We have taken it for granted that as the years go by, computer technology becomes faster, cheaper and more energy efficient. However, researchers Mellon Hassan Khan, David Hounshell and Erica Fuchs from Carnegie Mellon University (USA) argue that the future advancement of microprocessors faces new and unprecedented challenges.
In 1965, Fairchild R&D Director (and later Intel co-founder) Gordon Moore predicted in an article the continued systemic decline in cost and increase in performance of integrated circuits. This trend, later coined as ‘Moore’s Law’, persisted for more than 40 years, making possible a “dazzling variety” of new products on the market : from ICBMs to global environmental monitoring systems and from smartphones. even medical implants.
When Moore’s Law turned 50, it began to age rapidly, passing into old age and, finally, to this day, into oblivion. Moore’s Law leaves a thriving global multi-billion dollar electronics and semiconductor industry.
The death of Moore’s Law is not surprising, because the semiconductor industry has been telling conflicting stories for years. While it created new process nodes like clockwork, the capital requirements to develop those new devices increased almost exponentially.
“The current technological and structural challenges facing the industry are unprecedented and undermine the incentives for continued collective action in research and development that has underpinned the last 50 years of transformative global economic growth and social advancement,” the researchers explain in the journal. Nature Electronics.
As the authors explain, progress in semiconductor technology is undergoing change driven by variations in underlying technology and in product end markets. Achieving continuous improvements in performance through transistor miniaturization has become increasingly expensive and the emergence of new end markets has driven innovation in more specialized domains.
The split in technology paths has pushed the industry at a snail’s pace toward Moore’s Law.
“While these innovations will drive many dominant specific advancements, continuing to advance general-purpose computing capabilities at reduced cost with benefits across the economy will likely require entirely new device technology and semiconductor processes,” explains Hassan Khan, co-author of the paper. “The underlying science for this technology is still unknown, and it will require significant research funding, an order of magnitude much higher than what is invested today.”
One thing seems clear: in the short term, performance will go backwards. Computers are likely to have less performance a year from now.
Crushed by the physics of finance and the militarization of optimization, Moore’s Law has hit the wall, bounced, and reversed direction. It is time to go back: all things will get slower, more difficult and more expensive.
In the future, the game changes from “cheaper and faster” to “smarter and wiser”. Software optimizations will take the lead for decades to come, as Moore’s Law fades away altogether.
Reference: Hassan N. Khan et al. Science and research policy at the end of Moore’s law, Nature Electronics (2017). DOI: 10.1038 / s41928-017-0005-9
