This post is also available in: heעברית (Hebrew)

The first artificial intelligence processor that is a deep neural-network accelerator, delivering both high operational capability and high energy conversion efficiency for a full range of computer applications, was recently developed. The processor is intended for the application of artificial intelligence on different devices.    

The manufacturer is Cadence company, who developed the Tensilica DNA 100 Processor IP.

The processor provides up to 4.7 times improvement in performance and up to 2.3 more performance per watt, in comparison with other MAC system solutions of the same magnitude.

The DNA 200 was designed especially for neural network applications needed in autonomous vehicles, ADAS systems, reconnaissance systems, robotics, drones, VR/AR, smartphones, smart houses and IoT.

“The applications for AI processors are growing in an incredible pace, yet the use of advanced neural networks could widen the currently available output,” says Mike Demler, a senior analyst at Linley group. “in order to provide solutions for the AI capabilities required in products such as small IoT sensors operated on batteries or autonomous vehicles, there is a need for more efficient architecture. The innovative engine in the DNA 100 helps cope with such restrictions and provides high performance.”

The DNA 100 comes with a full AI software platform. The processor enables support for the most advanced AI systems and a variety of neural networks.  

This makes the DNA 100 ideal for vision applications, speaking, radar, laser and communication applications.

The DNA 100 processor has robust software ecosystem support for different network types, including classification, object detection, segmentation, recurrent and regression. The DNA 100 processor also supports the Android Neural Network (ANN) API for on-device AI inference in Android-powered devices.

“Our customers’ neural network inference needs span a wide spectrum, both in the magnitude of AI processing and the types of neural networks, and they need one scalable architecture that’s just as effective in low-end IoT applications as it is in automotive applications demanding 10s or even 100s of TMACs,” said Lazaar Louis, senior director of product management and marketing for Tensilica IP at Cadence, as reported in