Called as NM (neuromorphic) 500, Nepes’s NPU-based AI chip solution is one of the energy-efficient chip solutions that power the QuickAI platform at the center, helping the platform to be trained in the field to process and recognize image patterns in real time.

It also will work as a hub of a cluster of multiple devices to allocate as many numbers of neurons as necessary to run an AI system and solutions across these devices.

The neuromorphic chip is based on NeuroMem neural network IP of General Vision. Nepes has licensed the IP from General Vision and integrated into the Neuromorphic NM500 AI machine learning device. General Vision is a U.S-based NeuroMem Chip technology and algorithm inventor and licensor

The Neuromorphic NM500 comes in an energy efficient, small form factor.

Developed by U.S.-based fabless FPAG chip maker QuickLogic, the QuickAI platform is a sort of endpoint AI applications platform,  a comprehensive AI application development infrastructure.

That includes low-power AI chip solutions and reference designs, and development tools and kits.

The chip comes built with 576 neurons, each one holding up to 256 bytes of learned or loaded pattern signature. It boasts of unmatched power efficiency, consuming just 120 megawatt, or MW to perform 85 Giga operations per second (GOPS) at a 37 MHz bandwidth.

The speedy and energy-saving performances allow the chip not only to perform real-time, on-the-spot image patterns recognition and self-learning capability, but also simultaneously and individually assign many neurons across a cluster of various devices.

It takes just 8 µs to detect and learn them regardless of network size. Nepes is also offering Knowledge Studio software tool, which a set of customer education program kit.

QucikLogi’s QuickAI HDK will be available in the third quarter of fiscal 2018.

General Vision, QuickLogic, Nepes are now working together to expand their ecosystem of NeuroMem neural network IP and platform across a wide range of industries from industrial IoT smart factories and consumer IoT smart homes and cars.