Tech for Tomorrow/ Hardware
Adding Magic
to geospatial intelligence
G
eospatial intelligence relies on the timely processing of large volumes of sensor data. For many applications, this means that the data can be acquired, stored, collated and delivered for processing on large workstations or even super computers far removed from the data source. However, there is a set of applications which require real-time (or near real-time) processing on mobile, often airborne, platforms — and this implies a set of constraints which drives the need for a low size, weight and power (SWaP) solution based on high performance embedded computing (HPEC) components. The MAGIC1 Rugged Display Computer from GE Intelligent Platforms is an example of such a low-SWaP system, combining the latest Intel Corei7 processors and NVIDIA GPUs in a true rugged system. Currently running with the 384-core NVIDIA EXK107 Kepler GPU, this year will see integration with the Intel 4th generation Corei7 processor, further enhancing its processing capability. How it works: The high speed interfaces can acquire data from multiple sensors, enabling the unit to run complex GEOINT algorithms, combining the real-time sensor data with databases stored in solid state memory to increase in-theatre loiter time while streaming actionable intelligence to a range of users within actionable time frames.
MAGIC1 22 / Geospatial World January 2014
Unique features • CUDA-enabled compute node. • Multiple video standards. • Dual channel output. • Up to 256 GB solid state disk. • Baseplate, convection or forced air cooled.
The Intel and NVIDIA architecture scales to larger arrays of CPU and GPU for systems that can tolerate a larger size, weight and power envelope. The IPN251, for example, combines the Corei7 and NVIDIA Kepler GPU with MellanoxInfiniBand, further enhancing the bandwidth of the external interfaces and reducing the latency of data throughput, leveraging GPUDirect technology to eliminate unnecessary memory copies as data moves around the system. Best suited for: GE Intelligent Platforms is actively working with NVIDIA on its GeoINT programme to ensure that customers who require the highest levels of performance on rugged mobile platforms can access the technology needed to succeed. Typical applications for these rugged platforms fall into the categories of command, control, computing, communication & information (C4I), and intelligence, surveillance and reconnaissance (ISR). C4I applications include search and rescue (SAR) — both maritime and land — and situational awareness. ISR applications are often performing similar GEOINT data fusion and analysis tasks, but may have the added requirement to send the resulting information to a ground station or other mobile platform via a low bandwidth data link. Using modular open systems architectures (MOSA) based on the best of commercial technology, the GEOINT community can benefit from mainstream investments while retaining all of the SWaP — and ruggedisation features needed to deploy complex data processing algorithms in the harshest of environments.