The IoT revolution with cloud computing and explosive growth in data volumes has consequences, even if the internet connections are reliable. "Data confidentiality is a top priority, data must be transmitted securely, latency and response must be appropriate to the application, and the cost of data management must not exceed the desired scope - new systems must be easy to integrate into existing ones," explains Chetan Khona, Director Industrial, Vision, Healthcare & Sciences (ISM) at Xilinx. "For many applications, the trend is therefore moving from the cloud back to the edge."

To support this development, Xilinx has optimized its product portfolio for industry and healthcare in terms of adaptability and intelligence. Khona cites the 'Zynq' and 'ZynqUltraScale+' modules, which are optimized for IIoT and HcIoT (Health Care IoT) and form a powerful and scalable hardware basis, as examples. The hardware is complemented by the Industrial and Healthcare Stack, an interdisciplinary offering from Xilinx and its partners for the edge. In addition to the hardware, the stack includes embedded software for safety and security as well as for networking, intelligent control and embedded vision through to artificial intelligence at the edge. "We have developed the special 'Whole Application Acceleration' process for this purpose, which significantly outperforms the AI inference previously used in the industry," explains Khona. In an image processing application, for example, the throughput increases from the usual CPU-based processing of 6 frames/s to up to 60 frames/s if the CPU is supported by Xilinx components.

In addition to its chips, Xilinx has developed the 'Alveo' series of accelerator cards in order to be able to quickly process and store the huge amounts of data that are generated everywhere. They can be used to build domain-specific architectures on which changing algorithms run. In addition, a wide range of pre-programmed applications is already available for them. Xilinx also works with Amazon Web Services, Microsoft, Google, IBM and Alibaba to offer its customers cloud services.

According to Khona, such a comprehensive product range is urgently needed today because electronics development has changed completely. "What is needed are scalable platforms, so-called domain-specific architectures (DSAs), which are tailored to the respective application area with adaptable hardware solutions and can be easily adapted to changed or increased requirements. Khona cites FPGAs, SoCs, MPSoCs, RFSoCs and the new Adaptive Compute Acceleration Platforms (ACAP) as examples of his company's adaptable hardware. The necessary development tools Vivado and PetaLinuxSDX as well as the new Vitis and Vitis AI tools are available.