AMD Xilinx Makes Machine Vision AI Development Quick, Easy And Affordable

Consumer Tech AMD Xilinx Makes Machine Vision AI Development Quick, Easy And Affordable Marco Chiappetta Contributor Opinions expressed by Forbes Contributors are their own. I cover all things tech, from low-level chip designs, to storage, FPGAs, and the latest display technologies. New! Follow this author to improve your content experience.

Got it! Jun 22, 2022, 04:52pm EDT | Share to Facebook Share to Twitter Share to Linkedin Xilinx Kria Chip Xilinx Artificial Intelligence and Machine Vision have become and ever increasing part of our daily lives, both personally and professionally. Vision AI permeates everything from assembly line manufacturing to video chats, and touches our daily activities in many ways. As such, there is currently an immense demand for developers in the field.

To court some of those developers and offer an easy way to get started with machine vision, Xilinx (Pre-AMD acquisition) released the affordable KV260 Vision AI Development Kit. The KV260 features the company’s K26 SoM (System-On-Module) , along with a companion development board that breaks-out its IO connectivity, which consists of four USB 3 Type-A ports, HDMI and DisplayPorts, and headers for cameras and JTAG. A host of software tools, code samples, an app store, and pre-configured Linux images make getting started surprisingly simple.

KV260 Vision AI Dev Kit: Machine Vision With A Low Barrier To Entry The K26 SoM at the heart of the KV260 (and the recently-released KR260 for the Robotics market) features a Xilinx UltraScale+ multi-processor system on a chip (MPSoC), which is equipped quad Arm Cortex A53 CPU cores, a dedicated media engine for H. 264/265 video compression / decompression, and a programmable gate array with an impressive (for the sub-$200 price point) 256,000 logic cells. Also integrated is 4GB of DDR4 memory and a pair of 240-pin IO blocks, which are used to interface with the development board in the case of the KV260, but can also be used with various edge devices.

You can think of it as a configurable engine for AI-powered machine vision. All told, the kit offers support for up to 15 cameras, up to 40Gbps of aggregate Ethernet throughput, and four USB 2/3 ports to connect various peripherals. Xilinx KV260 Vision AI Development Kit Xilinx To best leverage the hardware, the KV260 Vision AI Development Kit is supported by Xilinx’s vast array of Vitis software development tools, which give developers the ability to work in their preferred native AI frameworks, like TensorFlow, PyTorch or Caffe, or programming languages like Python, C++ and OpenCL, while optimally leveraging the hardware resources on the KV260 SoM.

MORE FOR YOU Google Issues Warning For 2 Billion Chrome Users Forget The MacBook Pro, Apple Has Bigger Plans Google Discounts Pixel 6, Nest & Pixel Buds In Limited-Time Sale Event A Big, But Low Cost FPGA One of the key differentiators of the KV260 Vision AI Development Kit is its integrated FPGA with 256,000 logic cells. That FPGA is used to run what Xilinx calls AI accelerator applications, which is exactly what the kit is designed to do. Over at HotHardware, one of our team members just evaluated the KV260 and noted that, “the Xilinx Kria KV260 Vision AI Starter Kit is a speedy, affordable, and easy to grasp (for an experienced developer) way to get into machine learning accelerated on an FPGA”.

The Xilinx KV260 In Action HotHardware Although he initially hit a minor snag experimenting with a beta release of Ubuntu Linux for the KV260, he was up and running with the kit quickly and Xilinx’s code samples made testing face and object detection, among numerous other things, straightforward and simple. One of the major benefits of the KV260’s Vitis software platform and Xilinx’s code samples, is that they technically don’t require any actual FPGA programming experience. Virtually anyone can initially get started with the company’s pre-configured AI accelerators and sample apps and tweak them with applications written in C++ or Python using industry-standard tools like TensorFlow or PyTorch.

Of course, developers that want to take things to the next level and experiment with HDL design can do so with Vivado, but just getting started doesn’t require that level of expertise. Xilinx provides resources that explain how to migrate from NVIDIA CUDA and other frameworks to Vitis as well, so developers familiar with other, similar platforms can move to the KV260 with minimal friction if they so choose. The AMD Xilinx K26 Is For Robots Too The KV260 Vision AI Development Kit proved to be an affordable and easy way to enter the world of machine vision, but it’s not the only kit of its type to be offered by Xilinx, which is now part of AMD.

The KR260 Robotics AI Starter Kit , which is powered by the same K26 SoM, was released last month, but it features more Ethernet IO and the ability to cluster multiple kits together to drive entire robotics platforms. The KV260 Vision AI Development Kit is available directly from Xilinx for $199, while the KR260 with its additional IO has a somewhat higher, but still affordable, entry cost of $349. Follow me on Twitter or LinkedIn .

Check out my website . Marco Chiappetta Editorial Standards Print Reprints & Permissions.