Moore's Law may be dead, but there are still ways to improve chip performance

Source: This article is translated from QZ by Ben Conductor Industry Observation

In 1965, when computer technology was still in its infancy, Gordon Moore, a pioneering computer engineer, wrote an article that shocked the technologists of the time. Moore's theory was that computer performance would double every 12 months, while the cost of the technology would drop by 50% at the same time. Forty years later, the so-called Moore's Law remains as strong as ever.

But these are tough times for Moore. Last year, Intel, the computer chip maker he helped found, said its rate of doubling processing power had slowed to 30 months. In May 2016, a headline in MIT Technology Review read, “Moore’s Law is Dead.”

It’s true that the rate of improvement in computer performance is slowing. And that slowdown is a problem: Many of the next-generation products we’ve been promised depend on faster, more powerful, and cheaper chips, and their development supports the assumption that Moore’s Law will continue. If exponential growth slows or stops, then the development of virtual reality, artificial intelligence, self-driving cars, medical treatments, genetic engineering, and even the latest smartphones will be significantly delayed.

But from one perspective, reports of the death of Moore's Law may be greatly exaggerated.

How small can a chip be?

Moore's Law is not dead, but it looks like it will soon be dead. If it is to be revived, engineers and product designers must find new ways to break through.

The “must” is not a suggestion; it’s based on the laws of physics. Computer engineers have increased the performance of chips by shrinking their size, but that strategy is outdated. In chip design, we’ve run into a wall of physics and geometry: As a practical matter, it’s very difficult to make chips smaller and smaller.

Contemporary chip design has reduced the space between various components of a chip to a dozen nanometers. If you are not an engineer, please imagine the thickness of a piece of paper (about 1 mm, which is equivalent to 100,000 nanometers). The space inside the chip is about 1/8000 of the thickness of a piece of paper. Although it is possible to further reduce these dimensions to about 7 nanometers, industry estimates that it will cost $100 million just to develop a 7nm chip prototype, and only three companies in the world can afford to try: TSMC, Samsung, and Intel, which Moore founded. Intel just announced an investment of $9 billion to develop 7nm processors, which will take at least four years.

Once we get to 7 nanometers, we can't squeeze anything out of a smaller space. So, increasing the performance of computing technology will come down to how well we innovate in two other areas, which are thermal management and power density.

Heat and power are killers for designs and devices. They are also killers for innovation. We are basically stuck in the same place because of the size and heat and power limitations.

Step 1: Don’t design a chip that generates heat

To get computing power back up to where it once was, we have to push the boundaries of thermal management. Think about it this way: To make cars faster, we need more powerful engines and better tires. But right now, almost everything we do is make the tires blow out.

Heat issues have hampered some advances in computer engineering, such as stacking, a design approach in which the various parts of a computer system, such as processors, memory and power supplies, are stacked on top of each other. This shortens the distance commands and energy must travel within the machine, saving energy and increasing processing speed.

While stacked components are faster, they generate more heat than if they were separated. Their proximity severely limits engineers’ ability to maintain viable safe temperatures. As a result, chipmakers Qualcomm and Intel have abandoned the idea of ​​stacking. Babak Sabi, director of assembly and test development technology at Intel, told EETimes magazine: “No one is really stacking logic memory, unless someone comes up with a thermal solution… I don’t think anyone will adopt this approach.”

Old cooling technology relies on copper/aluminum tubes and copper/aluminum plates to conduct and dissipate heat. But these tubes and plates are heavy, which makes them inefficient in products like laptops, phones, and cars. They're also stiff and inflexible, which makes them a design nightmare. Try designing a sleek, sexy smartphone around a sheet of copper.

The good news is that thermal technology is advancing quickly, as it has held back progress in overall computer performance. Future thermal solutions may include gels, pastes, and newly engineered flexible fibers instead of heavy, rigid materials. For example, NASA is currently testing a new, lightweight, flexible heat sink material that looks and feels a lot like velvet.

Step 2: Get more out of your power investment

If thermal issues constrained Moore's Law, then power density issues completely paralyzed it.

Power density is the amount of power that can be drawn from a given space. If the battery is the same, greater power density can deliver more power, thus extending range. Going back to the race car analogy, if computer processing is the engine and thermal management is the tires, then power density is the fuel.

Our computers and other electronics are getting faster and more powerful, requiring more and more power to be packed into smaller and smaller spaces — but our battery technology is only slowly advancing. As the Samsung Galaxy Note 7S can tell you, even the slightest mistake in balancing greater power demands with tighter design specifications can be catastrophic.

The energy density issue is a huge stop sign for the next generation of mobile computing products such as robots, drones, space exploration equipment, and electric vehicles. For these fields, power density is everything. For more casual consumers, the lack of power density improvements is why you feel your phone battery draining quickly.

To complicate matters further, energy density and thermal management are related issues. Storage, battery charging, and traction power all generate heat. So every time engineers push one boundary, factors on the other side become more complex.

The future of chip technology

But it’s not all doom and gloom. I’m very optimistic that scientists and engineers will soon be able to keep up with Moore’s Law beyond thermal management and power density. One reason I’m confident is that the incentive to overcome these technical, engineering, and design challenges is driven by consumers. Consumers want longer battery life and laptops that don’t run too hot; they prioritize thinner, lighter products over more processing power. So when it comes to a risk/reward business decision, solving the heat and power issues can reap significant financial rewards.

Another reason for optimism is that the slowdown in innovation has created slack in the skills chain, meaning that every step forward we make in thermal or energy technology is likely to unlock corresponding advances in other areas.

When that happens, the emergence of new products and technologies will be fast and furious, which will simultaneously restore and destroy Moore's Law. Moore predicted that technological progress might not be linear, but in the end it could be more exciting.

[About reprinting]: Reprinting is limited to full text reprinting and the title and content of the article must be completely retained. No deletion or addition of content is allowed to circumvent the original protection, and the beginning of the article must indicate: Reprinted from the WeChat public account "Semiconductor Industry Observation icbank". Thank you for your cooperation!

【About Call for Papers】: Semiconductor elites are welcome to contribute (including translation and editing). Once accepted, they will be published with your name, and you will be rewarded with a red envelope! Signing a contract to become a column expert will earn you a thousand yuan in manuscript fees! Please indicate "Submission" in the title of your manuscript email, and indicate your name, phone number, unit and position in the manuscript. Welcome to add my personal WeChat account MooreRen001 or send an email to jyzhang@moore.ren

Click to read the original text and join the Moore Elite