The rise of AI and the Internet of Things will subvert the design of embedded systems



    The rise of AI and the Internet of Things will subvert the design of embedded systems

    Press Release

    Amazon Music Disney Promo

    Different from the general-purpose PC architecture, the definition of embedded system is the IT system designed for specific purposes. In recent years, the development of embedded system in specific fields has accelerated. Compared with the past, both depth and breadth have made great progress. The main reason is not only the advancement of IT technology itself, but also the application industry to continuously expand new functional requirements, especially in application. Under the increasingly competitive market situation, both consumer and non-consumer equipment suppliers must make good use of IT technology and strengthen their own competitiveness, so the market demand for embedded systems is increasing. With the demand and supply sides pulling each other, the development of the embedded industry has reached a peak in history, and it is expected to continue to grow in the next few years driven by AI and the Internet of Things.

    AI is the focus of the IT industry in 2017. Most research institutions and industries believe that AI will not only integrate with embedded systems, but in some applications, embedded devices with AI functions will be connected in series to become an AIoT system. And in the AIOT system, not only the upper-level cloud platform will have computing power, but also the embedded devices of the terminal, even the components in the device, will have a certain degree of AI design, which will form a huge business opportunity. Therefore, all major chip vendors have already Started to put into the layout of the AI chip.

    AI chips(icchipword) are widely used in embedded systems, from data centers, terminal devices (smartphones, tablet computers, wearable devices, etc.), vertical specific industries (manufacturing, transportation, medical, etc.). All of which are the target markets. In terms of architecture, floating-point arithmetic and synchronous parallel computing are very suitable for deep learning neural networks of artificial intelligence. Therefore, GPUs with these characteristics have become an important computing architecture for this wave of AI boom.