Xilinx Inc. Raises the Bar for Automotive Edge Computing with the World's Highest AI Performance-per-Watt Processor
【Summary】On Wednesday, semiconductor developer Xilinx Inc. announced its latest hardware product for automotive, robotics, aerospace applications and more called the Versal AI Edge. Xilinx says the Versal AI Edge series processor is the world’s most scalable and adaptable portfolio for next-generation distributed intelligent systems. It delivers 4 times the AI performance-per-watt versus GPUs.
As automakers around the world equip more vehicles with standard advanced driver assist systems (ADAS) and autonomous driving capabilities, there is a growing demand for AI-powered hardware that's powerful enough to process data from dozens of vehicle sensors in real time and run AI-powered machine learning algorithms for safe navigation.
A company at the forefront of developing this hardware is Silicon Valley-based Xilinx Inc (NASDAQ: XLNX). On Wednesday the company announced its latest hardware product for automotive, robotics and aerospace applications and more called the Versal AI Edge.
The Versal AI Edge series is the newest member of the company's Versal Adaptive Compute Acceleration Platform (ACAP). Xilinx says the Versal AI Edge series is the world's most scalable and adaptable portfolio for next-generation distributed intelligent systems.
ACAP is a fully software-programmable compute platform that combines Scalar Engines, Adaptable Engines, and Intelligent Engines to achieve dramatic performance improvements of up to 20X over today's fastest FPGA implementations and over 100X over today's fastest CPU implementations for automotive ADAS applications, according to Xilinx. For a self-driving car, the adaptable engines are for preprocessing and sensor fusion tasks.
The Versal AI Edge delivers performance up to 479 TOPS (INT4), that far exceeds that of conventional CPUs, GPUs, and Field Programmable Gate Arrays (FPGAs), according to Xilinx. It delivers 4 times the AI performance-per-watt versus GPUs, such as Nvidia's Xavier SoC of and 10X greater compute density versus previous-generation Zynq Ultrascale+ adaptive SoCs from Xilinx.
Xilinx's ACAPs can be modified at both the hardware and software level to adapt to the needs of a wide range of applications and workloads from edge to cloud. The company also includes the Xilinx Vivado design suite for developers.
"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."
Xilinx is globally recognized as the inventor of the FPGA and adaptive SoCs. Like rivals Nvidia and Intel, Xilinx is pushing into the automotive space as the demand for high performance SoCs that can support automated driving functions increases.
Xilinx's FPGA technology gives automotive engineers and software developers the flexibility to design the next generation of vehicle systems, including vehicles with full autonomous driving capability as FPGAs can be customized for specific functions.
The Xilinx AI Engine
Versal AI Edge adaptive compute acceleration platforms (ACAPs) deliver intelligence to a wide range of applications including autonomous vehicles and robotics.
Like its name implies, the latest product in Xilinx's Versal family is designed to perform AI edge processing with very low latency while consuming very little power. It's optimized AI engines are designed for machine learning. The AI Engine in the new Versal processor provides up to five times higher compute density for running vector-based algorithms, according to Xilinx.
The Versal AI Edge series takes the 7 nanometer Xilinx Versal architecture and miniaturizes it for AI computing at very low latency. It can deliver up to 14 TOPS of AI compute (INT4) using just 6 watts of power.
However, the Xilinx Versal AI Edge series processor family scales up to 479 (INT4) TOPS for advanced signal processing workloads for vision, radar and LiDAR sensors using 75 watts of power.
It includes native MIPI support for multiple cameras with up to eight-megapixel resolution, which is critical for camera-based level-2 autonomous driving systems and ADAS.
It also includes additional safety and security measures required for automotive applications. The Versal AI Edge devices support multiple safety standards, including automotive ISO 26262.
In developing the Versal AI Edge, Xilinx said it was responding to industry demand for next-generation machine learning applications that can deliver higher compute density while requiring minimal power. So the company began working on a new innovative hardware and memory architecture, which eventually resulted in the development of the Versal AI Edge.
AI-powered hardware that's capable of edge processing reduces the reliance on data centers for crunching vehicle data. But edge processing also helps to reduce latency by sending data less frequently from the vehicle to the cloud. Reducing latency is critical for improving the safety of self-driving vehicles that are required to make split driving second decisions.
Xilinx's AI Engine was developed with four primary goals: deliver 3 to 8X more compute capacity per silicon area; reduce compute-intensive power consumption by 50%; provide high-performance real-time DSP capabilities; and improve the development environment in order to make its easier for developers to create innovative products.
The Versal AI Edge also integrates new accelerator RAM with an enhanced memory hierarchy for running computationally intensive algorithms. According to Xilinx, these architecture improvements deliver better performance-per-watt versus GPUs, with lower latency for edge processing.
Supports Next-Generation Automotive Safety Systems & Automated Driving
The next generation of vehicles coming out with standard advanced driver assist systems (ADAS) and highway autonomous driving features is resulting in a greater need for automakers to perceive the entire 360-degree environment around the vehicle.
In the auto industry, there is a growing demand for AI-powered edge processing, including for object classification from camera images. Xilinx is addressing these needs through its re-programmable automotive hardware platforms. Developers can scale their device selection to meet specific processing needs.
"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."
Versal AI Edge ACAPs provide a design-entry point for any developer, including Xilinx's Vivado design tools for hardware developers, the Vitis unified software platform for software developers, Vitis AI for data scientists and acceleration libraries for 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.
More information on the portfolio—including product table, white paper, and video can be found at https://www.xilinx.com/versal-ai-edge.
In Oct 2020, semiconductor company AMD announced it was acquiring Xilinx in an all-stock transaction valued at roughly $35 billion. The acquisition will help AMD further expand into the automotive industry, allowing the company to better compete with chipmakers Qualcomm, NVIDIA and Intel.
Originally hailing from New Jersey, Eric is a automotive & technology reporter covering the high-tech industry here in Silicon Valley. He has over 15 years of automotive experience and a bachelors degree in computer science. These skills, combined with technical writing and news reporting, allows him to fully understand and identify new and innovative technologies in the auto industry and beyond. He has worked at Uber on self-driving cars and as a technical writer, helping people to understand and work with technology.
Tesla Agrees to Pay Model S Owners $1.5 Million For Reducing Battery Power in Their Vehicles
Tesla's Chinese Battery Supplier CATL Unveils a Sodium-ion Battery, a Major Breakthrough for EVs
LG Electronics & Magna International Finalize Agreement for New ePowertrain Joint Venture That’s Linked to the Secretive ‘Apple Car’ Project
Tesla is Moving its Retail Showrooms Out of Upscale Shopping Centers, According to Sources
Hyundai’s XCIENT Fuel Cell Trucks Hitting the Road in California in the Largest Deployment of the Technology in the U.S.
Lucid Motors Debuts on the Nasdaq and Raises $4.4 Billion as its Looks to Take on Tesla
General Motors Files Lawsuit Against Rival Ford for Naming its New Autonomous Driving Feature ‘BlueCruise’
Luxury-Electric Automaker Lucid Motors to Launch its Much-Anticipated IPO Next Week
- Ford is Taking its Electric Mach-E SUV on a 6,500-Mile Cross Country Trip to Demonstrate its Capabilities & Technology
- Battery Maker Panasonic Sells its Entire Stake in Tesla for $3.6 Billion
- Mercedes-Benz to Make its Sindelfingen Tech Center its Global ‘Electric Software Hub’, Plans to Hire 3,000 Developers
- Baidu Inc. to Deploy 1,000 ‘Apollo Moon’ Level-4 Robotaxis Over the Next 3 Years in China
- Automaker Stellantis Announces its Bold EV Plans, Will Invest $35 Billion to Electrify its Global Vehicle Lineup
- Volvo’s Concept Recharge is a Blueprint For the Brand’s EV Future
- BMW’s New X5 Plug-in Hybrid Will Ride on Sustainable Tires Made From FSC-Certified Natural Rubber & Rayon
- Ford Expanding Electric Lineup With Explorer, Possibly Bronco
- Lidar Pioneer Velodyne Unveils its Next-Generation Velabit Sensor
- Tesla Agrees to Pay Model S Owners $1.5 Million For Reducing Battery Power in Their Vehicles