Syntiant reveals a neural decision processor at CES 2023. The company, specifically providing deep learning solutions for AI at the edge, has demonstrated its NDP115 Neural Decision Processor.
The company offers special-purpose chips, which can be used in any and all industries, the company stated. These new processors are an addition to its family of processors and use the company’s Syntiant Core 2 inference engine. The Syntiant Core is said to be able to run multiple neural networks at loads below 1mW.
The NDP115 is small with dimensions of 2.1mm x 2.1mm and housing a 25-ball WLBGA package, with a 0.4-ball pitch. The company has said the NDP115 is powerful enough to deliver cloud-free audio and sensor processing that is highly accurate.
These attributes are designed for a large variety of edge products. These include consumer electronics, such as wearables and hearables, smart home applications, and industrial predictive maintenance applications.
“The NDP115 offers the multi-modal functionality of our Core 2 inference engine in a compact, cost- and power-efficient solution for ultra-power and size-constrained applications,” said Kurt Busch, CEO of Syntiant. “Combined with our machine learning software models, the purpose-built NDP115 enables developers to easily deploy full audio and sensor processing solutions that address all kinds of consumer and commercial use cases, from home security to industrial IoT.”
Key features include:
- Support for concurrent neural networks
- multi-sensor fusion
- The Syntiant Core 2 neural processor
- Embedded Arm Cortex M0 Microcontroller
- Support for up to 5 audio streams with the direct input of PCM audio
- Target modes for sensor control and integrations using the I2C controller
- 144KB 64-bit data RAM and 64KB 64-bit instruction RAM, delivered through embedded and programmable HiFi-3 DSP
- Up to 13 GDPIO Pins
- Onboard firmware decryption and authentication
- Flexible clock generation
- 25-ball WLBGA package
The NDP115 can run speech inference at 280 Microwatts, and natively run deep neural networks on many different types of architecture. These architectures include CNNs, RNNs, and fully connected networks. The systems are best utilized for close-talk, far-field, and keyword speech. There is also audio event classification applications and the support of the I2C controller. It is also recommended for Pulse Density Modulation (PDM) for such applications and interfaces as sensor fusion, multi-axis acceleration, magnetic field, tilt, and pressure.
To inquire further head over to the product page here.
Syntiant | syntiant.comSponsor this Article
For the past 8 years, I have been writing about embedded technologies, added to my technical, academic, and medical editorial experience, with companies like Elsevier and Cambridge University Press. I tell people to read what I write, not try to pronounce my last name. I am always available for comments and suggestions you can reach me at firstname.lastname@example.org and I promise I will take the time to reach back out to you. I live in the North East with my wonderful family.