AV800-D27-A28

AV800-D27-A28 400G Airborne GPU Computer
AV800-D27-A28 400G Airborne GPU Computer
AV800-D27-A28 400G Airborne GPU Computer
AV800-D27-A28 400G Airborne GPU Computer
AV800-D27-A28 400G Airborne GPU Computer

400G Airborne GPU Computer , 4xQSFP28 ConnectX-6, Xeon-D 2796NT (20C)

  • Intel® XEON D-2775TE 20 Cores Turbo 3.1GHz
  • 512G LRDIMM DDR4-2933 / 256GB RDIMM ECC DDR4-2933
  • Nvidia RTX A2000 2560 CUDA Cores 
  • 4x100GbE Single-Port QSFP28, Nvidia ConnectX-6
  • 4x 1GBase-T, 2x 10GBase-T LAN (Options)
  • 2x 2TB 2.5” Swappable SATA Drive with AES function
  • Hardware Secure Erase(AES) button, Swappable CMOS battery
  • IP65 Sealed with External Cooling Blade
  • MIL-STD-810G Thermal, Shock, Vibration, Humidity
  • MIL-STD 18V~36V EMI DC Input
  • Extreme Temperature -20°C to 60°C

 Intel Gold

Download
Get A Quote
  • Technical Profile
  • Specifications
  • Thermal Solution
MAIN FEATURE
  • Ultra High Performance Intel® Xeon Ice Lake-D, D-2796NT (20xCores)
  • NVIDIA MXM-GPU Quadro RTX A2000 2560 CUDA
  • 4x 100GbE QSFP28
  • Up to 512GB LRDIMM /256GB RDIMM
  • Dual Removable Anti-Drop Solid-State Disk
  • Hardware Secure Erase (AES) by option
  • IP65 Sealed with External Cooling Blade
  • MIL-STD-810 Thermal, Shock, Vibration, Humidity
  • MIL-STD 461 EMI/EMC 18V~36V DC-IN(28VDC)
  • Extreme Temperature Support -20°C~+60°C
Ice Lake_X12SDV
INTRODUCTION

satellite

AV800-D27 is 7STARLAKE ruggedized AI Inference platform specifically designed for Intel® Xeon® Ice Lake D-2796NT Processor(20 Cores, 2.0GHz Max Turbo 3.10GHz), NVIDIA MXM RTX A4500 and NVIDIA® ConnectX®-6 DX (200Gb/s, signal port)

 

With built-in acceleration Intel® Deep Learning Boost, AV800-D27 is able to deliver server-class  and high-end performance for rugged IoT devices. Combining stunning inference performance and high-bandwidth (200Gb/s) networks, AV800-D27 is the perfect HPC platform for versatile edge AI applications.

Designed with NVIDIA MXM RTX A4500(Ampere Architecture-Based 5888 CUDA cores, 46 RT Cores, and 184 Tensor Cores), AV800-D27 will help achieve powerful performance for data-intensive industries such as medical, defense, and transportation.

What’s more, with NVIDIA® ConnectX®-6 DX, AV800-D27 is capable of providing single port of 200Gb/s Ethernet connectivity and highly secure smart network interface card (SmartNIC), which accelerates performance, security, virtualization, SDN/NFV, big data, machine-learning for data center applications.

SR800-D27

 

I.    Ultra-High Performance Intel® Xeon® D Processor ICELAKE-D HCC:

圖片5_1.png

Intel® Xeon® processor D-2700 product family offers hardware and software scalability up to sixteen cores, making it the perfect choice for a broad range of high-performing, low-power solutions that will bring intelligence and Intel® Xeon® reliability, availability, and serviceability (RAS) to the edge.

 圖片8_0.png   圖片7_1.png

Enhanced performance per watt:

Intel® Xeon® processor D-2700 product family delivers exceptional value and unmatched performance density per watt. Its TDP of 100W, industry-leading 10 nm process technology and a compute-only design make it ideal for meeting the diverse needs of customers seeking mid-range low-power, high-density solutions.

II.    NVIDIA Quadro RTXA4500 MXM 

AV800-D27 supports 1x NVIDIA® Quadro RTXA4500 MXM Module; can power the planets most reliable mainstream workstations. Designed into a 115-watt package, RTXA4500 is powered by NVIDIA Ampere architecture, supplying innovative multi-precision performance to accelerate a vast range of modern applications. AV800-D27 w/ Quadro RTXA4500 GPU accelerates diverse cloud workloads. These include high-performance computing, data analytics, deep learning training and inference, graphics and machine learning. RTXA4500 MXM features multi-precision Turing Tensor Cores. It comes in a very compact MXM form factor, helping AV800-D27 deliver ground-breaking performance at scale.

23123.jpg

III.    NVMe U.2 Conduction Cooled Pack

A Quantum Leap in Speed : NVMe Gen 4.0x4

In the world of NVMe interface architecture, a direct connection to the CPU is established through the PCIe interface, allowing for efficient memory-like access as data transactions exclusively follow the PCIe bus protocol. The Super High-Performance Gen 4.0x4 NVMe drive stands out with an impressive read/write speed of 7,880MB per second, disrupting the conventional use of hard drives in SATA and SAS. This innovative technology ensures not only a superior bandwidth for data transmission but also guarantees optimal operational efficiency, especially when managing data-intensive tasks crucial for key operations.

Intelligent Passive Cooled solution

With increased speed and performance comes a higher generation of heat which causing higher chance of malfunction. In environments where active cooling is impractical, conventional removable SSD cages struggle to cope with the elevated heat levels, particularly from ultra-high capacity NVMe U.2 SSDs (up to 16TB per pack). Enter the 7StarLake NVMe U.2 Conduction Cooled Pack—a game-changer. Each U.2 SSD is equipped with its own standalone, and anti-corrosive aluminum cooling pack. This innovative solution efficiently redirects heat flow through thermal pads and heatsinks. The distinctive and high-efficiency modular design of the 7StarLake solution distinguishes it from traditional storage alternatives, ensuring exceptional thermal solution even in the face of the intense heat generated by demanding processing tasks.

AV800 tray

 

 

Steadfast Against Extreme Pressures

In addition to its outstanding thermal solution, 7StarLake ultra-compact Conduction Cooled Pack features a meticulously designed enclosure specifically tailored to accommodate U.2 SSD dimensions. This meticulous design guarantees that U.2 SSDs remain immune to vibrations and shocks, making them resilient in diverse scenarios such as airborne missions, naval operations, or ground deployments. The precision-engineered enclosure adds an extra layer of protection, ensuring enhanced resilience and improved durability of the U.2 SSDs even in the most demanding and challenging environments.

 

 

 
MIL-STD Environment
  • Operating Temperature High: 50°C, MIL-STD-810G, Method 501.5, Procedure I
  • Operating Temp Low: 0°C, MIL-STD-810G, Method 502.5, Procedure I
  • Non-Operating Temperature High: 70°C, MIL-STD-810G, Method 501.5, Procedure II
  • Non-Operating Temperature Low: -40°C, MIL-STD-810G, Method 502.5, Procedure II
  • Operating Altitude: Up to 15,000 ft., MIL-STD-810G, Method 500.5
  • Non-Operating Altitude: Up to 45,000 ft., MIL-STD-810G, Method 500.5
  • Humidity: MIL-STD-810G, Method 507.5, Procedure Ib (Natural Cycle B3)
  • Shock: MIL-STD-810G, Method 516.6, 30 g’s, Saw-tooth, 11ms & MIL-DTL-901E, Grade A, Class II; Type B
  • Vibration: MIL-STD-167, Type I, Deck Mounted Equipment
  • EMI/EMC: MIL-STD-461F, RE101, RE102 (Shipboard Level 1), RS103, CE101, CE102, CS101, CS114, CS116
  • Airborne Noise: MIL-STD-740-1 compliance: 43.7dBA (Idle), 52.5dBA (50%), 54.6dBA (80%).
NVIDIA CONNECTX-6 DX
connectX-6-dx

 

NVIDIA® ConnectX®-6 DX is a highly secure and advanced smart network interface card (SmartNIC) that accelerates mission-critical cloud and data center applications, including security, virtualization, SDN/NFV, big data, machine learning, and storage. ConnectX-6 DX provides up to two ports of 100Gb/s or a single port of 200Gb/s Ethernet connectivity and is powered by 50Gb/s (PAM4) or 25/10 Gb/s (NRZ) SerDes technology. ConnectX-6 DX features virtual switch (vSwitch) and virtual router (vRouter) hardware accelerations delivering orders-of-magnitude higher performance than softwarebased solutions. ConnectX-6 DX supports a choice of single-root I/O virtualization (SR-IOV) and VirtIO in hardware, enabling customers to best address their application needs. By offloading cloud networking workloads, ConnectX-6 DX frees up CPU cores for business applications while reducing total cost-of-ownership.

PRODUCT SPECIFICATIONS
Maximum total bandwidth 200Gb/s
Supported Ethernet speeds 10/25/40/50/100/ 200GbE
Number of network ports 1/2
Network interface technologies NRZ/PAM4
Host interface PCIe Gen4.0 x16, with NVIDIA Multi-Host™ technology
DPDK message rate Up to 215Mpps
Platform security Hardware root-oftrust and secure firmware update
Form factors PCIe HHHL, OCP2, OCP3.0 SFF
Network interfaces SFP+, QSFP+, DSFP

 

 

 

Appearance
Appearance-1
Appearance-2

System

Processor

Intel® Xeon® ICELAKE-D D-2775TE Processor, 20Cores, 40Threads, Base Frequency 3.1GHz

Memory type

256GB RDIMM ECC DDR4-2933 / 512G LRDIMM DDR4-2933 in 4 DIMM Slot

GPU

Graphics Card

NVidia® RTX A2000 2560 CUDA Cores 

Storage

HDD/SDD

2x SATA SSD

Hardware Secure Erase (AES) by option

Front I/O

X1

DC In connector

X2

2x USB3.0 Amphenol USB3FTV7AZNF312 connector

X3

2x 1GBase-T TV07RW-13-98S connector

X4

1x 1GBase-T TV07RW-13-98S connector

X5

1x 10GBase-T M20 RJ45 CAT6A connector (Options)

X6

1x 10GBase-T M20 RJ45 CAT6A connector (Options)

X7

4x 100G Fiber Ethernet Amphenol FSI MPO
VGA D-sub 15 connector with waterproof cap

Side I/O

SSD tray

2x Dual 2.5" HDD/SSD Easy Swap Tray

Power Button

1x Power Button with LED backlight

Power Requirement

Power Input

MIL-STD-461 EMI power supply, 18V~36V DC-IN (300W)
Applications
Applications Military Platforms Requiring Compliance to MIL-STD-810 Where Harsh Temperature, Shock, Vibration, Altitude, Dust and MIL-461 EMI Conditions.
Operating System
Operating System Windows 10 64Bit, Linux by request.
Physical
Dimension 405mm x 316mm x 195mm (W x L x H)
Weight 10KGS
Chassis Aluminum Alloy, Corrosion Resistant
Finish Anodic aluminum oxide (Color Iron gray)
Cooling Intelligent Active Cooled
Environmental  
MIL-STD-810G Test

Method 500.5, Procedures I and II (Altitude, Operation):

12,192M, (40,000 ft) for the initial cabin altitude (18.8Kpa or 2.73 Psia)

Method 500.5, Procedures III and IV (Altitude, Non-Operation):

15,240, (50,000 ft) for the initial cabin altitude (14.9Kpa or 2.16 Psia)

Method 501.5, Procedure I (Storage/High Temperature)

Method 501.5, Procedure II (Operation/High Temperature)

Method 502.5, Procedure I (Storage/Low Temperature)

Method 502.5, Procedure II (Operation/Low Temperature)

Method 503.5, Procedure I (Temperature shock)

Method 507.5, Procedure II (Temperature & Humidity)

Method 514.6, Vibration Category 24/Non-Operating (Category 20 & 24,Vibration)

Method 514.6, Vibration Category 20/Operating (Category 20 & 24,Vibration)

Method 516.6, Shock-Procedure V Non-Operating (Mechanical Shock)

Method 516.6, Shock-Procedure I Operating (Mechanical Shock)
Reliability

Conduction Cooling.

Designed & Manufactured using ISO 9001 Certified Quality Program.
MIL-STD-461

CE102 basic curve, 10kHz - 30 MHz

RE102-4, (1.5 MHz) -30 MHz - 5 GHz

RS103, 200 MHz - 3.2 GHz, 50 V/m equal for all frequencies

EN 61000-4-2: Air discharge: 8 kV, Contact discharge: 6kV

EN 61000-4-3: 10V/m

EN 61000-4-4: Signal and DC-Net: 1 kV

EN 61000-4-5: Leads vs. ground potential 1kV, Signal und DC-Net: 0.5 kV

CE and FCC
MIL-STD-1275

Steady State – 20V~33V,

Surge Low – 18V/500ms,

Surge High – 100V/500ms

Emitted spikes

Injected Voltage surges

Emitted voltage surges

Voltage ripple (2V)

Voltage spikes

Starting Operation

Reverse polarity
Operating Temp. -20°C to +60°C
Storage Temp. -40°C to +85°C
Relative Humidity 5% to 95%, non-condensing.