Defence Digital Signal Processing

Mercury Computer Systems, Inc.

Mercury Computer System’s common system architecture spans form factors and scales from two to hundreds of processors, leveraging industry-standard RapidIO® and RACE++® switch fabrics for data communications. Regardless of the architecture you select, you are assured of exceptional system performance, backed by the solutions expertise of Mercury's applications experts, computer scientists, and support and service team.

High-Density Computing
Systems from Mercury deliver extreme processing density with up to one TeraFLOPS or more of total processing power within a deployable chassis. Optimized for the deployment of high-end image and digital signal processing applications in the harsh, confined environments of airborne, ocean-going, or land-mobile platforms, Mercury’s exceptionally powerful solutions are based on open hardware and software standards.

Technologies that contribute to the high-performance density of Mercury systems include:

  • FPGA-based processing for greater performance per watt and unit of board real estate on certain critical functions
  • RapidIO and RACE++ switch fabrics for carrying the high data quantities common to these applications
  • Conduction cooling and ManagedAir™ solutions for removal of heat from dense configurations
  • Commercial off-the-shelf (COTS) products for proven industry-standard reliability and cost-effective implementation

FPGAs Accelerate Performance
Field-programmable gate arrays (FPGAs) in multiprocessing systems can outperform RISC processors by over 20x on certain tasks, reducing the overall system cost, size, weight, and power consumption. Mercury's FPGA compute node (FCN) solutions integrate a high-performance FPGA directly into the switch fabric, making it easier to use FPGAs for complex signal and image processing applications.

The Mercury RACE++ Series FCN products include critical, high-performance IP components for switch fabric interfaces, sensor I/O interfaces, and low- and high-level SRAM and SDRAM controllers. Two levels of memory controller allow developers to balance their requirements for ultimate memory performance with easier application development and integration. A default sample application allows for system checking and verification out of the box. Also included in the development software package is a simulation "test harness" that includes bus-functional modules of the RACE++ endpoints, the memory, and the I/O ports.

Scalable Performance
Scalability is built into each system from Mercury. Developers can prototype and test applications on small RACE++ systems with a handful of processors, then deploy finished applications on much larger systems running either the RACE++ or RapidIO switch fabric for optimal real-time performance. Additional processors and new technologies can be easily added to existing applications for greater performance or new functionality. Applications can also migrate freely across architectures and form factors with virtually no porting investment. You are never locked into any one architecture or form factor over time.

Multiple simultaneous transfers on the switch fabric enable data and instructions to flow reliably and predictably in all operating regimes. Adding processors, memory, or I/O nodes actually increases the total bandwidth of Mercury systems, keeping the system in balance as the application scales up or down.

High-Performance I/O Interfaces
The Mercury RACE++ Series MYRIAD™ family of I/O devices provides high-performance data acquisition, transfer, and storage for real-time applications including signal and image processing, telemetry, data collection, remote sensing, and simulation. In conjunction with RACE++ Series RINOJ streaming front-panel data solutions, these systems provide I/O functionality integrated into the switch fabric architecture for optimal performance. I/O solutions include Fibre Channel, HIPPI, and Common Data Link interfaces; RAID and JBOD systems (Ciprico and Storage Concepts); interfaces to high-speed data recorders from AMPEX, Sony, Enertec, and Metrum-Datatape, and advanced reconnaissance compression hardware.

Packaging for Extreme Environments
Conduction-cooling technology from Mercury extends the ultimate performance of defense products to the realm of extremely rugged applications. A common Mercury development and runtime platform is available for the full range of environments from commercial and laboratory settings to extremely rugged conditions at altitudes of up to 70,000 feet.

Mercury's deployment-ready solutions include 6U VME and 3U CompactPCI® (cPCI) processing boards, FPDP I/O interfaces, enclosure, power supply, systems, and system integration. Chassis options include ATR assemblies fitted with liquid cooling or conduction through the skin or frame of ships, aircraft, or land vehicles.

Proven COTS Success
Mercury is the principal force in real-time computing for hundreds of military programs requiring intensive signal and image processing. Premier integrators of defense electronics often rely on Mercury's commercial off-the-shelf (COTS) embedded computing products to power their leading-edge applications. Many of today's most prominent intelligence, surveillance, and reconnaissance platforms employ Mercury solutions, including the Predator and Global Hawk unmanned aerial vehicles, the Joint STARS aircraft, F/A-18 Hornet's SHARP reconnaissance pod, and many other signal processing solutions.

Integrated Solutions

MCP3 FCN Systems
High Performance Scalable to Small Form Factors
The MCP3 FCN board is a rugged, conduction-cooled 3U cPCI digital signal processing board designed for use in small spaces and harsh environments on deployed military platforms. It contains a Freescale™ MPC7447 microprocessor, a high gate-count Xilinx® Virtex-II® Pro P40 FPGA, a PMC site, and I/O options that provide the foundation for highly adaptable, high-performance digital receiver systems for a diverse range of applications. The PMC site can host Mercury’s dual-channel ADC PMC board or other standard conduction-cooled PMCs.

RACE++ VME Systems
Highly Scalable Computing for Rugged Environments
RACE++ VME systems provide real-time, embedded multicomputing in an industry-standard VME chassis. Scaling to hundreds of processors and gigabytes per second of communication bandwidth, RACE++ VME systems provide the processing power, I/O bandwidth, and deterministic low latency required by the most demanding real-time applications. RACE++ 6U VME systems include up to 80 PowerPC® G4+ compute nodes that provide up to 512 GFLOPS in a single chassis and are available in commercial, rugged, and conduction-cooled versions.

PowerStream® 7000 Systems
Extraordinary Performance Density for Military Deployment
The PowerStream 7000 system solves the embedded real-time computing industry's need for exceptional performance in a limited space, delivering one TFLOPS or more of total processing power within an air-cooled, deployable chassis. Based on the RapidIO communications fabric, the PowerStream 7000 system offers performance density exceeding 150 GFLOPS per cubic foot and up to 16 GB/s of streaming I/O. More processing power and I/O bandwidth in a smaller package enables full image and signal processing closer to the sensor in mobile platforms.

Processor Specifications

Processor Max* Processors Peak GFLOPS Peak GOPS Aggregate Bandwidth Bisection Bandwidth Maximum Memory per Processor
3U cPCI 24 194 768 3.0 GB/s 1.0 GB/s 512 MB
6U VME 80 512 2048 10.7 GB/s 2.1 GB/s 512 MB
Stream 7000
120 960 3840 75.0 GB/s 60.0 GB/s 512 MB

*Maximum number of processors at highest frequency available for each system. 3U includes FPGA processors.
Software Solutions for Development and Runtime

Mercury software is based on open industry standards and is implemented across runtime and development environments. Built upon a collection of open standards, the MCOE™ multicomputer operating environment gives users a high-performance, deterministic, multicomputer platform for their demanding applications. Mercury also provides a choice among popular operating system platforms, a mainstream integrated development environment (IDE), single and multiprocessor development and debugging tools, portable libraries, and system diagnostics.

Reliable Runtime Software
The Mercury runtime environment is the software foundation for Mercury's scalable, heterogeneous multicomputer systems. Mercury's SAL, VSIPL-Lite, and PixL™ algorithm libraries are hand-tuned for each supported processor, extracting maximum performance through optimal utilization of processor resources and by overlapping processing and I/O. To take advantage of high-speed switch fabric connections between processors, the Parallel Acceleration System (PAS™) delivers scalable high-performance multiprocessor communication and data reorganization with the industry-standard DRI API. Mercury's offline and online diagnostic software can test boards at power-up, monitor system health, and run diagnostics to identify and isolate system faults.

Multicomputer Application Development
Developing applications for a multicomputer requires not only strongly integrated tool suites, but also tools that provide insight into multiprocessor interactions. Mercury’s solution starts with the widely adopted Green Hills MULTI® integrated development environment, and adds a suite of multiprocessor analysis tools. For example, the TATL™ Trace Analysis Tool and Library is a unique and powerful tool for understanding and debugging multicomputer interactions in RACE++ multicomputers. This combination of mainstream tools and multiprocessor enhancements increases productivity while shortening the learning curve. This helps to reduce both your development costs and your time to market.

About Mercury Computer Systems
Mercury Computer Systems is the leading supplier of high-performance image and digital signal processing systems. These embedded computer systems transform streams of sensor data to visual information in real time for analysis and interpretation.

Mercury provides high-performance real-time embedded multiprocessing components to the defense electronics market. Since starting operations in 1983, our long-term focus has allowed us to develop an unmatched degree of expertise in technologies such as radar and sonar processing, signals intelligence, and EO/IR sensor fusion.

Mercury's degree of specialization allows us to maintain greater expertise in our selected fields than wider-ranging companies can afford, and Mercury's growth and strength gives us more engineering specialists than most competitors have employees. These internal resources are further augmented by a comprehensive service offering crafted from customer input and best practices. Through various support programs – Design, Program, Partner, Support, Solution, Consulting, and Training – Mercury and certified partners offer a wide range of services. These services include customized system design and integration, application software development, phone and field technical support, third-party product certification, and integration, training, and senior-level consulting.

A public company since 1998, Mercury Computer Systems is headquartered in Chelmsford, MA. Mercury also maintains engineering centers of excellence in Silver Springs, MD; Reston, VA; Carlsbad, CA; and San Diego, CA; with sales and support offices located in customer centers throughout the United States. Wholly owned subsidiaries, including technical expertise and sales support, serve local users in the United Kingdom, France, Germany, and Japan. A network of distributors and Mercury personnel support customers in other countries around the world.

Mercury Computer Systems, Inc.
199 Riverneck Road
Chelmsford, MA 01824-2820 U.S.A.
978-256-1300 ? Fax 978-256-3599

Visit company web site:

page up