Syntiant Launches its Latest Neural Decision Processor NDP115 at the CES 2023 Event

Syntiant Corp. unveiled NDP115 which is the newest addition to its neural decision processor family. NDP115 is a compact yet powerful processor with highly accurate real-time decision-making and sensor AI capabilities. The processor targets applications like self-driving cars, smart home devices, and wearable electronics. NDP115 is based on the company’s Syntiant Core 2 that can run multiple neural network loads at under 1mW power consumption.

Technical Specifications

• The Syntiant Core 2 enables multi-mode functionality in a tiny, cost and power- efficient form factor. It includes many processing elements, such as artificial neurons, that are used to process input data and make decisions.
• In addition to the Core 2 neural processor, the NDP115 includes an embedded Arm Cortex-M0 microcontroller. ARM Cortex M0 is a 32-bit microcontroller and is based on the ARMv6-M architecture.
• The NDP115 includes an embedded user-programmable HiFi-3 DSP with 144KB 64-bit data RAM and 64KB instruction RAM. This processor can handle up to 13 GPIO pins.
• The NDP115 also includes support for I2C and PDM interfaces, which can be utilized for various purposes such as sensor fusion, multi-axis acceleration measurement, tilt detection, magnetic field sensing, and pressure sensing.

Artificial neurons are mathematical models that are designed to simulate the behavior of biological neurons. They are the basic building blocks of artificial neural networks (ANNs). ANNs are a subset of machine learning (ML) models. Various ML applications, such as image recognition, natural language processing, and prediction, widely use artificial neurons. The Cortex-M0 supports a wide range of on-chip peripherals, including timers, serial interfaces, and ADCs, making it suitable for a wide range of applications. Programmers can use C or assembly language to program the microcontroller. A wide range of development tools and software libraries support the microcontroller. The Cortex-M0 also supports low-power modes, making it well-suited for battery-powered applications.

The NDP115 is based on the same architecture as the NDP120. The below diagram is a representation of NDP120’s architecture.

Applications of the NDP115

The company is promoting the NDP115 as a well-suited microcontroller for running deep neural networks (DNNs) on devices. This means that it has the capability to process DNNs locally, without the need to send data to the cloud for processing. The use of microcontrollers is particularly well-suited for “hearables” (i.e. wireless earbuds), wearables, smart home devices, and industrial internet of things (IIoT) devices.

The NDP115 retails at $3.25 per chip in 10,00 unit tray quantities. For more information, you can visit the product site.

Please follow and like us: