AMD Versal AI Core and AI Edge Series

Both the AMD Versal AI Core and Edge series hold promise for new space applications, but cater to slightly different needs. The Versal AI Core Series offers the highest compute density within the Versal family. Its strength lies in processing massive datasets and complex algorithms, making it ideal for spaceborne applications requiring high-performance AI, like on-board image processing for faster scientific data analysis from telescopes or rovers and real-time anomaly detection in spacecraft systems for enhanced reliability.

The Versal AI Edge Series shines in compact size and power efficiency, crucial for resource-constrained space environments. Its AI capabilities and adaptable fabric enable edge-based processing on satellites for tasks like target recognition or object classification, reducing data to be transmitted back to Earth and autonomous navigation and decision-making for smaller satellites or drones exploring space.

While lacking dedicated AI engines, the Edge series can still handle AI workloads efficiently, making it a strong contender for space applications demanding both performance and frugality.

Contact Our Experts 

AMD Logo

AI Core Series

Geared towards raw processing power, this series prioritizes AI Engines for maximum AI inference performance. Ideal for data centers and cloud computing, it excels at handling complex AI workloads much faster than traditional CPUs. This makes it perfect for scenarios where maximizing AI inference speed is the top priority, like cloud-based AI applications.

Other applications

5G Radio & Beamforming

AMD's Versal AI Core series tackles the demanding computational needs of 5G radio and beamforming with its built-in AI engines and powerful processing. These engines handle the complex signal processing required for beamforming in real-time, while also offering the flexibility to adapt to evolving beamforming algorithms. This allows 5G networks to achieve higher capacity, lower latency, and better efficiency by precisely directing radio signals towards specific devices.

5G mast in a cityscape at night

Video Processing for Smart Cities (MM)

Display portlet menu

Video Processing for Smart Cities

AMD's Versal AI Core series acts as the muscle behind smart city video processing applications. It packs a punch with specialized AI engines for real-time tasks like license plate recognition and facial detection, along with video decoders for handling different video formats. The series also boasts adaptable engines for video scaling, compression, and customization, allowing cities to efficiently analyze video feeds and extract valuable data for tasks like traffic management and public safety.

Smart City

Radar Processing

The AMD Versal AI Core series empowers radar processing for various applications by offering a powerful combination. Its built-in AI Engines accelerate complex tasks like filtering and target recognition, while the adaptable fabric allows for real-time customization of radar signal processing algorithms. This translates to improved radar performance, including faster detection, higher accuracy, and the ability to handle sophisticated techniques like Synthetic Aperture Radar (SAR) for high-resolution imaging.

Radar dish

AI Edge Series

This series packs a punch in a compact package. It combines various processing elements like CPUs, real-time processors, and specialized AI Engines. This versatility makes it perfect for intelligent systems at the network edge where efficiency and handling the entire data pipeline (sensor input to real-time control) are crucial. Applications include self-driving cars, smart factories, and medical devices. Safety features like ISO 26262 compliance ensure reliable operation.

Other applications

ADAS and Automated Driving

The AMD Versal AI Edge Series tackles the demanding needs of Advanced Driver-Assistance Systems (ADAS) and Automated Driving (AD) with a powerful combination of hardware and features. It integrates various processors for sensor fusion, real-time data manipulation, and AI tasks, all while delivering exceptional performance with low power consumption. This architecture prioritizes safety with adherence to ISO 26262 standards and offers flexibility to adapt to different sensor configurations. By handling the entire data flow from raw sensor input to real-time vehicle control, the Versal AI Edge Series simplifies system design and paves the way for the future of autonomous driving.

ADAS Aerial Shot

Collaborative Robotics

AMD's Versal AI Edge series processors are designed to be a one-stop shop for collaborative robots by offering features like combining sensor data, precise control of multiple axes, and built-in safety features. This allows robots to perceive their environment, move accurately, and operate safely alongside people, all while using artificial intelligence to improve their performance and efficiency.

Robotics in Industry 4.0

Unmanned Aerial Vehicles 

AMD's Versal AI Edge series is a boon for unmanned aerial vehicles (UAVs) by cramming powerful processing, sensor fusion, and AI capabilities into a single, compact chip. This lets UAVs handle multiple data streams like navigation, radar, and cameras simultaneously, enabling real-time target tracking, intelligent flight path optimization, and even on-board threat identification – all with minimal size, weight, and power consumption.

A drone flying across a field

Key Features

  • Processing System: Both Core and Edge series offer multiple processor types for handling various application tasks, ensuring smooth operation of the entire system.

  • Programmable Logic: This key feature allows for creating custom computational blocks within the chip, making it adaptable to ever-evolving algorithms and specific application needs (present on both Core and Edge series).

  • AI and DSP Engines:
    • Core Series - This series boasts specialized AI Engines for superior compute density in vector-based operations used in machine learning and advanced signal processing. It also includes traditional DSP engines for broader processing needs.
    • Edge Series - While the Edge series might lack dedicated AI Engines, it leverages its processing system and adaptable fabric to handle AI workloads efficiently, making it suitable for applications where size and power consumption are critical.

  • Next Generation I/O: Both series come equipped with high-speed I/O interfaces for seamless data transfer between the chip and external devices.

  • VDU (Video Decoder Unit): This feature is present only in the Core series and assists in processing and decoding various video formats, making it valuable for video analytics applications.

  • Programmable Network on Chip: This built-in network allows for efficient communication between different parts of the chip on both Core and Edge series, ensuring smooth data flow and maximizing overall performance.

Technology

AI/ML Overview

Avnet Silica brings extensive expertise in implementing machine learning on the edge, in the cloud, and on-premises. We support our customers in understanding and building their AI-based applications.

Artificial Intelligence depicted by connections and lights
Learn More