The smart world AI video analytics solution provides what's described as  deterministic low latency performance for  demanding AI video applications.

Addressing the demands of the modern data center, Xilinx, Inc. (NASDAQ: XLNX) also announced a range of new data center products and solutions, including a new family of Alveo SmartNICs, smart world AI video analytics applications and accelerated algorithmic trading reference design for sub-microsecond trading, and the Xilinx App Store.

Today's most demanding and complex applications, from networking and AI analytics to financial trading, require low-latency and real-time performance. Achieving this level of performance has been limited to expensive and lengthy hardware development. With these new products and solutions, Xilinx eliminates software developers' barriers to quickly create and deploy software-defined, hardware-accelerated applications on Alveo accelerator cards, the firm claimed.

Building the Smart World with New AI Video Analytics Platform

Xilinx is launching an AI video analytics platform with an "ecosystem of partner solutions" built to accelerate the most complex and latency-sensitive AI video inferencing applications. The Xilinx smart world platform, powered by the Xilinx Video Machine-learning Streaming Server, delivers full application acceleration and can support multiple neural networks on a single Alveo accelerator card deterministic sub-100ms pipeline latency. The result is the industry's lowest Total Cost of Ownership (TCO) for demanding AI video analytics applications. So it is claimed.

Video analytics doesn't sound as appealing as it is. Low latency inference is a driver market. For example, in the US, work-related injuries cost $171 billion in 2019. Injuries resulted in 105 million days lost. AI analytics can keep an eye on the workers in an industrial environment and sound an alarm if a non-authorized person comes too close to a dangerous machine.

The second example of the implementation of video analytics involves a global incident and emergency market. The market is expected to grow from 117.2 billion in 2020 to 156.1 billion in 2025. AI analytics platforms can, for example, keep an eye on busy crossroads.

AI analytics can keep an eye on the items in a retail store. Retail loses $100 billion a year worldwide from loss "shrinkage". This fancy word is just an acronym for shoplifting. Gartner estimates that AI retail business value hits an astonishing $78 billion. It sure sounds like that AI investment would pay off very fast.

2020 and 2021 are the years that burden us all with COVID-19. Xilinx video analytics can help here too, it claims.  Global critical care's economic burden is estimated to be $270 billion, and nursing staff represented 30 percent of the total value. Regardless of a massive investment in critical care, the intensive care unit mortality rate is still close to 20 percent. The Analytic AI system by Xilinx can help sound the alarm and alert the medical staff a bit earlier and potentially save lives.

Xilinx has a Total Cost of Ownership (TCO) advantage over its main competitor, Nvidia. The system based on two Xilinx Alveo cards has a 29 percent lower total cost of ownership versus four Nvidia T4 cards. Both systems use 32 cameras with 1080 30 fps resolution, Resnet 50 &TinyYoloV3. The Xilinx system was built using one U30 card and one U50 card, while the Nvidia system uses four T4 cards.

The numbers are impressive as Xilinx has a 29 percent lower total cost of ownership. With 32 streams at 1080p 30 fps, 71 percent lower latency, and 77 percent lower latency to four Nvidia T4 cards and 16 camera streams.

Talking to people behind these products, we could not resist mentioning that product managers did a great job estimating demands in these unusual years.

Xilinx smart world ecosystem solutions available today include:

• Aupera is offering turnkey smart city and smart retail solutions that combine Aupera's intelligent, video processing with Alveo accelerators.
  
  
• Mipsology is providing a toolset for easy migration of existing AI applications from GPU-based architectures to the Alveo platform, as well as plug-and-play, high-performance AI inference acceleration.
  
  
• DeepAI is delivering AI training at the edge on Alveo accelerators with up to a 10x performance per cost advantage compared to GPU-based solutions.