Where to from here?, page-1162

https://www.linleygroup.com/newsletters/newsletter_detail.php?num=6083

BrainChip will offer its spiking-neural-network (SNN) technology in a new processor called Akida, which it optimized for low-power edge AI. Akida is the Greek word for spike, so it’s an appropriate description of the temporally sparse activations the processor uses to enable event-driven computation. The company previously offered its SNN technology as an FPGA accelerator, but at the recent Linley Fall Processor Conference, it announced plans to by the end of the year tape out the chip for TSMC’s 28nm technology and to start sampling in early 2020. It also sells Akida’s core architecture as configurable intellectual property (IP).
Akida is built around a mesh-connected array of 80 neural processor units (NPUs). The interconnect is a proprietary design that supports core-to-core packet transfers as well as intrachip communications among the various function blocks. The cores employ digital logic that mimics the spiking behavior of biological neurons, but the chip also integrates spike converters, which enable it to run popular convolutional neural networks (CNNs) such as MobileNet. The converters can generate spikes from audio, image, lidar, pressure, temperature, and other sensor data, as well as from Internet packets and multivariate time-series data. Although Akida is primarily an inference engine, its native SNN mode allows it to learn new spiking patterns, too.
BrainChip has developed a unique product that’s much more versatile than its low-power competitors. OEMs can use Akida in smart cameras or to power gesture-based and voice UIs in edge devices. Its audio-pattern-detection capabilities are well suited to industrial tasks such as vibration monitoring, and its ability to recognize data patterns associated with common Internet exploits is ideal for protecting IoT devices. Although many researchers are continuing to investigate SNN applications, Akida offers a solution that addresses a wide range of problems.