Syntiant 推出适用于始终在线视觉边缘 AI 应用 的NDP200 神经决策处理器

Deep Learning Device to Perform Person Detection and Object Classification at Ultra-Low Power

IRVINE, Calif., September 22, 2021 – Syntiant Corp., a provider of deep learning solutions making edge AI a reality for always-on applications in battery-powered devices, today introduced its Syntiant® NDP200 Neural Decision Processor™ (NDP), the company’s first chip designed for vision processing that can provide highly accurate inference at under 1mW.

Packaged with the Syntiant Core 2™, the NDP200 can run multiple applications simultaneously with minimal power consumption, including AI vision features such as person detection, object classification, motion tracking and occupancy monitoring. With its multi-modal capabilities, the NDP200 also can perform highly accurate processing on-device, including multi-sensor fusion, voice command recognition, acoustic event detection, infrared detection, anomaly and tamper detection, as well as other audio, motion and pressure sensing applications.

“Our NDP200 brings extraordinary inference capabilities to a wide variety of always-on image and sensor applications,” said Kurt Busch, CEO of Syntiant. “Our Core 2 architecture minimizes data movement by coupling memory and MAC functions that generally provides 100x improvement in efficiency and performance. As an example, an application that lasts three and a half days on a battery-powered device using an Arm A53 processor would last one year running on the NDP200.”

Ideal when presence and motion detection are required in ultra-low-power applications, the NDP200 brings always-on neural processing to all types of consumer and industrial battery-powered products, including IP and security cameras, mobile phones, tablets, smart speakers, smart displays, as well as smart home applications and security devices.

“The combination of Syntiant’s NDP200 and PixArt's ultra-low-power image sensor offers the ability to develop highly accurate person and object detection systems that consume very little power with almost no latency,” said Sen Huang, CEO of PixArt Imaging, a leading provider of smart sensor SoC solutions. “These features are extremely critical to the growing need for imaging sensor applications found in battery-operated IoT devices, such as video doorbells and security cameras.”

Syntiant Core 2 is a highly flexible, ultra-low-power deep learning inference engine that moves larger neural networks into always-on domains, with capacity to generate shared embeddings, run ensembles and other neural architectures concurrently or in cascades. Designed to avoid inefficiencies in stored program architectures, the Core 2 provides neural designers with full control of multiple concurrent independent networks, while also enabling machine learning deployment without any intermediary compilers.

“We worked with Syntiant to deploy high performing, highly efficient machine learning image models running on the NDP200,” said Mil Danailov, ML engineer at Connected, a leading software product development firm. “We also were very impressed with the ease-of-use of the company’s training development kit, which includes tools for running bit-exact simulations directly within high-level modeling languages. The NDP200 brings wide functionality, low power and high accuracy to any image or sensor edge AI application.”

The NDP200 is equipped with an Arm Cortex M0 processor and a HiFi-3 DSP to support feature extraction and signal processing for image and voice enhancements.

Key product features include:

• Syntiant Core 2™ neural network
• Hardware acceleration over 6.4GOP/s
• Supports convolution neural networks, recurrent neural networks, fully connected networks, and with ability to process multiple heterogenous networks concurrently
• Supports more than 7 million parameters
• Embedded user programmable HiFi 3 DSP
• Deeply embedded Arm Cortex-M0 Processor with dual-timers, and UART functionality
• 11-wire direct image interface
• SPI and I2C for multi-modal sensor fusion
• Dual PDM microphone interface
• I2S serial interface with PCM
• Stand-alone operation
• Onboard firmware decryption and authentication
• Flexible clock generation
• 40-pin QFN package (0.4mm pitch)

About Syntiant

Founded in 2017 and headquartered in Irvine, Calif., Syntiant Corp. is a leader in delivering end-to-end deep learning solutions for always-on applications in battery-powered devices by combining purpose-built silicon with an edge-optimized data platform and training pipeline. Syntiant’s advanced chip solutions merge deep learning with semiconductor design to produce ultra-low-power, high performance, deep neural network processors for edge AI applications across a wide range of consumer and industrial use cases, from earbuds to automobiles. The company is backed by several of the world’s leading strategic and financial investors including Intel Capital, Microsoft’s M12, Applied Ventures, Robert Bosch Venture Capital, the Amazon Alexa Fund and Atlantic Bridge Capital. Syntiant continues to be recognized as an industry leader, being named to Gartner's April 2020 Cool Vendors in AI Semiconductors; Fast Company’s prestigious list of the World’s Most Innovative Companies for 2020; as well as a CES® 2021 Innovation Awards Honoree and CES® 2020 Best of Innovation Awards Honoree. More information on the company can be found by visiting www.syntiant.com