The most popular and comprehensive Open Source ECM platform
Let us know more
Photonic Computing: Leapfrogging Electronics to Propel the Next Generation of AI
Is it too good to be true? Startup Boston startup Lightmatter is developing a specialized photonic chip for processing AI algorithms. The company unveiled their Mars SoC chip in August.
The Mars SoC chip is based around an optical tensor core with low latencies of only 100 picoseconds, about 1000 times faster than standard 100ns electronics-based chips. The chip uses standard CMOS manufacturing and needs no specialized fabrication techniques. Unlike some other quantum chips, it runs at room temperature.
The chip is also super energy efficient, requiring only one microwatt of power. Again, this is a 1000 times improvement over standard electronic-based chips.
Lightspeed has focused their efforts on designing the photonic chip for processing AI algorithms. Nick Harris, Lightmatter CEO, said that “a lot of deep learning relies on this specific operation that our chip can accelerate. It’s a special case where a special purpose optical computer can shine. This is the first photonic chip that can do that, accurately and in a scalable way. AI is really in its infancy, and to move forward, new enabling technologies are required. At Lightmatter, we are augmenting electronic computers with photonics to power a fundamentally new kind of computer that is efficient enough to propel the next generation of AI.”
