digital technology is an enormity. Yet Moore forecasts have been fulfilled to the letter. At least, until now.
By the way, there is a history of this prediction, in which Moore was inspired. Another computer genius, Douglas Engelbart, precursor of graphic interfaces and hypertext and inventor of the mouse, had anticipated, although no figures, the phenomenal miniaturization was coming.
Aside from its extraordinary longevity and successful prediction, Moore’s law is also notable for misunderstandings, confusion and myths that surround it. A major is in its very definition.
What says
Moore predicted that the density of transistors would double every two years and nothing else. The issue is a bit more complex and has to do with the economy of the silicon industry (silicon is the key semiconductor chip manufacturing; there is also the name Silicon Valley, literally, Silicon Valley).
Since Robert Noyce invented the integrated circuit, Moore had not one but two observations. The first was obvious: the chips that had been manufactured by Fairchild doubled the number of components each year. The second was more subtle.
Moore knew that simple circuits the cost per component was inversely proportional to the number of components. That is, more components, less cost. But he also noticed that if you were adding components, a point is reached where the cost began to increase. In fact, this is the first chart on your article.
Moore postulated that every moment of technological developments there is a minimum cost, which is up when circuit has a certain amount of components. At the time of writing your article, that number was 50, and estimated that in five years increase to 1000.
So yes, roughly Moore’s law postulates that the density of circuits will double every two years. But the economic aspect is crucial, because there would be no point scoring 100 times more transistors in an electronic brain was costing several million dollars. And I would not have because, as also notes the text of Moore, chip manufacturing is an industry scale.
“Every time we opened a new microprocessor architecture factory must be installed, and make it difficult 6 billion dollars. And still counting the cost of research and development of this new architecture. To amortize all that is necessary to make and sell many chips “, he explains by phone Germán Loureiro, technical manager of Intel Argentina, whom I called . to check the most difficult aspects of Article
In total, the correct statement of Moore’s Law, as it appeared in the magazine Electronics, said: “The complexity to achieve the minimum component cost doubles approximately every two years. “
By the way, have noticed that the word” component “appears in the preceding paragraphs insufferable excess redundant. But it is the term used by Moore in his article, and uses it because it is more appropriate than “transistors”. In fact, the components of a chip can be “transistors, resistors, diodes or capacitors” lists Moore on a review of the original article published in 1995 in a journal of the Society of Engineers of Photo-Optical Instrumentation.
What
Moore is not a law of physics or chemistry. It is not based on the observation of natural phenomena, but of a very unnatural, such as the manufacture of integrated circuits.
This is rather a keen observation that this doctor of chemistry extracted his work at Fairchild, and then refined in Intel, the company he co-founded with Noyce in 1968.
Moore has been cautious on many occasions, including in his article in 1965 about how long it would possible to sustain this fabulous increase in the density of circuits. At first he thought it might be maintained for at least 10 years. In practice certain observation time was tested for five times. But Moore always insists that any physical quantity can grow exponentially forever, and this is 100% true. But here is where an interesting twist of his law (which, again, is not a law) arises.
The article of 1965 contains another great prophecy, which is rarely mentioned and whose accuracy stunned the Moore himself. Long before Bill Gates or Steve Jobs predicted that in the future we would have “home computers, automatic controls for automobiles and personal portable communication devices.” “The electronic wristwatch just need a display to be viable today,” he concludes. If the statement was not in a prestigious electronics magazine in 1965 would have sounded to delirium.
When this prediction began to come true, in the mid-70s, the industry was based on the law Moore progress to plan and prepare roadmaps for future releases. So we could be looking at a self-fulfilling prophecy, and now.
But do not think so, and this for two reasons. First, continue to increase the density of the circuits was not options, it was an obligation imposed by competition. Second, on the issue of costs. This easy, the whole race of silicon is reduced to increase the number of components to achieve the minimum cost that greater density would be profitable. Moore predicted what would that number with 50 years in advance.
Two years or 18 months?
The postulate of Moore usually stated as the amount components on a chip will double every 18 months. But were not two years?
Yes, they are two years and were always two years (except, as I said, in the first version of the article). The medley is that several years later, based on the assumption Moore, another Intel executive, David House, determined that computing power, which depends not only the number but also the speed of transistores- would double every 18 months.
Aside from misunderstanding House observation is important because, ultimately, one of the reasons why we put more components on a chip is to increase the computing power of computers , smartphones and everything else, allowing increasingly complex and ambitious applications. Many of the things we do with our smartphones starting with almost all jueguitos- would have been impossible with the first personal computers, whose microprocessors had 29,000 transistors. That was much, for the time, but today’s chips have more than 1000 million components. Yes, no typo: 1 billion on a silicon wafer of about 250 square millimeters. It is the most famous law of electronics in action.
The premise of House suggests that the industry not only used Moore’s Law as a roadmap, but became dependent on it. If the growth of computing power (and other factors) stagnates, not only most of the technical progress it would stop, but economically very significant industry would shrink to a fraction of its current size. Since silicon is multiplying almost all other industries, the impact could be catastrophic.
In a way, civilization has become addicted to Moore’s law, and one of the clouds on the horizon deserved celebration of this is that in 5 or 10 years would no longer be able to continue tightening components on chips. Here, several articles about the imminent collapse of Moore’s Law (in English, and quite técnicos):
*http://techland.time.com/2012/05/01/the-collapse-of-moores-law-physicist-says-its-already-happening/
* http://www.extremetech.com/computing/165331-intels-former-chief-architect-moores-law-will-be-dead-within-a-decade
* http://www.extremetech.com/computing/178529-this-is-what-the-death-of-moores-law-looks-like-euv-paused-indefinitely-450mm-wafers-halted-and-no-path-beyond-14nm
Es inevitably, on the other hand. Although so far has not occurred, there will come a time when the letter of the law hard Moore belied. But is that what counts?
Step into another scene . pivotal character in this story
we already think of something
Another anecdotal data: Moore’s Law is not called for 10 years in 1975. The engineer Carver Mead, colleague and collaborator Moore, coined the phrase, as you can see, himself. This would be enough to mention it here because, to baptize, contributed much to its popularization. But also, Mead was very linked to the creation of the article in 1965. Moore shared graphics with him and consulted about electronic issues. But there is more.
In an interview he gave on Wednesday Spectrum magazine of the Institute of Engineers Electrical and Electronics (IEEE), Mead believes that the fact that we can not continue forever shrinking transistors does not mean that progress in electronics will stop. In his view, since it seems to reach a limit, a discovery that changes everything arises.
For obvious reasons, Moore’s law can not last forever, but, perhaps, when no is no more space to miniaturize, because we will run into the very limits of matter, is a revolutionary way to continue increasing complexity while reducing cost us happen.
In any case, the remark made history was the hallmark of personal computing and remained in force for 5 decades in which everything else in our everyday world changed completely.
Learn more
Those who want to delve into the present and, above all, the future of Moore’s Law and electronics, here an excellent collection of articles from the IEEE (also in English):
http : //spectrum.ieee.org/static/special-report-50-years-of-moores-law
.
No comments:
Post a Comment