Xilinx unveiled its Vitis AI software platform in late 2019, to simplify the development of enterprise chip-level hardware architectures and accelerators, using common software development tools and programming languages. Since its unveiling, the Vitis AI development environment has evolved and expanded to incorporate more tools, design examples, libraries, and templates designed to make the most of the diverse range of devices in Xilinx Alveo Computing, Accelerators and Expansion Cards. Today, the company is releasing the latest version of the suite, Vitis AI 1.4, which supports Xilinx’s 7nm Versal ACAP platforms, including the Versal AI Core Series VCK190 Evaluation Kit, the Versal Development Board VCK5000 and the recent Kria 16nm Adaptive Systems-on-Modules (SOM) Portfolio.
With this release and the integration of support for additional hardware platforms, Xilinx provides a kind of one-stop shop for developers who can handle a wide range of machine learning and inference workloads. IA, to a multitude of power envelopes and performance levels. The Xilinx Versal AI Core Series VCK190 Evaluation Kit, for example, is capable of MLPerf inference performance levels that far exceed current server processors. In a closed-division ResNet50 benchmark test, the versatile VCK5000 development board also outperformed NVIDIA’s T4 board with a result of 5,921 fps. And the Kria KV260 Vision AI Starter Kit under $ 200 can be used for vision application development with an inexpensive, low-power kit that can literally fit in the palm of your hand.
In addition to supporting additional hardware, a key component of Vitis AI, the AI Model Zoo, has been improved. The diversity of the AI Model Zoo has been increased and now offers models for 4D radar detection, image-LIDAR sensor fusion, panoramic 3D detection, depth estimation and super resolution.
Xilinx has taken a number of steps over the past few years to enable developers and engineers to work within their native AI frameworks, such as TensorFlow, PyTorch, and Caffe, while targeting the broad spectrum of computing and computing devices. business accelerators. Standard programming languages such as Python, C ++, and OpenCL are also supported and optimized for Xilinx hardware. All in all, Xilinx’s expanding software ecosystem and the breadth of development kits allow developers of virtually any skill level to take advantage of powerful adaptable accelerators that they may have previously been out of reach. Today’s release of Vitis AI 1.4 further strengthens Xilinx’s position and makes its hardware and software platforms even more accessible to developers of all backgrounds.