How Intel will keep Moore’s Law running for years to come

Moore’s Law, the observation that the number of transistors on a computer chip doubles every 24 months, has taken a beating as progress miniaturization circuits falter. But chip giant Intel has charted a way to keep the idea alive with a plan to pack 50 times more transistors into processors than is possible today.

Advances in Moore’s Law, named after Intel co-founder Gordon Moore, spread chips from expensive mainframe computers of the 1960s to personal computers of the 1980s and now to smartphones, watches, cars, televisions, washing machines and just about anything that is supplied with electricity. .

Moore’s Law worked by shrinking transistors, the computer elements on a chip. Intel plans to keep reducing them, but also to increase density by stacking the chips in multi-layer packages.

“We firmly believe that the density of transistors is much more important to come”, Intel chief architect Raja Koduri said, in a speech given Monday for the Hot Chips conference for advanced processor disclosures. “The vision will materialize over time – perhaps a decade or more – but it will materialize.”

Koduri’s optimism reflected the enthusiasm of many other companies at Hot Chips, an engineering conference where researchers detail progress. AMD, Nvidia, Google, Microsoft, IBM and a group of startups have shown how they are advancing both general-purpose chips and those dedicated to tasks such as artificial intelligence, graphics and networking.

How Intel plans to advance the chip

Koduri outlined several steps to cram more transistors into a chip than possible with 10nm chips like its Tiger Lake processor hitting laptops this fall. First will come the more traditional approach, reducing the transistors and bringing them closer together. This will triple the density of the transistors, predicts Koduri.

Next come new transistor designs that continue the current transformation of flat circuit element transistors into 3D structures. These steps, called nanowires and stacked nanowires, are expected to quadruple the density.

Next come packaging innovations, with chips stacked in a layered cake of processor elements. This should quadruple the density again. The total calculation increases the density by a factor of approximately 50.

Years of Intel difficulties

Intel’s optimism contrasts with the tough times to keep Moore’s Law in effect.

Intel, once the clear leader in chipmaking, has struggled in recent years. Its switch from a manufacturing process with transistor characteristics measuring 14 nanometers to 10 nm later took five years instead of two. A nanometer is a billionth of a meter, and with 14nm wide circuit elements, Intel can pack about 7,000 the width of a human hair.

Next, Intel delayed its move from 10nm to 7nm manufacturing for six months, and Apple is getting rid of Intel chips from its Macs. To help it adjust, Intel has adopted a more flexible design process that allows it to rely more on other chipmakers like its main rival, Taiwan Semiconductor Manufacturing Corp.

Moore’s Law, but at what cost?

TSMC, which switched to 7nm manufacturing about two years ago and makes Apple’s iPhone chips, said “Moore’s Law is alive and well” last year. But unlike the past, Moore’s Law steps now impose new costs on companies that want to use the most advanced manufacturing processes.

Microsoft’s Xbox One in 2013, Xbox One X in 2017, and Xbox Series X coming this year all have chips that are roughly the same size, which in the past would have meant that chips cost roughly the same size. same price. Now, however, “it’s a lot more expensive for the newer,” Microsoft chip designer Jeff Andrews said.

Another challenge besides cost is that new chips often only speed up specific computer operations. It’s useful for tasks like artificial intelligence and graphics, but it makes life more difficult for software programmers who have to contend with processors that work in different ways.

Intel is trying to bridge this chip divide with a new software layer it calls oneAPI. It’s a notable decision: Intel is a hardware specialist, but it embraces software as a critical step in making its chips useful.

“More and more, hardware architecture teams need to be made up of software experts,” Koduri said.

New chip ideas

At Hot Chips, processor makers have also detailed a host of innovations. Among the biggest:

• The Intel Tiger Lake processor uses a new embodiment of power-saving technology called DVFS, or dynamic voltage and frequency scaling. Different parts of the chip can run faster for high priority tasks or slower to save power. Intel is now juggling priorities between its multiple processor cores, the memory system, and the communications fabric that binds them all together.
• AMD’s competing Ryzen 4000 series chips, named Renoir and now coming to PC, are the first eight-core processing chips for ultra-thin laptops. AMD initially planned for a six-core design, but realized that a neat design could accommodate eight for better performance on tasks like video and photo editing, said architect Sonu Arora. They use half the horsepower for a given level of performance than their predecessors.
• IBM’s Power10 processors, which have 18 billion transistors and are slated to hit massive Unix servers next year, can be bundled into a single powerful server with up to 240 processing cores. In addition, a “pod” of interconnected servers can share up to 2 petabytes of memory. It’s useful for massive business IT challenges like data mining and inventory database management.
• Startup Lightmatter has unveiled its Mars chip to speed up the work of AI such as image recognition. It combines about a billion conventional transistors with tens of thousands of components that use light instead of electricity to transfer data and perform calculations. The idea behind this photonic technology is to reduce energy consumption.