AMD Powers AI Inference on COM Express Board

congatec, a leading technology company based in Germany, has recently introduced a new line of COM Express Compact Computer-on-Modules designed specifically for AI applications. The conga-TCR8 Type 6 COM Express modules are powered by the latest Ryzen Embedded 8000 Series processors from AMD, a company that has positioned itself as a key player in the data center chip market. These processors boast up to eight Zen 4 cores, a XDNA neural processing unit (NPU), and Radeon RDNA 3 graphics, providing a remarkable performance of up to 39 tera operations per second (TOPS) for AI inference in edge applications.

Targeted at high-volume and price-sensitive applications that demand a blend of advanced AI capabilities, superior graphics performance, and robust computing power, the congatec COM Express modules cater to a wide range of industries. OEMs operating in sectors such as medical imaging, test & measurement, AI-supported POS/POI systems, and professional gaming can leverage these modules to drive innovation and ensure long-term investment security.

In a move that underscores congatec's commitment to innovation and adaptability, the company has appointed its third CEO in as many years. This leadership change comes at a time when congatec is expanding its product offerings and strengthening its position in the competitive tech market.

One of the key features of the conga-TCR8 Computer-on-Modules is their scalable TDP thermal envelope, ranging from 15 to 54 W. This versatility makes the modules ideal for upgrading existing designs, allowing companies to enhance their products by simply swapping out modules. By doing so, businesses can bring their offerings up to date with the latest technology, thereby improving product lifecycle, return on investment (ROI), and sustainability.

The new conga-TCR8 modules are available with a choice of four different AMD Ryzen Embedded 8000 processors, each featuring six or eight Zen 4 cores. These modules support up to 128GB of DDR5-5600 memory with error correction code (ECC), making them suitable for data-intensive and data-critical applications. The Radeon RDNA 3 GPU, with up to 12 compute units, can be utilized for AI tasks, while also supporting immersive graphics output on up to four displays with resolutions as high as 8k.