ASHBURN, Va. – The Curtiss-Wright Corp. Defense Solutions division in Ashburn, Va., is introducing the VPX3-4936 3U OpenVPX GPGPU processor module for deep learning, neural networks, artificial intelligence (AI), and machine learning.
The processor features the combination of the NVIDIA Ampere graphics processing unit (GPU) and a configurable Gen4 PCI Express switch. The NVIDIA Ampere architecture boosts performance and efficiency over the previous NVIDIA Turing generation, including more flexible concurrent execution of floating point and integer streams.
Example applications for the embedded computing processor include intelligence, surveillance, and reconnaissance (ISR); electronic warfare (EW), high-performance radar; signals intelligence (SIGINT), sensor fusion, and unmanned vehicles.
The Ampere device's third-generation Tensor cores deliver as much as four times the acceleration of AI and machine-learning algorithms, and its -RT cores and CUDA core architecture provide twice the performance compared to the previous generation.
While delivering close to 18 TFLOPS FP32 peak performance and 68 dense/136 sparse Tensor TOPS, the NVIDIA Ampere also improves power efficiency, yielding 154 GFLOPS per Watt. The module's PCI Express Gen4 architecture also doubles the host interface bandwidth, eliminating data throughput bottlenecks.
Pin-compatible with Curtiss-Wright's Turing architecture VPX3-4935, the VPX3-4936 enables system designers to boost math-intensive processing algorithms without increasing size, weight, and power (SWaP). The board's PCI Express architecture also supports Non-Transparent Bridging (NTB) and daisy chain options for system flexibility.
The rugged VPX-4936 module is designed in compliance with the U.S. Army's C5ISR/EW Modular Open Suite of Standards (CMOSS) and is aligned with the Sensor Open Systems Architecture (SOSA) technical standard to support compute-intensive ISR and EW systems.
For more information contact Curtiss-Wright Defense Solutions online at www.curtisswrightds.com.