Artificial Intelligence at the Edge

Xilinx

In edge AI applications, there is no room to be second place – almost in time is not an option. Performance at the edge, however, is not the only design consideration. These complex systems also require power efficiency, discreet form factors, peripherals, and above all, the ability to keep up with rapidly evolving AI models and frameworks. Xilinx adaptive platforms and solutions enable developers to meet all these challenging requirements in single device.

With Xilinx, latency at the edge is no issue. Xilinx enables optimized hardware acceleration of both AI inference and other performance-critical functions by tightly coupling custom accelerators into a dynamic architecture silicon device. This delivers end-to-end application performance that is significantly greater than a fixed-architecture AI accelerator like a GPU. With a GPU, the other performance-critical functions of the application must still run in software, without the performance or efficiency of custom hardware acceleration.

Reprogrammability is at the heart of all Xilinx technology. New structures and layers in AI models are being introduced all the time. With adaptive silicon, Xilinx allows domain-specific architectures to be updated, optimizing to the newest models without the need for new silicon. ASICs and ASSPs, however, have fixed architectures with long development cycles that simply cannot keep up with the speed of AI.

 

Processing at the edge is critical in most factories, hospitals, and smart city applications, where latency and determinism have a significant impact on revenue.  Avoidance of unexpected machine downtime, shorter patient wait/examination/diagnosis times, and quicker processing in high demand environments, like a sporting event, all directly tie into efficiency, productivity, and ultimately, money. As well, in intelligent applications such as factory automation, data has a shelf life. The more data can be consumed real-time, the more actionable the data.

Xilinx adaptive technology is proven to beat competitive offerings in latency reduction, power efficiency, and AI inference performance. Xilinx also offers the only comprehensive edge solution stack focused on functional safety, real-time vision and control, industrial networking, machine learning, mixed criticality software domains, and robust cybersecurity anchored in a hardware-root-of-trust without compromising on performance/watt, reliability over harsh conditions, and longevity of supply.

 

Zynq UltraScale+ MPSoC ZCU102 Evaluation Kit

The ZCU102 Evaluation Kit enables designers to jumpstart designs for Automotive, Industrial, Video and Communications applications. This kit features a Zynq UltraScale+ MPSoC device with a quad-core ARM Cortex-A53, dual-core Cortex-R5 real-time processors, and a Mali-400 MP2 graphics processing unit based on Xilinx's 16nm FinFET+ programmable logic fabric. The ZCU102 supports all major peripherals and interfaces enabling development for a wide range of applications.

Buy now

 

Ultra96-V2 I-Grade

The Ultra96-V2 is an Arm-based, Xilinx Zynq UltraScale+ MPSoC development board based on the Linaro 96Boards Consumer Edition (CE) specification. Available in both commercial and industrial grade options, the Ultra96-V2 is a perfect entry-level board for prototyping and developing edge compute applications.

Buy now

Spartan-7 FPGA SP701 Evaluation Kit

The SP701 Evaluation Kit, equipped with the best-in-class performance-per-watt Spartan®-7 FPGA, is built for designs requiring sensor fusion such as industrial networking, embedded vision, and automotive applications. The SP701 features high I/O availability and I/O expansion capability via Pmods and FMCs. The included XC7S100 FPGA is the highest density device of the Spartan-7 family. With 102K logic cells and 400 I/O pins, the XC7S100 pairs expandable connectivity with high processing capability, all in a form factor ideal for edge applications.

Buy now

MicroZed SOM I-Grade

MicroZed is a low-cost SOM that is based on the Xilinx Zynq-7000 SoC, available in commerical and industrial-grade options. The Zynq-7000 SoC features a dual-core Arm Cortex-A9 processing system tightly coupled with programmable logic in a single device for performance-per-watt optimization in applications such as machine vision and motion control. In addition to the Zynq-7000 SoC, the module contains the common functions and interfaces required to support the core of most SoC designs, including memory, configuration, Ethernet, USB, and clocks. On the bottom side of the module, MicroZed contains two 100-pin I/O headers that provide connection to two I/O banks on the programmable logic (PL) side of the Zynq-7000 SoC device. When plugged onto a user designed baseboard or carrier card, these 100-pin connectors provide connectivity between the Zynq-7000 SoC PL I/Os and the user circuits on the carrier card. MicroZed also includes on-board power regulation that supports 5 V input with an option to support 12 V input.

Buy now

From forward camera and surround view systems to imaging RADAR and LiDAR sensors, Xilinx adaptive solutions provide extremely reliable real-time performance at low power consumption for automotive edge applications. The adaptable architecture of Xilinx devices provides the ability to parallelize operations in programmable logic, reducing clock rates compared to GPUs and CPUs, and lowering overall power consumption to provide greater reliability and performance while staying within automotive thermal envelopes.

In landscape of automotive edge applications is still evolving at a rapid pace. Xilinx offers the only solution that can rapidly adapt to the latest AI networks from hardware to software in single device—meaning no need to waste precious design cycles and resources to redesign as necessary with ASIC, CPU or GPU-based solutions.

Whether reacting to dangers on the road or ensuring that automated features run smoothly as expected, Xilinx adaptive technology is there to help automakers meet strict demands for computation latency, performance, power efficiency, and adaptability for edge sensors to optimally handle a growing number of complex scenarios.

 

Zynq UltraScale+ MPSoC ZCU102 Evaluation Kit

The ZCU102 Evaluation Kit enables designers to jumpstart designs for Automotive, Industrial, Video and Communications applications. This kit features a Zynq UltraScale+ MPSoC device with a quad-core ARM Cortex-A53, dual-core Cortex-R5 real-time processors, and a Mali-400 MP2 graphics processing unit based on Xilinx's 16nm FinFET+ programmable logic fabric. The ZCU102 supports all major peripherals and interfaces enabling development for a wide range of applications.

Buy now

 


Spartan-7 FPGA SP701 Evaluation Kit

The SP701 Evaluation Kit, equipped with the best-in-class performance-per-watt Spartan®-7 FPGA, is built for designs requiring sensor fusion such as industrial networking, embedded vision, and automotive applications. The SP701 features high I/O availability and I/O expansion capability via Pmods and FMCs. The included XC7S100 FPGA is the highest density device of the Spartan-7 family. With 102K logic cells and 400 I/O pins, the XC7S100 pairs expandable connectivity with high processing capability, all in a form factor ideal for edge applications.

Buy now

Zynq UltraScale+ RFSoC ZCU111 Evaluation Kit

The Zynq® UltraScale+™ RFSoC ZCU111 evaluation kit enables designers to jumpstart RF-Class analog designs for wireless, cable access, early-warning(EW)/radar, and other high-performance RF applications. This kit features a Zynq UltraScale+ RFSoC supporting 8 12-bit 4.096GSPS ADCs, 8 14-bit 6.554GSPS DACs, and 8 soft-decision forward error correction (SD-FECs). Complete with Arm® Cortex®-A53 and Arm Cortex-R5 subsystems, UltraScale+ programmable logic, and the highest signal processing bandwidth in a Zynq UltraScale+ device, this kit provides a rapid, comprehensive RF Analog-to-Digital signal chain prototyping platform.

Buy now

UltraZed-EV SOM

UltraZed-EV SOM is a high performance, full-featured, System-On-Module (SOM) based on the Xilinx Zynq UltraScale+ MPSoC EV family of devices that have an integrated H.264 / H.265 video codec unit is capable of simultaneous encode and decode up to 4Kx2K (60fps). Designed in a small form factor, the UltraZed-EV SOM on-board dual system memory, high-speed transceivers, Ethernet, USB, and configuration memory provides an ideal platform for embedded video processing systems. The UltraZed-EV provides easy access to 152 user I/O pins, 26 PS MIO pins, 4 highspeed PS GTR transceivers along with 4 GTR reference clock inputs, and 16 PL high-speed GTH transceivers along with 8 GTH reference clock inputs through three I/O connectors on the backside of the module. These connectors provide USB 2.0, USB 3.0, PCIe Gen2, DisplayPort, SATA 3.0, FMC-HPC and more! 

Buy now

 

 

 

  1. On-Demand Webinar:  Accelerating Predictive Maintenance with Python and Neural Network-based Edge AI

    Learn how Xilinx-powered systems can incorporate intelligence to maximize productivity and lower downtime in the factory by applying predictive maintenance. Xilinx’s Python and Neural Network-based Edge AI Solutions simplify the implementation of hardware accelerated predictive maintenance to continuously monitor assets, analyze data, and intelligently plan service for these systems.

    LEARN MORE     

  2. Programmable Logic: Accelerating Automotive Applications

    As the demands on automotive capability increase, Xilinx Zynq UltraScale+ SoCs and supporting tool chains provide designers with the ability to achieve not only the desired performance for edge computing, but also the quality, functional safety and security required.

    LEARN MORE     

  3. Zynq UltraScale+ MPSoC Kit Selection Guide

    See how the ZCU102 kit compares with other Zynq UltraScale+ MPSoC kits and SOMs available from Xilinx and Avnet.

    LEARN MORE   

  4. Vitis-AI 1.1 Flow for Avnet VITIS Platforms - Part 1

    This guide provides detailed instructions for targeting the DNNDK samples from the Xilinx Vitis-AI 1.1 flow for Avnet Vitis 2019.2 platforms.

    LEARN MORE   

  5. Vitis-AI 1.1 Flow for Avnet VITIS Platforms - Part 2

    This guide provides detailed instructions for targeting the VART samples from the Xilinx Vitis-AI 1.1 flow for Avnet Vitis 2019.2 platforms.

    LEARN MORE   

AI at the Edge - Industrial

Learn More

AI at the Edge - Transportation

Learn more

Element14 Supplier Tech Spotlight

Learn More

Hackster Supplier Blog

Learn More

  
 

Xilinx Zynq UltraScale+ Kit

Looking for a platform with enough processing horsepower to jumpstart your next edge AI design? Learn more about the high-performance Zynq UltraScale+ Kit in our unboxing video.

Watch now