6U OpenVPX GPGPU Processor Card based on industry-standard MXM modules, is fitted with a pair of 240 core GPU devices each with 2GB of GDDR5 memory
The VPX6-490 GPGPU Application Accelerator provides two NVIDIA Fermi architecture GPU devices on a 6U VPX board. Based on industry-standard MXM modules, the VPX6-490 is fitted with a pair of 240 core GPU devices each with 2GB of GDDR5 memory. The VPX6-490 leverages new OpenVPX profile standards to deploy the bandwidth of two 16-lane Gen2 PCI Express interfaces which is critical to maximizing the performance that can be extracted from the GPUs.
The VPX6-490 is designed to work with Intel processor-based SBCs such as the Curtiss-Wright VPX6-1959 Intel Core i7 Broadwell SBC, as well as the CHAMP-AV9 dual quad-core 4th Generation Core i7 DSP board.The addition of GPU technology complements Curtiss-Wright High Performance Embedded Computing (HPEC) portfolio of products that address the complete data flow from sensor to operator: FMC/XMC data acquisition, FPGA-based processing modules, high-density multiprocessor boards, GPGPU co-processors and a large catalog of system level interface products to connect to other computers or display information to human operators.
- 6U OpenVPX form factor
- One or two NVIDIA Fermi architecture GPUs
- Dual DVI graphics outputs
- Temperature sensors, voltage and current sensors
- IPMI support
- 12V power
- Rugged air-cooled, conduction-cooled and air flow-through versions
- CUDA 4.1 support for Linux
- Applications: Intelligence, Surveillance, Reconnaissance (ISR), Electronic Warfare (EW), Industrial Compute intensive video rendering, CUDA math engine for 6U VPX subsystems