Home Enterprise Xilinx Alveo SmartNIC SN1000 Announced

Xilinx Alveo SmartNIC SN1000 Announced

by Adam Armstrong

Today Xilinx, Inc. announced several products and solutions for the data center. These include the Xilinx Alveo SmartNICs, Smart World AI video analytics applications, an accelerated algorithmic trading reference design for sub-microsecond trading, and the Xilinx App Store. This new hardware and solutions are bringing composable data centers closer to reality.

Today Xilinx, Inc. announced several products and solutions for the data center. These include the Xilinx Alveo SmartNICs, Smart World AI video analytics applications, an accelerated algorithmic trading reference design for sub-microsecond trading, and the Xilinx App Store. This new hardware and solutions are bringing composable data centers closer to reality.

Xilinx Alveo SmartNIC SN1000

Applications and workloads are getting even more demanding. Complex applications such as AI-based or financial trading require low-latency and real-time performance. While this can be done now, it takes some specific and expensive hardware. Xilinx is looking to change this with today’s announced products. Xilinx will be offering composable, software-defined hardware accelerators that will be more adaptive to the changing needs in the data center.

First up is the new Xilinx Alveo SmartNICs, the SN1000. According to the company, the SN1000 is the industry’s first composable SmartNIC offering software-defined hardware acceleration for all function offloads. The Xilinx Alveo SmartNIC SN1000 is based on the company’s 16nm UltraScale+ architecture and powered by the low-latency Xilinx XCU26 FPGA and a 16-core Arm processor. Leveraging an open architecture, the SN1000 offloads CPU-intensive tasks in order to optimize networking performance. Through a combination of the Vitis Networking platform and programming languages, users can compose new network features (and adjust exiting ones) as needed without replacing any hardware. The SmartNIC has dual-QSFP ports for 10/25/100Gb’s connectivity and comes in an FHHL form factor.

Xilinx, with the help of ecosystem partners, is launching an AI video analytics platform, Xilinx Smart World Video Analytics. AI video analytics can be extremely important and beneficial in several areas such as healthcare, retail, and smart cities. The new platform is powered by the Video Machine-learning Streaming Server (VMSS) and uses a single Alveo accelerator card to deliver whole application acceleration and can support multiple neural networks. The company states that this results in the lowest TCO for demanding AI video analytics applications. Existing solutions of the Xilinx Smart World ecosystem 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.

The Vitis development platform now includes an accelerated algorithmic trading (AAT) reference design. This brings low-latency, sub-microsecond trading performance without custom hardware.

Finally, the company announced it is launching an FPGA App Store. Like any other type of app store, this one will have apps ready to be deployed though the focus is on Smart World AI video analytics to anti-money laundering and live video transcoding versus Candy Crush.

Availability

The Xilinx Alveo SmartNIC SN1000 is expected to be generally available in March of this year. There are several Xilinx Smart World ecosystem solutions available today. The AAT reference design is available today. The FPGA App Store is also available today.

Xilinx

Engage with StorageReview

Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | Facebook | RSS Feed