Unleashing AI Edge Computing Capabilities

  The rapid growth of the scale of the Internet of Things has made centralized data storage and processing models face difficult bottlenecks and pressures. Although cloud computing can meet some needs in the short term, as the amount of data generated by applications and devices is getting larger and larger, the application scenarios are becoming more and more complex, and the issues of bandwidth, latency, security, and connectivity are limited. Increasingly, the advantages of cloud computing are no longer obvious.
  ”This is also an important reason why edge computing exists at the same time as cloud computing in the era of the Internet of Things.” Thomas Lantzsch, senior vice president and general manager of the Internet of Things business unit of Intel Corporation, said in an interview with the author “Big and all,” “edge computing” is “small and beautiful.” Starting from the data source, the application of “real-time and fast” is complementary to the application of “cloud computing”. Comparing the two, cloud computing focuses on big data analysis of non-real-time and long-period data, which can provide basis for business decision support; edge computing focuses on analysis of real-time, short-period data, which can better support real-time intelligent processing of local business With execution.
  With the application innovation and implementation of 5G, artificial intelligence and cloud computing, the development speed of the Internet of Things seems to have exceeded expectations. The data shows that the potential market size of global data will exceed 300 billion US dollars by 2024. By 2025, more than 55% of all data will be created by IoT devices.
  ”From the perspective of silicon and chips, the overall market for edge computing in 2018 is about $ 8 billion, and the growth rate of edge computers is expected to almost double by 2023. In the computer vision market, from 2 billion to 3 billion in 2018 Increased by 25% to 8 billion to 9 billion US dollars. If the software platform and equipment solutions built by the company are added, the overall market opportunity will reach 800 billion to 900 billion US dollars, “said Thomas Lantzsch. It is interconnected. As billions of smart devices are popularized in houses, factories, hospitals, cars, etc., the rise of the Internet of Things technology is bound to be another revolution in this era.
  In less than three years since joining Intel, Thomas Lantzsch has been observing the development of the Chinese Internet of Things for two reasons: the rapid innovation speed of the Chinese market and the great potential of the Chinese Internet of Things market. “In terms of the speed of innovation and the size of the market, China is the ideal soil and market for my IoT career.”
  In order to better understand and deeply cultivate the Chinese market, Thomas Lantzsch made a lot of work adjustments: First, specifically set up China Team. Second, the decision makers of computer vision are transferred to China. Third, move the vertical sector industry that is primarily responsible for customer response and solutions to China. Fourth, from the perspective of time or human dimension, China, India, and Malaysia are divided into the same time zone, which is convenient to meet customer needs.
  ”Although Intel’s current strategy is mainly to provide chip technology, but the IoT solution solution is not enough chips, and the chip and software to provide solutions is our main direction.” Thomas Lantzsch revealed that the future platform There should be certain commonality, which enables Intel to provide a platform for horizontal computing, and to solve the application of various vertical industries such as retail, industry, smart city and vision, transportation, public sector, education, medical, automotive, finance, etc. For the vertical industry, their choice will ultimately fall to the research and development and promotion of technology products.
  On the other hand, with the promotion and application of the Internet of Things in various industries, the market urgently needs a solution to collect, process, and store the complex data generated by these Internet of Things devices. At the same time, the Internet of Things usually manages and controls devices at the edge, and a lot of data needs to make real-time decisions at the edge, which places higher requirements on the intelligence of edge devices. “Intel’s strongest competitiveness is not just AI architecture.” We hope that Intel’s technology architecture can be future-proof, which can protect customers’ software investment and configure high-performance CPUs for it. “.

  Recently, the Intel IoT Division held the “2019 Intel Artificial Intelligence and IoT Eco Partners Summit” and “2019 Intel IoT Summit” in Xiamen, and launched the latest version of the Intel Distribution of OpenVINO Toolkit (2019 R3) (referred to as OpenVINO R3), and the edge artificial intelligence (AI) ecological think tank, actively promote the development and evolution of the IoT ecosystem.
  OpenVINO is a very important free development tool suite from Intel. It contains Intel hardware platform and software tool suite. Through OpenVINO, you can build different CPU architectures, such as Xeon, Smart, etc., in the process of use. Benchmarking for different architectures.
  It is worth mentioning that OpenVINO is four versions a year, and each revision and upgrade will bring new capabilities and bring new performance improvements.
  Just launched the version of OpenVINO in 2018, Intel uses the first generation of Xeon servers, which can do deep learning and manual reasoning through the platform. In 2019, Intel upgraded the version of OpenVINO, but the underlying hardware platform remained unchanged, but the performance increased by 2.1 times.
  ”Intel has a large number of products and technologies to support the expansion of our AI and IT-related technologies.” Thomas Lantzsch said that Intel has adopted an open API, so OpenVINO can also be extended to other architectures. In the future Internet of Things market, Intel is more responsible for the role of providing chips and software solutions, and ultimately the actual ecosystem participants are OEMs, ODMs, and system integrators to help end customers with the market’s real-time conditions.
  In the view of Thomas Lantzsch, OpenVINO has a large user base in the Chinese market. To this end, Intel has launched a development cloud environment for deep learning and artificial intelligence reasoning-integrating all hardware platforms of Intel’s ODM partners, VPU-based hardware, and integrated CPU. And CPU + GPU to achieve platform cloudization and end-to-end security issues. Moreover, each customer uses its own proprietary model during the deployment and testing of the workload. Any third party, even Intel, cannot obtain model information for developing cloud users, which is completely private and secure.
  Over the past year or so, Intel has organized more than a hundred development trainings around the world. A dedicated team develops training content and is open online. It also regularly conducts developer training and has dedicated teams around the world to do OpenVINO training-not only providing content, but also sample code.
Write at the end

  ”Internet of Things deployment or AIoT deployment is a booming market, not just one enterprise can occupy all, whether upstream or software to application, or system integration to network service provision,” said Thomas Lantzsch. “From Intel’s perspective, we Should pay attention to and implement its own strategy. Intel has accumulated 50 years of experience at the chip level. No matter from production to design, we have the most complete set of chip products. At the same time, we are also a very complete software company. The combination, combined with ecological partners, has great room for development. “