Hardware Offers Faster Computation for AI with Much Less Energy

Another dip into analog computing?

 Hardware Offers Faster Computation for AI with Much Less Energy

MIT News, Adam Zewe, July 28, 2022

Researchers at the Massachusetts Institute of Technology (MIT) have developed an analog processor technology that uses inorganic phosphosilicate glass (PSG) to allow for faster computation with greater energy efficiency. Analog machine learning is enabled by increasing and decreasing the electrical conductance of protonic programmable resistors, controlled by the movement of protons into and out of a channel in the resistor. The researchers used PSG to create a programmable protonic resistor that is 1 million times faster than the researchers' previous fastest device; it also can operate at room temperatures using much less energy. Said MIT's Murat Onen, "Once you have an analog processor, you will no longer be training networks everyone else is working on. You will be training networks with unprecedented complexities that no one else can afford to, and therefore vastly outperform them all. In other words, this is not a faster car, this is a spacecraft."