SCIENCE

Mercury Computer Systems announced greatly improved radar subsystem performance through two new innovations: a general purpose Graphics Processing Unit (GPGPU) product based on the NVIDIA “Fermi” architecture, and a 10 Gigabit Ethernet (10GE) standards-based real time sensor interface module.

These products enable unprecedented levels of Size, Weight and Power (SWaP) optimization for radar applications through the highest TeraFLOP-per-slot compute performance metric and the highest I/O channel density per slot available in the defense industry today. This extraordinary level of performance is required to meet the stringent demands of modern radar, including the ability to search and track smaller, more numerous and faster targets in the harshest environments.

“Last year, Mercury announced the first fielded GPGPU-based ISR subsystem, which is flying today,” said Didier Thibaud, senior vice president and general manager of Mercury Computer Systems’ Advanced Computing Solutions business unit. “We are leveraging the high-performance, rugged and upgradeable aspects of this GPGPU innovation into our next-generation radar subsystems and extending it with massive I/O. Together with our industry-leading rugged OpenVPX Intel modules, these new capabilities enable our SWaP-optimized radar subsystems to ‘do more with less’ so we can help our customers meet the challenges of the modern battlefield.”

Mercury’s new subsystem enhancements are based on powerful building-block components in 6U OpenVPX standard form factors. Like all of Mercury’s open architecture building blocks, these modules can be configured with other components (such as Intel rugged processor cards and switch modules) into advanced Application Ready Subsystems (ARS) as part of larger radar systems. Each ARS is a customized design with unique, application-specific capabilities. The individual modules can also be adapted to specific program needs by Mercury’s Services and Systems Integration (SSI) group.

Powerful Building-Block Modules

Mercury’s radar Application Ready Subsystems are powered by the Ensemble 6000 Series 6U OpenVPX Intel-based modules and enhanced by the new GSC6200 GPU Processing Module and the Ensemble IO Mezzanine Series IOM-200 XMC and IOR-280 RTM building-block modules. Using two NVIDIA GeForce GTX 460M GPGPUs based on their latest Fermi GPU architecture, the GSC6200 delivers a combined 384 processing cores, 3GB of high-bandwidth GDDR5 SGRAM and more than 1 TFLOP of peak theoretical performance per 6U OpenVPX slot. It supports CUDA™ (Compute Unified Device Architecture), OpenCL and Mercury’s MultiCore Plus™ MathPack C/C++ software development environments.

Crucial algorithms for radar signal processing such as adaptive beam-forming, pulse compression, constant false alarm rate (CFAR) and cross correlation are greatly accelerated by the native, highly parallel architecture of the GPU. The high gigaflop-per-Watt performance delivered by the GSC6200 helps reduce board count, which in turn reduces system size and weight. Further, Mercury’s innovative board design houses the GPGPUs in an easily upgradeable MXM form factor, enabling customers to quickly validate and deploy the most current GPGPU technology available.

The IOM-200 XMC features four 10Gbps Ethernet I/O channels on a single XMC card and either a high-performance Altera Stratix IV 230 or 360 FPGA, which can be used for algorithm partitioning and acceleration. When combined with Mercury’s Ensemble LDS6520 Intel-based processor module, up to eight channels of 10 Gigabit Ethernet can be supported per 6U OpenVPX slot, which ensures that all processing cycles are being utilized.

In addition to advanced technical capabilities, Mercury’s scalable signal processing solutions deliver program-level advantages to radar customers for multi-mode and multi-mission applications. Integrated, tested and validated by Mercury, these subsystems help customers reduce both technical and business risks by supporting compressed development cycles and faster deployment of new programs.