AFRL to use rugged HPEC system from GE
NewsJanuary 07, 2014
HUNTSVILLE, AL. Officials at the High Performance Systems Branch (RITB) with the US Air Force Research Laboratory's (AFRL's) Information Directorate selected computing systems from GE Intelligent Platforms for the development and eventual deployment of advanced neuromorphic architectures and algorithms for implementing adaptive learning, large-scale dynamic data analytics, and reasoning.
The GE High Performance Embedded Computing system (HPEC) uses NVIDIA GPU accelerators for real-time processing of high bandwidth data from RF sensors. It will support the Department of Defense's (DoD's) High Performance Computing Modernization Program (HPCMP), and will also be used for developing next-generation radar programs such as Gotcha wide-area synthetic aperture radar (SAR).
It is housed in a 6U OpenVPX rack mount chassis and can deliver 20 teraflops (20 trillion floating point operations per second) in computing horsepower. The system is also scalable and may be expanded to include additional racks and compute nodes.
“This embedded system, provides a path forward to apply large scale neuromorphic computing models for Air Force’s state-of-the-art ISR platforms and systems,” said Mark Barnell, HPS Program Manager, AFRL RITB.
Each rack in the modular GE system design is comprised of five SBC625 single board computers (SBCs) which have quad core Intel Core i7 processors, and RDMA-capable Mellanox 10 Gigabit Ethernet/InfiniBand adapters. The SBCs are coupled with modules that have the latest NVIDIA GPU accelerators based on the NVIDIA Kepler computing architecture, which deliver a total of 13,440 cores. Inter-board communication is achieved through GE’s 20-port IBX400 InfiniBand switch.
For more information, visit http://defense.ge-ip.com/products/hpec/c560.
