AMD combines Ryzen Embedded processors and Versal AI Edge AI

AMD combines Ryzen Embedded processors and Versal AI Edge AI chips into a platform for unmanned vehicles, medicine and industry

AMD is quite active in its development of chips for embedded systems, as these solutions are used in industrial, automotive, commercial and medical sectors, in remote digital gaming systems and in other areas. AMD today introduced the new Embedded+ platform, combining Ryzen Embedded processors on the Zen+ architecture, as well as Versal adaptive SoCs on a single board.

  Image Source: AMD

Image Source: AMD

The Embedded+ platform combines the capabilities of Ryzen Embedded processors and Versal AI Edge scalable SoCs into a turnkey or near-ready solution. With this platform, the manufacturer is targeting those areas where high computing power and at the same time high energy efficiency are required. The platform allows you to work with various AI algorithms, as well as manage data received in real time from various sensor systems, which are actively used in modern production scenarios and environments.

Embedded+ solutions are designed for AMD ODM partners. They can be extremely useful in use cases where maximum responsiveness is required between system components and software. Therefore, Embedded+ can, for example, be used in self-driving systems, diagnostic medical equipment, as well as in high-precision manufacturing systems. The AMD Embedded+ architecture supports a variety of workloads designed for different processor types, including x86-compatible and Arm solutions. In addition, it can work with those artificial intelligence tools and FPGA structures that provide flexibility and scalability to embedded computing solutions across a variety of industries.

AMD Embedded+ architecture provides broad compatibility with a variety of sensor types and related interfaces. It supports direct connection to standard peripherals via Ethernet, USB, HDMI and DisplayPort interfaces. The platform’s embedded AMD Ryzen processors can handle input from common image sensors, including RGB, monochrome, and even advanced neuromorphic types, while supporting industry standard image sensors such as MIPI and LVDS.

Used as part of the AMD Versal AI Edge SoC platform, in turn, can process signals in real time using AI algorithms from sensors like LiDAR and others that are used in modern embedded systems in the manufacturing, medical and automotive sectors. In addition, these chips claim to support GMSL and Ethernet interfaces and protocols.

In addition to the Embedded+ architecture itself, AMD presented one of the ready-made solutions based on it. The compact Sapphire Technology VPR-4616-MB platform consists of a Mini-ITX motherboard on which the AMD Versal AI Edge 2302 SoC is installed with 34 AI engines providing 22 TOPS performance, as well as a Ryzen Embedded R2314 processor with four cores and four threads, as well as six integrated Radeon Vega graphics units. The board also offers a variety of interfaces, including two SO-DIMMs for DDR4 modules supporting up to 64 GB of memory, one PCIe 3.0 x4 M.2, one SATA III, one 2.5-Gigabit network adapter and an M.2 Key E 2230 slot PCIe x1 for Wi-Fi wireless modules. The platform claims to support the Ubuntu 22.04 operating system.

In addition, AMD has introduced a number of different expansion cards that will increase the capabilities of the Embedded+ architecture. This, for example, includes the Octo GMSL camera I/O board, which allows you to simultaneously interact with several cameras. It can be used by high-throughput machine vision systems, which in turn are an integral part of advanced driver assistance systems (ADAS) and automated surveillance systems. Such solutions often require the integration of multiple image inputs for real-time processing and analysis, and the Octo GMSL board is specifically designed to meet this requirement.

In addition, AMD is introducing a dual-link Ethernet NIC for Embedded+ that supports 10/100/1000 Mbps connection speeds, as well as a Dual 10 Gb SFP+ card that includes 16 GPIO ports to meet even higher bandwidth requirements. The board is designed to provide sufficient data transfer rates for applications such as real-time video streaming and large-scale aggregation of image sensor data. All this expands the possibilities of using the Embedded+ architecture in peripheral and industrial scenarios.

The Sapphire VPR-4616-MB platform is already available for purchase and can be offered in the form of fully assembled configurations with RAM and ROM, power supply and housing.

About the author

Dylan Harris

Dylan Harris is fascinated by tests and reviews of computer hardware.

Add Comment

Click here to post a comment