General Micro Systems (GMS) Introduces “4×4” Artificial Intelligence Quad GPGPU Server for Defense and Government Applications
WASHINGTON–(COMMERCIAL THREAD) – General Micro Systems today launched its “4×4” quad GPGPU Velocity artificial intelligence (AI) server designed for battlefield applications such as command posts, tactical operations centers (TOC) and protected platforms such as autonomous vehicles, large body airborne sensor fusion, surveillance, data mining, synthetic vision and any platform requiring massive compute resources from a small air-cooled rack-mounted server. Available immediately, Velocity 4×4 ensures mission success with the value of a supercomputer AI algorithm processing data center in a single rack-mounted server equipped with industry standard connectors.
The NVIDIA A100 GPGPU AI Velocity 2U “4×4” quad server does the trick, balancing the server’s four high-performance areas for maximum data processing: CPUs, GPGPUs, 100Gb Ethernet sensor LAN ports, and 22 SSD bay. Velocity 2U has NVIDIA quad capability Ampere Core GPGPU coprocessors paired with four 28-core Intel Scalable Xeon data center processors, each with 1TB DDR4 ECC memory and 4TB total. The 112 processor cores and the four GPGPUs are connected through the server’s x16 PCIe Gen 3 fabric with 128 Gbps of onboard bandwidth, which is more than enough to keep up with the processing and storage of the AI sensors.
Velocity is designed from the ground up to balance the flow of information through the server, with the movement of data:
to / from the 16 Mellanox 100Gb Ethernet ports connected to real-time sensors or other subsystems
via the four Intel processors and the four A100 Ampere Core GPGPU coprocessors
to Broadcom RAID NVMe 22 SSD hardware array
GPGPUs, 100Gb Ethernet ports, CPUs, and SSDs are intelligently grouped and balanced to maximize throughput, reduce latency, and load processors and ports equally. The optional FIPS-140-2 encrypted SSDs are the fastest NVMe x4 drives available, streaming at 32 Gbps per drive. Data flows smoothly through the server, never stopping, further increasing overall performance.
The Velocity server is commercially available. The need for extreme “right now!” “ State-of-the-art computing and AI algorithm data processing on the platform is acute, as Army programs like Project Quarterback, ATLAS, NGCV, and other Futures Command projects require decisions in real time and sharing of multi-platform sensors. Sending data for processing to a command post, the cloud or CONUS is just too slow and the data links too congested.
“When we worked with the military to provide the local area network of sensors, server processor, and smart displays for the Multi-Function Video Display (MVD) mine clearance program on MRAPs, we saw the desire firsthand. insatiable to learn more about -processing the platform, ”said Ben Sharfi, founder and chief architect of GMS. “The Quad AI GPGPU ‘4×4’ server grew out of what GMS learned while working with the military on MVD and follow-up programs. It is also inspired by the conduction cooled GMS TITAN CSSP and CSVP 2U airborne servers used in the Navy’s P-8 Poseidon aircraft. In fact, the Quad AI GPGPU “4×4” server shares a similar commercial architecture with the Navy’s GMS servers, used to run the Minotaur software mission system for Maritime patrol and reconnaissance. Army and Navy systems require massive AI workloads in very small racks. Both programs are based on GMS servers.
“With only a tiny 2U slot available, we deliver four of NVIDIA’s latest available Ampere GPGPUs, four of Intel’s best data center processors, 16 Mellanox 100 Gigabit Ethernet ports and 264TB of NVMe storage controlled by Broadcom RAID,” Sharfi added. “That’s more than anyone has seen on the battlefield to date, in just 2U. For example, in real time, the Quad AI Velocity “4×4” server can accept high-resolution image data from 1600 high-resolution 1 Gbit sensors, then process, analyze and perform fast Fourier transformations for frequency domain computation. before converting the data back. in the time domain and storing 264TB. The capabilities of the Velocity 2U are far beyond what the military has seen before on the battlefield. ”
The main performance specifications for the Velocity 2U server include:
4x Intel Scalable Xeon 28-core data center processors (112 cores total)
4 PCIe Gen 3 x16 slots for 4 dual-slot 250W AI GPGPU coprocessors (or equivalent)
4TB Balanced DDR4 ECC SDRAM
16 100 Gb Ethernet LAN sensor ports
22 SSD bay with up to 264TB of storage, RAID-capable NVMe x4 PCIe Gen 3 drives
144 PCIe Gen 3 lanes via 8 user I / O module slots
Operates from 0 ° C to +40 ° C (optional -20 ° C to +75 ° C)
2U design at just 30 “deep
Hot-swappable and redundant CRPS power supplies
GMS’s proven RuggedDNA ™ cooling, conditioning and hardening techniques are carried over into Velocity “4×4” servers from 40 years of deployed military VME, VPX and small form factor systems. The servers exceed typical data center specifications for extended temperatures, high shock and vibration, and are ideal for stationary and mobile (vehicle) applications. Servers can operate over a wider temperature range of -20 ° C to + 75 ° C while providing the long life required for Department of Defense (DoD), government or other high availability applications and high reliability.
GMS Rugged Server Family Provides Critical Real-Time AI Sensor Data Processing and Logging
The Velocity server is one of a family of rugged 1U and 2U servers. All Velocity and Vortex shortrack sister server variants have 1 GbE LAN port, video console, optional hardware or software RAID, BMC with dedicated Ethernet port, TPM and optional 10/40 GbE (copper or fiber). A DVD / Blu-ray player is also optional, as are serial ports and audio I / O. Single or dual redundant power supplies are standard, 110 VAC or 220 VAC, as well as some DC voltages. Optional features such as zeroing via GMS SecureDNA ™ or secure hardware erase via FIPS-140 or Opal 197 disks, enable data at rest encryption on removable SSD media.
Designed in America
GMS proudly designed the Velocity and Vortex servers in America with the same design philosophy as the company’s TITAN 1U and 2U conduction-cooled MIL-SPEC servers, and the conduction-cooled small form factor (SFF) chassis and systems deployed by GMS. In Velocity servers, however, the fans controlled by TwoCool ™ BMC reliably attenuate heat while minimizing costs.
Or: Booth # 7751, AUSA Annual Meeting and Expo, Washington, DC, October 11-13, 2021
About General Micro Systems:
General Micro Systems (GMS) is the rugged server company. The company is known as the industry expert in small form factor, modular, compute-intensive and rugged integrated systems, servers and switches. These powerful systems are ideal for demanding C5ISR defense, aerospace, medical, industrial and energy exploration applications. GMS is an IEC, ISO, AS9100, NIST-800-171, and MIL-SPEC provider with infrastructure and operations for long-life, specification-controlled and configuration-managed programs.
Designed from the ground up to provide the best performance and functionality in the world’s toughest environments, the company’s highly customizable products include GMS Rugged DNA ™ with patented RuggedCool ™ cooling technology. GMS is also the leader in high-end deployable Intel processors® processors and a proud Intel® partner since 1986. For more information, visit www.gms4sbc.com.
General Micro Systems and the General Micro Systems logo are trademarks of General Micro Systems, Inc. All other product or service names are the property of their respective owners.
© 2021 General Micro Systems, Inc. All rights reserved