Mercury unveils processing module with integrated AI

1 min read

Mercury Systems has launched the SCFE6931 processing module, said to be the first in the industry to incorporate integrated artificial intelligence (AI) processing functionality.

The system features dual Xilinx Versal AI Core adaptive compute acceleration platform (ACAP) processors, and the 6U OpenVPX heterogeneous processing module delivers performance improvements up to 20× more than other FPGA implementations and 100× more than the current fastest CPU implementations.

As a consequence theses modules are able to offer significantly more processing power for a wide variety of digital signal processing-intensive (DSP) applications such as radar, 5G wireless, electronic warfare (EW) and signals intelligence (SIGINT).

“Demanding radar, artificial intelligence and similar processing-intensive applications rely on rapid technology adoption to keep pace with evolving threats,” said Neal Austin, vice president and general manager, Mercury Microelectronics. “Mercury’s new ACAP-based signal processing modules meet our customers’ demand for greater processing power needed for real-time tactical decision making.”

The Versal ACAP AI processing power and architecture is abe to maximise performance, regardless of application or data type, by incorporating scalar processing, vector processing and next-generation FPGA fabric into a single 6U module.

Designed to be delivered in a variety of cooling options, the SCFE6931 is intended for applications that require high-performance operation in harsh environments. In addition, the module’s OpenVPX, SOSA-aligned design enables system integration.

As will all Mercury FPGA boards, the SCFE6931 module is built around EchoCore IP to provide design verification testing infrastructure functionality right out of the box, optimising time-to-market and reducing development time.

“Versal ACAPs have been architected to achieve new thresholds of system-level performance for a variety of aerospace and defence applications where size, weight and power (SWaP) are critical,” said Manuel Uhm, director of silicon marketing, Xilinx.