Menu

Nvidia GPU technology for AI/ML and video processing at the edge

UAV and CCTV video streams provide increased security in many locations today.

With VoyagerGPU, you can analyse video in real time for inference applications but also train your AI/ML models at the edge of the network.

Additionally VoyagerGPU can transcode up to 16 HD video streams for transmission over a low-bandwidth network

Combined Xeon D CPU and Nvidia GPU spanning two Voyager modules

 

Figure 1: VoyagerGPU Front View Figure 2: VoyagerGPU Rear View

Key Features:

  • GPU options include
    • NVIDIA Turing-based GPU options:
      • Quadro T1000
      • Quadro RTX 3000
      • Quadro RTX 5000
  • CPU options include
    • Xeon D 8, 12, 16 cores
    • Atom Denverton 8-core
  • Storage is provided by dual SATA 2.5” bay plus VIK+ for operating system
  • 10 GB networking capabilities

Specifications

Physical Specifications

■ 7.9” L x 7.4” W x 4.2” H (201 x 188 x 107 mm)

Electrical Specifications

■ 12 VDC Input
■ Power consumption:
■ NVIDIA Quadro T1000: 130 W
■ NVIDIA Quadro RTX3000: 170 W
■ NVIDIA Quadro RTX5000: 200 W

Operating Temperature

■ 0°C to 50°C

Construction

■ Aluminum chassis

GPU Options

■ NVIDIA Turing-based GPU options:
■ Quadro T1000 with 768 CUDA Cores
■ Quadro RTX 3000 with 2,304 CUDA Cores, 36 RT Cores & 288 Tensor Cores
■ Quadro RTX 5000 with 3,072 CUDA Cores, 48 RT Cores & 384 Tensor Cores

Ports

■ 4 x HDMI ports
■ 2 x 10 Gb SFP+ ports
■ 2 x 1 Gb Ethernet ports
■ 2 x USB 3.0 ports
■ 1 x VGA port
■ 1 x Voyager Ignition Key (VIK+) port
■ 1 x Console Management port
■ 1 x Dual SATA 2.5” slot

Compliance

Designed to:
■ MIl-STD-810
■ MIL-STD-461
■ FCC CFR 47 Part 15 Subpart B Class A
■ RoHS Directive
■ REACH