Skip to main content Skip to main content

EnsembleSeries™ GSC6204 - Powerful GPU Co-Processor

6U OpenVPX GPU NVIDIA Quadro Co-Processor


By off-loading compute-intense functions to GPU co-processors, you can build in the scalable processing power to handle the most complex workloads. Commercial and defense compute-intensive applications including radar, EO/IR imagery, AI, urban air mobility, autonomous driving, cognitive EW and sensor fusion generate huge volumes of sensor and network data that requires big processing resources to extract its information in real time. Powered by NVIDIA® Quadro® TU104 GPUs, GSC6204 co-processing modules deliver the floating-point math compute performance required for efficient off-loading.

 MOTS.png


Extreme Processing Performance

GSC6204 modules incorporate the NVIDIA Turing GPU architecture to bring the latest commercial processing advancements and scalability to the embedded domain. Powered by dual Quadro TU104 processors and incorporating NVIDIA’s NVLink™ high-speed direct GPU-to-GPU interconnect technology and a PCIe architecture, these modules embed the same massive parallel processing capability found in contemporary data centers into your edge application. For enhanced performance, each GPU features NVIDIA’s Tensor Cores for mixed-precision matrix multiply and accumulate calculations in single operations, which are a key component of many AI, deep learning, and signal processing and fusion workloads.

Embedded Cognitive Computing and AI for Military Applications


Embedded Composable Data Center

composable-PCIe-HPEEC.png

GSC6204 GPU co-processing engines are a critical component of a truly composable high-performance embedded edge compute (HPEEC) environment with processing power unmatched by competing approaches. Combined with Mercury’s HDS6605 Intel® Xeon® scalable server blades, SCM6010 fast storage, SFM6126 wideband PCIe switches and streaming IOM-400 I/O modules forms a truly composable high-performance embedded edge compute (HPEEC) environment. The HPEEC approach leverages the same tools, architecture and software used in the data center, enabling your application to scale from the cloud to the edge with ease.


The Anatomy of a Rugged GPU Co-Processor Module

exploded-view-with-call-outs.png

GSC6204 modules protect electronics, keeping them cool for long, reliable service lives and delivering consistent switch fabric performance across a broad temperature range. Our proven fifth generation of advanced packaging, cooling and interconnects technologies enable GSC6204 modules to be the first GPU powered OpenVPX co-processors that are truly rugged for applications requiring protection from harsh environments. For extreme environmental protection, GSC6204 modules are available with optional MOTS+ technologies.


Affordable, Interoperable and Upgradable

Cloud-Fog-Edge 1100px.png

Place the best technology in your application faster by building your next HPEEC processing solution with compute modules that are made to open architectures including Sensor Open System Architecture (SOSA) for interoperability, scalability and reuse. Further drive affordability and performance and lower your program risk by using the architecture that evolves at the speed of technology, follows industry technology roadmaps and leverages data center hardware, software and IP.

 

 

Learn More About MOSA for Weapon Systems


Gain the Competitive Edge

Speed-SWap-Software-Security-Silicon-Hex-chart_1140px.png

The HPEEC architecture enables you to deploy smarter and more automatous platforms and compute-intense sensor processing solutions that require data center capabilities closer to the data source – at the edge. Efficiently process streaming data over a distributed, heterogeneous PCIe-enabled and scalable common data center architecture. Then leverage the data center software and tools to drive your program velocity, while lowering your risk through much greater IP reuse.


Resources

The Military Digital Convergence - enables next-generation military platforms

Embedded Cognitive Computing & Artificial Intelligence - part 1

High-performance OpenVPX Forced Air-cooling Architectures


Specifications

  • NVIDIA Tensor and Ray Tracing Cores for AI and graphics applications
  • 6144 CUDA cores and 32GB GDDR6 memory
  • PCIe Gen 3 and NVIDIA NVLink high-speed interconnects
  • SOSA aligned and VITA 65 compliant
  • NVIDIA, CUDA and OpenCL software support
  • Optional MOTS+ extreme environmental protection
  • Designed, built, coded and supported from trusted DMEA facilities

NVIDIA.png   

  • 6U OpenVPX, standard 1-inch pitch module
  • Weight 2 to 3Kg depending upon configuration

Other HPEEC Building Blocks


Mercury Systems, Inc.
50 Minuteman Road
Andover, MA 01810
(978) 256-1300

Copyright © 2020 Mercury Systems, Inc. All rights reserved. From FORTUNE. ©2020 FORTUNE Media IP Limited All rights reserved. Used under license. FORTUNE® is a registered trademark of FORTUNE Media IP Limited and is used under license. FORTUNE and FORTUNE Media IP Limited are not affiliated with, and do not endorse products or services of, Mercury Systems, Inc.. Legal / Privacy Policy