Kria K26 SOM: The Ideal Platform For Vision AI At The Edge

An out-of-the-box ready, low-cost development kit for deploying smart vision applications.


With various advancements in artificial intelligence (AI) and machine learning (ML) algorithms, many high-compute applications are now getting deployed on edge devices. So, there is a need for an efficient hardware that can execute complex algorithms efficiently as well as adapt to rapid enhancements in this technology. Xilinx’s Kria K26 SOM is designed to address the requirements of executing ML applications efficiently on edge devices. In this white paper, the performance of various ML models and real-time applications is investigated and compared to the Nvidia Jetson Nano and the Nvidia Jetson TX2. Xilinx’s results show that the K26 SOM offers approximately a 3X performance advantage over the Nvidia Jetson Nano. It also offers greater than a 2X performance/watt advantage over the Nvidia Jetson TX2. The K26 SOM’s low latency and high-performance deep learning processing unit (DPU) provides a 4X or greater advantage over Nano, with a network such as SSD MobileNet-v1, making the Kria SOM an ideal choice for developing ML edge applications.

Click here to read more.

Leave a Reply

(Note: This name will be displayed publicly)