Penn Engineers have developed the first programmable chip that can train nonlinear neural networks using light—a breakthrough ...

It’s a dual-core, RISC-V chip running at 400MHz. There’s 6 MB of SRAM on the CPU, and there’s 2MB for convolutional neural network acceleration. There is, apparently, WiFi on some versions.

Co-fonder and CTO, Paul Masters, described a novel way to arrange the elements of a chip to do both machine learning "training" -- where the neural network is developed -- as well as "inference ...

Penn Engineers have developed the first programmable chip that can train nonlinear neural networks using light — a breakthrough that could dramatically speed up AI training, reduce energy use and even ...

For now, the company is selling the device as a way to train "biological AI," meaning neural networks that rely on actual neurons. In other words, the neurons can be "taught" via the silicon chip.

Taichi has demonstrated high efficiency and complex AI task handling, using a novel hybrid optical approach to overcome ...

Intel is firing back at Nvidia's growing prominence in artificial intelligence with the launch of its Nervana Neural Network processors ... NNP-T1000 and NNP-I1000 chips for deep learning training ...