More and more of Inndisk’s customers are developing products for AI and IoT. For example, one company evaluated an AI solution with software development tools and product lines for different computing power requirements. However, the solution’s high power consumption was not suitable for a fanless environment. The system was inflexible and not durable enough for use in the industrial market.

Innodisk’s optimised solution offers the advantages of low power consumption, high flexibility and long-term availability. The Innodisk Machine-Vision Solution Kit is designed for AI and video analytics in edge applications and is based on AMD’s Kria SOM module. Featuring adaptive SoCs with integrated FPGA fabric, the Kria SOMs offer high performance with low latency in a small package. The K26 SOM-based module design reduces hardware development time and simplifies inventory management. In addition, the powerful Vivado development tool helps software developers work with popular AI frameworks in a short time.

AI technology is constantly changing. The architecture of a deep neural learning network is constantly evolving, and specifications such as the time-sensitive networking (TSN) protocol need to be adapted. AMD’s FPGA product is therefore an ideal solution due to its high flexibility. This is an important feature in vertical markets, especially in factory automation. AMD offers development tools such as Vivado™, Vitis™ and Vitis-AI with extensive technical documentation. This helps to understand FPGA products and accelerate time-to-market. Due to the small and diverse product characteristics of industrial markets, FPGAs are the most suitable AI solution for industrial AI applications as they offer the three advantages of low power consumption, high flexibility and long-term availability.