®

Syntiant  Always-On
Vision, Sensor, and Speech 
Processor

NDP200
SYNTIANT-200-front-back copy.png

NDP200 Neural Decision Processor™

The Syntiant® NDP200™ is a special-purpose processor for deep learning and is ideal for always-on applications in battery-powered devices. The NDP200 applies neural processing to run multiple applications simultaneously with minimal battery power consumption. Built using the Syntiant Core 2™ programmable deep learning architecture, NDP200 is designed to natively run deep neural networks (DNN) on a variety of architectures, such as CNN, RNN, and fully connected networks, and it performs vision processing with highly accurate inference at under 1mW.  NDP200 brings a level of ML performance that delivers 25x the tensor throughput than the Syntiant Core 1™ found in the Syntiant NDP100™ that are currently shipping in high volume. A programmable Tensilica Hifi3 DSP Is also added for feature extraction and signal processing. 

Always-on Vision

Person detection and object classification at ultra-low power

Ease of Use

Flexible methods to access neural processing engine where neural designers have full control of network configurations

Multi-sensor fusion

sens.png
PDM, SPI, I2S and I2C interfaces for microphone and sensor applications 

Syntiant Core 2

Concurrent neural networks to run multiple applications simultaneously with minimal power consumption

Heterogenous Compute

DNN + DSP support both machine learning and traditional image and voice processing algorithms

Applications

The NDP200 enables ultra-low power vision, sensor and speech interfaces in the battery-powered systems and supporting always-on person presence detection and object classification use cases

mobile phones • smart home • security cameras •  video doorbells  • smart displays 

smarthome.png
smart display.png
security camera.png
video doorbell_edited.png

Key Features

+    Syntiant Core 2™  Deep Neural Network

 

+    Neural network supported concurrently: fully-connected, 1D & 2D convolution, Depth wise convolution, Recurrent neural network including LSTM and GRU, average and max pooling

 

+    Up to 896k neural parameters in 8bit mode, 1.6M parameters in 4bit mode, and 7M+ In 1bit mode

 

+    11-wire direct image interface

 

+    Dual PDM digital microphone interface

 

+    I²S serial interface with PCM

 

+    SPI and I²C controller and target for multi-modal sensor fusion

 

+    26 GPIO pins

+    Embedded programmable Tensilica HiFi 3 DSP

 

+    Up to 100MHz internal operating frequency

 

+    Embedded Arm Cortex-M0 for device management with dual timers and UART functionality

 

+    Low power PLL for flexible clock input

 

+    Onboard firmware security and authentication

 

+    Software Development Kit (SDK) integrates in any software environment

 

+    Training Development Kit (TDK) to enable the user of standard frameworks such as TensorFlow for customer-programmed applications

 

+    40-pin QFN package (0.4mm ball pitch)