Monday, April 20, 2015

How Moore's Law Changed History (and Your Smartphone) (PCMagazine)

After 50 years, Intel co-founder Gordon Moore's prediction about semiconductor miniaturization still holds up.

Gordon Moore/Credit: Intel

When I was a kid, I got an idea from a puzzle book which I believed in my childlike way would make me fabulously wealthy in just a few weeks. The trick was to convince my mother to pay me a progressively doubling wage for doing household chores for a month—a penny on the first day, two cents on the second, four on the third, and so on.
Unfortunately for me, my mom was too bright to take me up on the offer. So I missed out on collecting $10.7 million and change after 30 days of doing the dishes, taking out the trash, and the like. Fortunately for all of us, the semiconductor industry did accept Intel co-founder Gordon E. Moore's challenge to do something similar 50 years ago—and we continue to reap the benefits today.
Back in 1965, Moore, then the director of R&D at Fairchild Semiconductor, was asked by Electronics magazine to submit an article making a prediction for developments in the semiconductor component industry over the following decade. So for the 35th anniversary issue of Electronics published on April 19, 1965, Moore noted that the number of transistor and resistor elements on computer chips had been doubling roughly every year—and that he expected this to keep happening for the next 10 years.
It was a prediction that ended up extending far past its first decade. And what later became known as "Moore's Law" would prove to be perhaps the most reliable and enduring guide to the pace of technological advance in not just the semiconductor business, but in the computing industry as a whole.
But back in April 1965, Moore's observation, accompanied with a simple graph he'd sketched, wasn't even regarded as cover story material in Electronics. Instead, you had to thumb forward to page 99 to find the Intel co-founder's prophetic pronouncement amidst the writings of other industry experts. For just 75 cents, you'd have had the very first iteration of Moore's Law in your grubby hands, but finding it wasn't easy.
Electronics Magazine Cover/Credit: Intel
In an interview earlier this year, Moore, who co-founded Intel in 1968 with the late Robert Noyce, explained how the whole thing came about after Electronics asked him for the article.

"I took the opportunity to look at what had happened up to that time. This would have been in 1964, I guess. And I looked at the few chips we had made and noticed we went from a single transistor on a chip to a chip with about eight elements—transistors and resistors—on it," said the 86-year-old chairman emeritus of Intel.
"The new chips coming out had about twice the number of elements, about 16. And in the laboratory, we were creating chips with about 30 elements and we were looking at the possibility of making chips with twice that many, around 60 elements on a chip. Well, I plotted these on a piece of semi-log paper starting with the planar transistor in 1959 and noticed that, essentially, we were doubling every year.
"So I took a wild extrapolation and said we're going to continue doubling every year and go from about 60 elements at the time to 60,000 in 10 years."
To make things even more interesting, Moore himself did not at the time regard this remarkable prediction of exponential growth in integrated circuit (IC) complexity to be the most important point of his article. Instead, he was trying to stress that making ICs was going to get cheaper to do over time—potentially, a whole lot cheaper.
"I was just trying to communicate the point that this was the direction semiconductors were going," he said. "And this was going to give a tremendous cost advantage, which wasn't true at the time. The early integrated circuits cost quite a bit more than the pieces to assemble the similar circuits out of individual components.
"But one could see the trend was going in the direction that this was going to be the cheaper way eventually. That was my real objective—to communicate that we have a technology that's going to make electronics cheap. But I didn't expect this binary order of magnitude increase, the thousand-fold increase in complexity to be very accurate."
The accelerated pace of early IC manufacturing slowed a bit in ensuing years. Moore later revised his timetable for the doubling of transistor density in microchips from one year to two. More recently, we've seen the cadence at which Intel and other leading semiconductor manufacturers ramp new process technologies range from about 18 to 32 months.
Making Moore's Law Happen

Of course, the continued relevance of Moore's Law hasn't just happened by magic. Moore's prediction, while remaining accurate for 50 years, isn't actually a natural "law" like the Second Law of Thermodynamics. Shrinking the size of elements on computer chips so consistently and for so many decades has required incredible innovations like CMOS, silicon straining, VLSI, immersion lithography, high-k dielectrics, and most recently, FinFET or tri-gate "3D" transistor process technology. Those advances didn't just appear out of thin air because Moore's Law demanded they happen—brilliant, doggedly persistent people working in labs at universities and companies like Bell, Shockley Semiconductor, Fairchild, Intel, Toshiba, IBM, Advanced Micro Devices, TSMC, Samsung, and elsewhere invented them and thushelped extend Moore's Law, not the other way around.


Moore made his original prediction about the pace of miniaturization in computer chip components as a way to highlight the attractive economics in semiconductor manufacturing. By his own admission, he wasn't terribly confident that it would hold up over time. But in the ensuing decades, Moore's Law has become as much a challenge to the industry to keep it alive as an axiom for how chipmaking works.
I never managed to trick my mom into paying me a fortune to make my bed for a month. But 50 years ago, Moore set the wheels in motion for an entire industry to dedicate itself to producing increasingly complex integrated circuits to make progressively more powerful computers and devices. As a child, I thought it would be really cool if I could use math to turn a penny into millions. Way cooler—and far more beneficial to all of us—is that we've turned transistors that were once the size of a human hand into nanoscale electronic switches powering smartphone supercomputers we can fit in our pockets.
Moore's Law has presided over that astonishing process of human ingenuity for the past 50 years. It's still holding up even as we approach atomic-scale limitations in making circuitry even smaller than it is today—a possible end of the road for Moore's enduring insight, or perhaps just the next great challenge.
When a prediction as elegant and impactful as Moore's Law holds up for 50 years, people tend to notice. PCMag asked several notable industry analysts to discuss the importance of Moore's famous observation and its legacy for the computer industry and the world.
Charles King, principal analyst, Pund-IT:

"The 50th anniversary of Moore's Law highlights an interesting issue—that the tech industry loves to talk about the importance of innovation but continuing relevance is a much rarer commodity. Originally, Gordon Moore was making a simple observation about the brevity of time required for significant technological advancement. But for half a century now, semiconductors, microprocessors, and related computing components have continued to evolve at the torrid pace Moore observed.

"The nature of physical materials means that Moore's Law will eventually hit a literal stopping point—you can't scale semiconductors smaller than a single atom. But I believe new innovations will arise that will allow digital technology to advance and improve at the 18-24 month pace Gordon Moore originally observed. As a result, I believe Moore's Law will remain relevant for years or even decades more to come."
Jack Gold, principal analyst, J.Gold Associates:

"Fundamentally, Moore's Law (and its effect/consequences) has probably done more to fundamentally change how all of us live, work, and learn than anything else in history. Fifty years ago, who would have thought that we would be carrying supercomputers in our pockets? Who would have foreseen the wireless and mobile revolution? Who would have thought that companies like Google could instantly parse the entire Internet and give us information ... and who would have envisioned tens of billions of users on the Internet to begin with?

"Moore's Law is not dead yet and it's unlikely to die for at least a couple more generations of chips. But I suggest it won't really die for many more years, as scientists and engineers continually find new ways to extend it. I am confident they will find new ways (e.g., 3D chips, organics) to continue the fundamental premise of the law, even if it's not quite what Moore had in mind."
Patrick Moorhead, principal analyst, Moor Insights & Strategy:

"Fifty years after Gordon Moore penned his observation now known as Moore's Law, why does it still matter? It matters because it is still the rallying cry in the tech industry to double transistors densities every two years. This then leads to higher performance and lower power technology.

"While Moore's Law is fairly well-known and (mostly) understood within the tech industry, thinking beyond the tech industry, why should people care? What has Moore's Law done for the economy, and for society, for our daily lives? Moore's Law has impacted almost the entire fabric our lives. It has enabled most of the comforts many of us take for granted, like quality of our health care and medications, safety of our cars, the ability for us to get almost everything we want from our very fingertips.
"In the past decade alone, we've seen countless firsts: The first YouTube video, the founding of Twitter, the introduction of the original iPhone and the iPad, the first Netflix movie streamed and much more. Where do you think we'd be without Moore's Law driving the evolution of technology?
"The reality is that without Moore's Law, we wouldn't have the technology that we love today ... smartphones, tablets, PCs, game consoles, 4K displays. We wouldn't have content like Twitter, Netflix, and YouTube.
"Assuming Moore's Law continues to drive the industry for the next 10-15 years, what might be possible? I think we will see some amazing breakthroughs. Self-driving cars will be commonplace. We will all have some kind of robot helping us in our homes and hospitals. We could finally cure cancer. There's a chance we can extend out lifespans considerably.
"Moore's Law isn't the beginning or the end of innovation. It takes breakthrough architectures, collaboration, software, and other advancements to make all these innovations come to life. But without the industry driving to Moore's law, we wouldn't even have had a chance at accomplishing all the wonderful things we have today and building even better technology in the future."
To celebrate the 50th anniversary of Moore's Law, Intel has put together a timeline and some facts and figures illustrating just how far we've come in the past several decades—and how mind-bogglingly small chip circuitry is today compared to the transistors Moore and his colleagues were working with in 1965.
Just to get things started, the chip giant noted that Moore's original purpose of highlighting just how inexpensive making semiconductor components was about to become has proven prophetic beyond anyone's wildest dreams.
"Around the time that Gordon Moore made his now-famous observation, Fairchild Semiconductor, where he worked at the time, sold transistors for $150 each. Today the price-per-transistor of Intel's Core i5 processor is $0.00000014. Put another way, you could buy 70,000 Core i5 transistors for a penny," an Intel spokesman said.
Here are some more fun facts and figures illustrating the impact of Moore's Law, courtesy of Intel:
  • CARS: If automobile fuel efficiency improved at the pace of Moore's Law, a person could easily drive a car for his entire life on a single tank of gas. With the pace at which transistors have shrunk, your car would be the size of an ant.
  • CONSTRUCTION: If a skyscraper fell in price at the pace of Moore's Law, a person could purchase one for less than the cost of a PC today. And if a skyscraper increased in height at the pace of Moore's Law, it would be 35 times the height of Mt. Everest.
  • REAL ESTATE: If house prices fell at the same rate as transistors, a person could purchase a home for the price of a piece of candy.
  • SPACE EXPLORATION: The Apollo space program to land humans on the Moon cost $25 billion. If prices fell at the pace of Moore's Law, today that program would cost about as much as a small private plane. The trip to the Moon in 1969 took three days. If Moore's Law applied to space travel, that trip would now take one minute.
  • AIR TRAVEL: A flight from New Zealand to New York would be over in the time it takes to fasten your seatbelt, if Moore's Law were operative.
Moore's Law-driven innovation has resulted in astonishing leaps in the power of Intel's current processors compared with its own technology of several decades ago, the chip giant notes:
  • If an Intel-based Android phone were built using 1971 technology, the phone's microprocessor alone would be the size of a parking space. Try taking a selfie with that.
  • Compared to Intel's first microprocessor, the Intel 4004, today's 14nm processors deliver 3,500 times the performance, at 90,000 times the efficiency, and at 1/ 60,000th the cost.
  • The first semiconductor transistors were the size of an eraser at the end of a pencil. As a result of Moore's Law, more than six million of today's tri-gate transistors could fit in the period at the end of this sentence.
  • Today's transistors are invisible to the naked eye. To see a single transistor, you'd have to enlarge a typical chip to the size of a house.
Amazing stuff and here's how it all went down at Intel in the years following the first formulation of Moore's Law:
Moore

No comments:

Post a Comment