Mercury’s high density server-class processing solutions exceed objectives in US Air Force-led open architecture study

Mercury Systems, Inc., announced the results of its participation in the US Air Force-led Next Generation Radar (NGR) Processor Study on September 2. The goal of this program is to assess the capability of current embedded computing open architectures to perform airborne radar signal processing on future USAF platforms. Chelmsford, MA-based Mercury demonstrated that its scalable and rugged high-density server modules exceeded the target benchmarks in the study, enabling affordable capability evolution of radar processors through open, non-proprietary and standards-based systems for airborne solutions.

“Mercury’s exceptional performance in this study reflects our ongoing investment and commitment to delivering the most powerful OpenVPX^™ processing modules in the embedded computing industry,” said Ian Dunn, Vice President and General Manager of Mercury’s Embedded Products Group. “Now in its fourth generation, our powerful, open standards-based Ensemble^®HDS660x modules meet the DoD’s expectations for affordability, security, modularity and OSA requirements and can be deployed in harsh environments to take cloud processing for military and aerospace applications right to the tactical edge.”

As part of the study Mercury was required to satisfy stringent performance requirements when executing USAF-provided ground moving target indicator (GMTI) and synthetic aperture radar (SAR) benchmarks. These benchmarks posed real-world processing requirements representative for previously fielded and future planned airborne systems. A dense configuration of three Mercury HDS6603 high density server processing modules surpassed the study’s most demanding performance thresholds, providing in excess of 4 TFLOPS of computing power resulting in the most capable COTS processing system yet to be tested.

Introduced by Mercury last year, the Ensemble^® HDS6603 is the fourth generation of Mercury’s high density Intel^® Xeon^®-based processing modules, with dual 12-core Intel “Haswell” e5 server-class processors, enhanced rugged packaging, 64 GB DRAM memory and dual QPI interconnect to support 24 cores with hyperthreading. Leveraging Intel’s “Ivy Bridge” architecture, each module features high-speed fabric interfaces for sensor data exchange and a dual x16 PCIe Gen3 expansion plane. All Ensemble high density processing modules are available in air-cooled, conduction-cooled, Air Flow-By^™ and Liquid Flow-By^™ packaging options.

An HDS6603-based system meeting the USAF-provided pod-class size, weight and power (SWaP) requirements supports nearly 100% growth beyond the already challenging study performance objectives. This open architecture solution also can easily be scaled to a cabin-class solution of more than 20 TFLOPS of processing power while remaining well within the SWaP requirements provided by the USAF for this class of application.

Source: Mercury Systems Inc.