Neural Decision Processors™

NDP102 - Always-On Sensor Applications

The Syntiant® NDP architecture is built from the ground up to run deep learning algorithms. Embedded with the Syntiant Core 1™, the NDP102 is a low-power, low-latency device that achieves 100x the efficiency levels of traditional CPUs and DSPs.

The NDP102 runs sensor applications and can operate as a standalone, always-on component of a larger system while consuming less than 100μW of power. The native neural network processing capabilities embedded in the Syntiant Core 1 eliminate the need for intermediate compilers and, works seamlessly with popular machine learning frameworks such as TensorFlow, which greatly reduces time to market and helps assure expected performance.

Ultra-Low-Power

Micro-watt level active power consumption while always-on and listening

System-on-chip

Programmable general purpose IO & SPI Master; boots from a serial flash with the ability to exist as the system's sole processor

Neural Processing

Neural network with over half a million parameters providing large scale parallelism for sensor-based applications

Ease of Use

Multiple flexible methods to access neural processing engine with full customization of applications and post-processing

Key Features

Under 100uW active power consumption in always-on sensor applications
1.71mm x 2.51mm 20-pin eWLB (0.4mm ball pitch)
Syntiant Core 1 neural processor
Hardware feature extraction up to 100Hz frame rate and up to 40 frequency bins per frame
Fully connected neural network, supporting up to 589k parameters
Deeply embedded ARM cortex-M0 processor with 112KB SRAM
Six GPIO pins with programmable direction and drive strength

I2C serial interface for sensor applications
On-chip programmable clock divider
External serial flash boot
Integrated clock multiplier and dividers support low-frequency clock source or external clocking
Onboard firmware security and authentication
Software Development Kit (SDK) integrates in any software environment
Training Development Kit (TDK) to enable the use of standard frameworks such as TensorFlow for customer-programmed applications