Syntiant Neural Decision Processors (NDPs) represent a family of processors meticulously designed for always-on AI applications, boasting ultra-low power consumption and high-performance capabilities. Anchored by Syntiant's proprietary neural processing technology, these processors excel at executing neural networks on edge devices where computational resources and power are constrained.
NDPs offer distinct advantages over traditional CPUs and GPUs for always-on AI applications. Their ultra-low power consumption, as low as 140 μW, positions them as an ideal fit for battery-powered devices. This is complemented by their ability to achieve up to 100 Trillion Operations Per Second (TOPS), rendering them well-suited for a diverse range of AI tasks.
Their compact form factor enhances versatility by facilitating integration into various devices, while their sub-millisecond latency ensures real-time AI responses. Notably, NDPs are designed for ease of use, making them accessible to developers with varying levels of AI expertise.
NDPs find applicability across an array of always-on AI scenarios. They empower voice recognition in voice assistants and voice-activated devices, facilitate sensor fusion by amalgamating data from multiple sources, enable real-time object detection for various entities, and empower machine learning model training and execution on edge devices.
Syntiant's NDPs are characterized by their proprietary neural processing technology optimized for edge devices with limited power and computational resources. Available in diverse models, each with unique performance characteristics, they are supported by a comprehensive software development kit (SDK) that simplifies AI application programming and deployment. These processors are already integrated into various products such as smart speakers, earbuds, and security cameras.
In the burgeoning landscape of always-on AI applications, Syntiant NDPs stand poised to play a pivotal role. Their amalgamation of ultra-low power consumption, high-performance capabilities, and user-friendliness positions them as a versatile solution catering to an extensive range of devices.
Syntiant’s Deep Learning Computer Vision Models Deployed on Renesas RZ/V2L Microprocessor