Xilinx has launched a version of its Versal FPGAs for edge AI applications with dedicated AI accelerators.
The Versal AI Edge family built on a 7nm process and boosts edge AI computing performance by 10x compared to the previous Zynq Ultrscale FPGAs by adding dedicated AI accelerators alongside the FPGA fabric.
A new memory architecture with a hierarchy supports new AI frameworks and provides up to 4X AI performance-per-watt of GPUs as well as lower latency for edge applications.
The Versal adaptive compute acceleration platform (ACAP) was launched n 2018 to provide application specific FPGAs with AI for particular markets, starting with data centres and telecoms.
The edge AI parts are aimed at automated driving with the highest levels of functional safety, collaborative robotics, predictive factory and healthcare systems and multi-mission payloads for the aerospace and defence markets supporting vision, radar, LiDAR, and software defined radio. The devices support multiple safety standards across industrial (IEC 61508), avionics (DO-254/178), and automotive (ISO 26262) markets, where vendors can meet ASIL C random hardware integrity and ASIL D systematic integrity levels.
"Edge computing applications require an architecture that can evolve to address new requirements and scenarios with a blend of flexible compute processing within tight thermal and latency constraints,” said Sumit Shah, senior director, Product Management and Marketing at Xilinx. “The Versal AI Edge series delivers these key attributes for a wide range of applications requiring greater intelligence, making it a critical addition to the Versal portfolio with devices that scale from intelligent edge sensors to CPU accelerators.”
“The market opportunity at the edge is growing exponentially and AI chipsets that serve these unique applications are expected to more than double from 2021 to 2025,” said Dan Mandell, senior analyst, IoT and Embedded Technology at VDC Research. “By creating a design for AI-specific tasks that focuses on performance acceleration while remaining scalable and with low power, Xilinx’s Versal AI Edge series is a compelling solution to address these critical markets.”
The architecture matches the algorithm to a range of compute engines in the chip, with Scalar Engines for embedded compute, Adaptable Engines for sensor fusion and hardware adaptability, and Intelligent Engines for AI inference. These AI engines scales to 479 TOPS for 4bit integer operations, although most performance benchmarks target 8bit operations.
Connectivity blocks range from LPDDR-4266, 32Gb/s transceivers to meet all the necessary protocols in edge applications, 40G multirate Ethernet, PCIe Gen4 with CCIX, and native MIPI support for vision sensors up to eight-megapixel resolution.
Development tools for edge AI have been a key area of focus for Xilinx, and developers can use the Vivado design tools for HDL hardware, the Vitis unified software platform for software, Vitis AI for data science and domain-specific operating systems, frameworks, and acceleration libraries for the platform’s target applications.
Versal AI Edge series design documentation and support is available to early access customers, with shipments expected during the first half of 2022, and will include a roadmap for automotive- and defence-grade devices.
Related FPGA articles
- Samsung and Xilinx team up for 5G commercial deployments
- Programmable hardware and software platform with scalable AI
- Customer shipments start for Versal ACAP
Other articles on eeNews Europe