Building the world’s first Optimization Processing Unit to help unlock solutions to some of society’s most profound problems.
From the nanoscale confines of a computer chip to the labyrinthian patchwork of roads, rails, and flight paths, our world is overrun by waste. Not grime and garbage, but rather underutilized and misused time and energy. These trashed seconds and watts may be invisible, but they affect our lives in very real ways.
Think of your commute. Do you go down one road to avoid traffic? Or wait it out, hoping that the line of cars in front of you gets moving? Do you follow directions from an app? Or do you go off script and take a shortcut? Every choice you make, and every choice your fellow commuters make, has an impact on the time you spend on the road and the amount of fuel you burn.
Now think of the same scenario for a logistics company delivering millions of packages worldwide, every day. The challenge to sort packages, find the most efficient delivery route, use the least amount of fuel, and ensure that packages arrive within a prescribed time window, is extraordinary. It is an overwhelming optimization problem — so overwhelming that these companies exist knowing that each day is inherently wasteful.
These types of problems, known as combinatorial optimization problems, exist in industries like telecom, pharma, finance, logistics, those that rely on machine learning, and more. Solving them is notoriously difficult because of the sheer number of possibilities that must be sifted through to determine the most efficient solution. They represent a challenge that digital computers, by their very nature, will never master.
Researchers have long searched for a viable computing approach to combinatorial optimization problems.
That search may now be over. Sync Computing, a startup born out of the MIT Lincoln Laboratory, is pioneering the world’s first Optimization Processing Unit (OPU) that could provide a solution to these stubborn optimization challenges.
Think of it as an “algorithm in hardware form,” says Sync’s co-founder Jeff Chou. Inside the OPU is a system that naturally wants to find a path to optimal energy use. This path, when found, can be interpreted as the solution to the problem.
This differs from a digital computer which must test every single scenario before reaching a reliable conclusion, an approach that is both time and energy intensive. The system Chou and his co-founder Suraj Bramhavar have invented uses simple electronic oscillators to take advantage of nature’s unfailing ability to optimize energy.
Similar approaches to solving combinatorial optimization problems have been attempted with quantum computers. But scaling quantum computations to anything beyond experimental levels is an exceptional challenge. The machines are temperamental and expensive, requiring exotic materials and engineering to keep the computer operating reliably at close to absolute zero. The Sync platform, in contrast, uses inexpensive off-the-shelf electronics components and can operate at room temperature.
The startup’s elegant use of everyday componentry means that it can experiment and scale quickly and with extraordinarily low financial risk. It will help the team create new solutions, faster.
Much of what Sync is pioneering runs counter to decades of problem-solving strategies dominated by a standardized digital architecture. Their computing solutions that power their OPU are not taught in most computer science classes. And that is precisely what makes it so exciting. Bramhavar sums it up well, “We are embarking on a completely alternative form of computing that does not rely on the standard digital architecture. If we can prove it out, it will open up an entirely new way of solving some of our trickiest problems.”