Silicon daddy: Moore's Law about to be repealed, but don't blame physics
Insider channels Deep Throat from All the President's Men
Hot Chips Moore's Law, which promises exponentially increasing transistor counts due to chip-manufacturing process shrinkage, is about to hit the wall. As Intel Fellow Mark Bohr once told The Reg, "We just plain ran out of atoms."
But there's one industry veteran, however, who looks at the reason for the repeal of the semiconductor industry's defining law from a different angle, after its nearly 50-year run.
"When Moore's Law ends, it will be economics that stops it, not physics," declared Robert Colwell of DARPA's Microsystems Technology Office, speaking on Monday at the Hot Chips conference at Stanford University.
"Follow the money," he advised.
According to Colwell, who was Intel's chief chip architect from 1990 to 2001 and an Intel Fellow, there's absolutely no doubt that Moore's Law will eventually be repealed. "Let's at least face the fact that [Moore's Law] is an exponential, and there cannot be an exponential that doesn't end," he said. "You can't have it."
Colwell believes that 2020 will be the earliest date for the Law's demise. "That's only seven years away," he reminded his audience. "I'm thinking seven nanometers. You could talk me into 2022 – you might even be able to talk me into five nanometers, I don't know, but you're not going to talk me into one nanometer, you're not going to be able to talk me into femtometers or whatever."
Looking back over the past decades of CMOS development – "CMOS was really good stuff" – Colwell believes that the exponential improvement in semiconductor density was unprecedented in history.
"I don't think there's ever been a technology development like this one," he said. "Ray Kurzweil ... goes down a different path. He says, 'No, no, no, no; Moore's Law is just one of a set of exponentials over history. It's just the latest one. Don't worry about it.' I say baloney. I don't agree with him at all."
As a chip architect at Intel, Colwell says that his job was to "stay out of the way" of Moore's Law, seeing as how the vast amount of historical improvements in semiconductors have been attributable far more to scaling than to progress in microarchitectures.
"If nature's going to give you this bounty of lots more transistors – oh, and they're all faster, and oh, and they're lower power – don't fight it," he said. "Find a way to design a machine so as to leverage that fact."
Harkening back to 1980 when he was at Bell Labs working on a 1MHz 32-bit processor, Colwell reckons that the improvements in semiconductor scaling since then have led to a 3500X improvement in chip performance. But what about design innovations such as pipelining, caches, superscalar architectures? Not so much, he says – a 50X improvement at best.
The next few decades most certainly won't see a repeat of that 3500X improvement due to process shrinkage. After Moore's Law ends – or, as Colwell put it, when "the fundamental Energizer Bunny of the silicon engine stalls out" – improvements in chip performance will be realized not through the further shrinking of the CMOS manufacturing process, but through clever, innovative rethinking of microarchitectures and system design. But those improvements won't be exponential.
It's not that chip designers aren't talented folks, it's more that the challenge of making performance improvements that can match those achieved by process shrinkage is a daunting one. More to the point, making such improvements attractive and affordable enough for customers to plop down cash for them – and making them in such a way that keeps chips sufficiently profitable for manufacturers – will be the gating challenge.
Colwell postulated a future chip designer who accepted the fact that Moore's Law had run its course, but who used a variety of clever architectural innovations to push the envelope. If that designer's chip could provide a 50 per cent improvement in performance or power consumption, Colwell says, it would likely find a market. "But how about 20 per cent? How about 10 per cent? How far down are you willing to go and still think that you've got something you can sell?"
From his point of view, such small improvements wouldn't justify the cost of creating a 10 per cent chip. Customers wouldn't buy it, so chipmakers wouldn't manufacture it.
To those who are of the belief that Moore's Law will be saved by clever chip architects, he says, "I'm happy for your optimism." Not that the progress in chip architectures will hit the wall as soon as Moore's Law does – "We'll play all the little tricks that we didn't get around to" – but the halcyon days of exponential performance improvements will be at an end. And with them, Moore's Law.
Options, profits, and jellybeans
Certainly, there are other routes that chip and system designers may take – 3D stacking, improved packaging, battery breakthroughs, faster memory, higher-bandwidth I/O, better architectures, snappier system interfaces, new materials such as carbon nanotubes, probabilistic or approximate computing, quantum computing – you name it. But none will show the 3500X improvement over time as was accomplished by CMOS and Dennard scaling.
"In the end, if you want to know what's going to happen," Colwell advised, "follow the money. That's not meant cynically, it's meant to remind you of how expensive this business is."
If you work for, say, Intel, and you're asking your company for billions of dollars to make a significant step forward in chip design, your company needs to decide whether to make a bet on you and your design team in the hopes that some multiple of those billions comes back in the form of profits. If your company doesn't see that profit potential, it won't make that investment – and the profit potential in small increments of performance or power improvements just isn't there, Colwell says.
And it's critical to remember, he said, that chip companies don't make the bulk of their profits from the top-of-the-line chips, but instead from the huge numbers of run-of-the-mill follow-on chips that they peddle.
"At least the way Intel did things," he said, "is that we would design a big, fat, hot flagship – we didn't call it that at the time, but that's what it was. Then we would sell a couple of million of those – and those were cool, but they were sort of tolerated. They were tolerated within Intel because they were necessary to get to the parts [Intel management] really wanted to get to," meaning the mass-market chips.
As an example, Colwell said that Intel sold three million Pentium Pro microprocessors for "about a thousand bucks" over about a year and a half. "That's not a terrible business," he said, "but in terms of Intel's business, that ain't so good." What Intel management cared about, he said, was taking the basic design of those high-end chips and translating that microarchitecture down to a smaller, higher-yield process technology, "then do it again, then do it again."
Those iterations would tweak the original design somewhat, but not much. After all, he said, the design teams that worked on those shrinking chips were smaller, had fewer resources, a shorter design cycle, and didn't have the original – expensive – engineers working with them. As a result, the shrunken chips were therefore cheaper, which helped provide higher margins for the iterative designs.
"What they were doing," he said of those secondary design teams, "was cranking out the parts that really made the profits for the place."
That iterative, shrinking process will not be possible after Moore's Law is repealed. "If you start designing a flagship at seven nanometers, you may or may not care about how many of those chips you sell at seven nanometers if you're at a place like Intel and you're designing the flagship-style machine," he said. "But what you do care about is, 'Can I proliferate this to five nanometers and make a hell of a lot of money there?' And if the answer turns out to be 'No,' suddenly my interest in seven nanometers has gone down."
The chip industry, after all, like any other industry, is about profit. Colwell recounted a conversation he once had with former Intel CEO Andy Grove, discussing profitability in the chip biz. "He said, 'Hey, look – if I could make jellybeans and make more of a profit than CPU chips, I'd do it. We're going to make a profit; we answer to the stockholders'."
And as Moore's Law arrives at the inevitable end of its exponential growth, those profits are going to be harder and harder to make. Follow the money. Or the jellybeans. ®