NVIDIA 3U and 6U OpenVPX embedded computing modules for artificial intelligence introduced by Curtiss-Wright
VPX3-4933 and VPX6-4953 modules are for radar, SIGINT, EO/IR, data fusion, and other uses where teraFLOPs of accelerated processing are necessary.
ASHBURN, Va. – The Curtiss-Wright Corp. Defense Solutions division in Ashburn, Va., is introducing two NVIDIA Quadro Pascal (P5200) GPGPU/Inference Engine-based OpenVPX embedded computing modules for compute-intensive applications in artificial intelligence (AI); intelligence, surveillance and reconnaissance (ISR); and electronic warfare (EW).
The rugged VPX3-4933 and VPX6-4953 modules are for high-performance radar, SIGINT, EO/IR, data fusion ingest, processing, display, and other applications where trillions of floating point operations per second (teraFLOPs) of accelerated processing are required. Curtiss-Wright is introducing these products through its reseller agreement with WOLF Advanced Technology.
The VPX3-4933 module is 3U OpenVPX GPGPU processor with one NVIDIA Quadro P5200 GPU that delivers 8.7 teraFLOPS performance. For more demanding applications, the 6U VPX6-4953 hosts dual P5200s, doubling available compute power to 17.4 teraFLOPS.
To meet demanding rugged military and aerospace specifications, both GPGPU boards feature a chip-down design. The P5200 GPU has 2560 CUDA cores, and provides a maximum memory bandwidth of 243 gigabytes per second.
Curtiss-Wright embedded computing systems are supported by the OpenHPEC Accelerator suite of software development tools that enable system developers to develop their software quickly.
For high-speed, low-latency, peer-to-peer communications, the OpenHPEC Accelerator Suite also includes Dolphin’s PCI Express communication library, which hides the complexities of directly programming the embedded computing system’s PCI Express devices.
For more information contact Curtiss-Wright online at www.curtisswrightds.com.