At 50 Years Old, The Challenge To Keep Up With Moore's Law
Fifty years ago this week, a chemist in what is now Silicon Valley published a paper that set the groundwork for the digital revolution.
You may never have heard of Moore's law, but it has a lot do with why you will pay about the same price for your next computer, smartphone or tablet, even though it will be faster and have better screen resolution than the last one.
Most of us are used to the cycle of technology. I visited a Best Buy in San Francisco to find shelves stacked with the latest digital gear, and asked a few customers about what they expect from their new devices — whether it's a laptop, a smart phone or tablet.
"Thinner, lighter, faster," says Zeplin Lui.
"I want better resolution," Josie Meng tells me.
Ivo Mijak wants "higher processing speed."
Yet, the same group of customers did not know why their new gadgets would be faster, lighter and have better screen resolution. When I asked them if they'd ever heard of Moore's law, they looked perplexed.
Mijak took a guess. "Diminishing returns ...?"
The Michigan Micro Mote (M3) is the world's smallest computer. A temperature sensor is one of the three types of the M3.
Wrong. It's actually the opposite. What became known as Moore's law was initially laid out in a paper by Gordon Moore. He was one of the founders of the giant computer chip maker, Intel, and of Fairchild Semiconductor, where the first commercial silicon chips were made.
Moore, who comes off as the humble, low-key boy from Pescadero, Calif., that he is, says he had an observation back in 1965 about the way each generation of silicon chips had more transistors than the last.
"It had kind of doubled every year — two, four, eight, 16, 32," says Moore, looking back. "So I said, 'OK it's gonna double every year for 10 years — go from 60 components to 60,000 components on a chip.' "
Moore published his prediction in Electronics Magazine. It was as much about science as it was about the economics of the growing electronics industry.
"I wanted to get across the idea that this is the way we're going to make cheap electronics. The nature of the whole industry is the more stuff you can put on a chip, the cheaper it is per unit of stuff," Moore says.
One way to think about what Moore meant is to refer to the 1989 film Field of Dreams. The main character Ray Kinsella (played by Kevin Costner) hears a voice that tells him, "If you build it, he will come." The voice in the film is referring to building a baseball field that will attract deceased baseball stars.
But, for Moore, the "he" that will come was all the inventors who will find ways to use the growing power of silicon chips and the consumers who will buy the products.
"That's what he was saying," says David C. Brock, who has just released a biography of Gordon Moore. "He was making the case that, if the chip makers invested in the chemical printing and made microchips that embodied cheaper electronics, that those chips would be met with these ballooning markets as chips suffused every area of society."
Over the years, Moore revised his prediction a bit — instead of doubling every year, the number of transistors on a chip doubles about every two years. And it takes more money to keep it going.
Last year, Intel spent $11 billion on research and development and another $10 billion to update the chip factories.
Mark Bohr has been at the company for 37 years and is a leading scientist on the team that keeps the process going. "I'm afraid Moore's law is in a sense a relentless boss," he says. "We have to continually learn how to make transistors smaller to achieve these benefits — improved performance, lower power and lower cost per transistor."
And the market continues to find new uses for the power and efficiency of those chips. At the center of an exhibition on the history of computing at the Computer History Museum in Mountain View, Calif., is a big display explaining and commemorating Moore's law.
"This whole museum is about miniaturization and Moore's law. It really is," says Dag Spicer, the senior curator. Spicer told me to look at a really recent computer and it's actually hard to see. "We're looking at a display of probably the smallest computer in the world right now," he says.
And when he says small — he really means it.
The display shows a thimble with a bunch of little computers in it. Each one is about the size of a grain of rice.
"The vision they're trying to achieve is: What would happen if we had computers that were so cheap and so small that we could just put them anywhere we wanted to gain information?" Spicer explains. "So this could be in a forest or in a lake or in the human body, in a spacecraft."
The point is you keep getting smaller and inventors and entrepreneurs will find uses — and so it has been for the past 50 years. The question is, how much longer can this go on?
"I think we're running out of gas here with Moore's Law," Spicer says. "Because no physical process can continue doubling every year forever — it's physically impossible."
Scientists at Intel think they can keep it going another 10 years. Other experts are less optimistic.
You may be wondering exactly how many transistors are on the latest chips. Chips vary, but let's start with the one inside the latest iPhone. I put that question to some customers at the San Francisco Best Buy.
Matteo Capratta guessed 400. Josie Meng thought it might be 100. Daniel Sayre came closest with a guess of 10,000. Jordon Bhulma on the Best Buy sales staff, says, "They don't pay me enough" to know.
In fact, the chip on your iPhone that's the size of your thumb holds 2 billion transistors. And that isn't the most powerful chip — there are chips with 8 billion transistors.
So, it does make you wonder how much longer this can last. And if it stops — if our computers no longer get faster and cheaper with each upgrade — will the digital revolution continue?
Maybe it will — maybe we will just find new ways to take the technology we have, and innovate.
9(MDE1MTIxMDg0MDE0MDQ3NTY3MzkzMzY1NA001))
Transcript
ROBERT SIEGEL, HOST:
And we mark a milestone this week in All Tech Considered. The 50th anniversary of a prediction that set the tone for the digital revolution.
MELISSA BLOCK, HOST:
Ever wonder how Silicon Valley keeps making that next computer, smartphone, tablet faster than the previous model, and yet smaller and cheaper, too? Well, an article called "Cramming More Components Onto Integrated Circuits" by Gordon Moore explained it. It was published back in 1965, and it proved so authoritative that it became known as Moore's law.
SIEGEL: Whether the explosive pace of technological advance can keep going, well, that's an open question.
NPR's Laura Sydell reports there may be a moment when it all stops.
LAURA SYDELL, BYLINE: We're kind of used to it now - the cycle of technology. When people come to this Best Buy in San Francisco, they have expectations about their new device, whether it's a laptop, a smartphone, a tablet.
What do you expect will be different?
ZEPLIN LUI: Thinner, lighter, faster.
JOSIE MENG: Yeah, I want better resolution, faster computer.
IVO MIJAK: It'll have a higher processing speed, higher screen resolution.
SYDELL: Zeplin Lui, Josie Meng and Ivo Mijak expect the improvements, but they don't really know why it happens.
So have you ever heard of something called Moore's law?
LUI: Ahh, no, Moore's law?
SYDELL: Moore's law.
LUI: Moore's law.
MENG: Moore's law - no.
MIJAK: Ah, yeah.
SYDELL: You know what it is?
MIJAK: Yeah, somewhat, yeah, yeah.
SYDELL: What do you think it is?
MIJAK: Like diminishing returns, you know, this kind of thing.
SYDELL: Wrong. It's actually the opposite. Here is someone who can explain.
GORDON MOORE: OK, I am Gordon Moore. I don't know what else you want to know about me.
SYDELL: What you want to know is that this is the guy who came up with Moore's law. Moore was trained as a chemist, and he still has the demeanor of that small-town kid from Pescadero, Calif., that he is. Moore was one of the founders of Intel and of Fairchild Semiconductor. That's where the first commercial silicon chips were made. Back in 1965 he started to notice something about the way each generation of chips had more transistors.
MOORE: It had kind of doubled every year, you know? Two, four, eight, sixteen, thirty-two. So I said, OK, it's going to double every year for 10 years - go from 60 components to 60,000 components on a chip.
SYDELL: Moore published this prediction in Electronics Magazine. It was as much about science as it was about the economics of the growing electronics industry.
MOORE: I wanted to get across the idea that this is the way we're going to make cheap electronics. Now, the nature of the whole industry is the more stuff you can put on a chip, the cheaper it is per unit of stuff.
SYDELL: One way to think about what Moore is saying is to refer to that old film, "Field Of Dreams."
(SOUNDBITE OF FILM, "FIELD OF DREAMS")
KEVIN COSTNER: (As Ray Kinsella) If you build it, he will come.
DAVID C. BROCK: Yeah, yeah, I think that's what he was saying.
SYDELL: David C. Brock has just released a biography of Gordon Moore. The he that will come is all the inventors who will find ways to use the growing power of silicon chips and the customers who will buy the products.
BROCK: He was making the case that if the chipmakers invested in the chemical printing and made microchips that embody cheaper electronics, that those chips would be met with these ballooning markets as chips suffused every area of society.
SYDELL: Over the years, Moore revised his prediction a bit. Instead of doubling every year, the number of transistors on a chip doubles about every two years, and it takes more money to keep it going. Last year, Intel, the company that Moore co-founded, spent $11 billion on research and development and another $10 billion to update the chip factories.
Mark Bohr, who has been at the company for 37 years, is a leading scientist on the team that keeps the process going.
MARK BOHR: Well, I'm afraid Moore's law is in a sense a relentless boss. You know, we have to continually learn how to make transistors smaller to achieve these benefits - improve performance, lower power and lower cost per transistor.
SYDELL: And the marker continues to find new uses for the power and efficiency of those chips. At the center of an exhibition on the history of computing at the Computer History Museum in Mountain View, Calif., is a big display explaining and commemorating Moore's law.
DAG SPICER: This whole museum is about miniaturization and Moore's law. It really is.
SYDELL: Dag Spicer is the senior curator here. Spicer takes me to look at a really recent computer. It's actually hard to see.
OK, so what is it that we're looking at here?
SPICER: We're looking at a display of probably the smallest computer in the world right now.
SYDELL: And when he says small, he really means it.
SPICER: The display shows a little thimble, for example, with a bunch of these little computers in them. They're each the size of a grain of rice.
SYDELL: Got that? A grain of rice.
SPICER: The vision they're trying to achieve is, what would happen if we had computers that were so cheap and so small that we could just put them anywhere we wanted to gain information. So this could be in a forest or in a lake or in the human body, in a spacecraft.
SYDELL: The point is you keep getting smaller and inventors and entrepreneurs will find uses - and so it has been for the last 50 years. The question is, how much longer can this go on?
SPICER: I think we're running out of gas here with Moore's law because no physical process can continue, you know, doubling every year forever. It's physically impossible.
SYDELL: Scientists at Intel think they can keep it going another ten years. Other experts are less optimistic. You may be wondering exactly how many transistors are on the latest chips. Chips vary, but let's start with the one inside the latest iPhone. I put that question to some folks at the San Francisco Best Buy.
The latest iPhone has a silicon chip which is about the size of your thumb.
MATTEO CAPRATTA: OK.
SYDELL: Do know how many transistors are on that chip?
CAPRATTA: I do not. I do not.
SYDELL: Even a guess?
CAPRATTA: Four-hundred.
SYDELL: Guess?
MENG: Let's see, 100?
DANIEL SAYRE: Transistors? I don't think - I think - isn't transistors a 20th-century concept, and - are they on the smart chips? Ten-thousand - I don't know.
SYDELL: OK, Matteo Capratta, Josie Meng and Daniel Sayre. I put the question to one of the sales people, Jordan Bhulma.
Say, an iPhone - the latest iPhone, how many transistors are on a silicon chip?
JORDAN BHULMA: No, but I don't work in mobile, so...
SYDELL: How about in your computers?
BHULMA: No, they don't pay me enough.
SYDELL: They pay me enough to know. That chip on your iPhone that's the size of your thumb holds 2 billion transistors, and that isn't even the most powerful chip. There are chips with 8 billion transistors. So it does make you wonder, how much longer can this last? And if it stops, if our computers no longer get faster and cheaper with each upgrade, will the digital revolution continue? Maybe it will. Maybe we'll just find new ways to take the technology we have and innovate with that. Laura Sydell, NPR News, San Francisco. Transcript provided by NPR, Copyright NPR.
300x250 Ad
300x250 Ad