Product of the Week: Cincoze GP-3000 Industrial AI & Machine Vision Computer

June 07, 2021

Story

Size, weight, and power consumption make it difficult to get data center-class AI performance to industrial machine vision, in-vehicle computing, and intelligent security and surveillance (ISS) applications at the edge. Enter the 105 mm x 190 mm x 360 mm Cincoze GP-3000.

The Cincoze GP-3000 is a GPU edge computing system from Cincoze for AI industrial and machine vision. The embedded computer features 8th or 9th generation Intel Core i3/i5/i7 (up to six cores) or Xeon (up to eight cores) processors, respectively, workstation-grade processors based on the Coffee Lake architecture that pair with the Intel C246 chipset. More importantly for demanding AI workloads, however, the platform supports an add-on GPU Expansion Box (GEB) that accepts dual 250 W full-length graphics cards from NVIDIA or AMD while maintaining a total system power of 720 W.

The GP-3000 includes all of the I/O necessary for an industrial AI or machine vision deployment, including:

• Display: 1x VGA, 1x HDMI, 1x DisplayPort
• Connectivity: 5x GbE LANs, 6x USB 3.2, 2x RS-232/422/485, 2x Mini PCIe
• Storage: 2x DDR4-2666 SO-DIMMs (up to 64 GB), 2.5” Storage Bay, 1x M.2 Key M, 1x M.2 Key E

Meanwhile, the GEB is equipped with up to one PCIe x16 lanes, one PCIe x4, and one PCIe x1. An internal USB 2.0 is also present.

The Cincoze GP-3000 in Action

Working with the GP-3000 industrial AI and machine vision computer begins with selecting one of two GEB configurations: Either the GEB-3301 (1x PCIe 16, 1x PCIe x4) or the GEB-3601 (2 PCIe x8 lanes, 1 PCIe x4, 1 x PCIe x1). Pairing this with the base CP-3000 system creates a platform as compact as 209 mm x 195 mm x 370 mm.

Up to two high-end 328 mm GPU cards present in the scalable GEB expansion boxes can chew through AI workloads by communicating with the SO-DIMM memories or host Intel® processor on the GP-3000 over the fast PCIe interconnects. The processing parallelism of these GPUs can be augmented by additional PCIe add-on cards like frame grabbers or high-speed I/O cards, as well as additional GbE/10 GbE LAN, RS-232/422/485, or PoE, or Optical DIO modules supported by Cincoze’s combined multiple I/O interface. 

And, if a performance upgrade is ever required, users simply need to replace the GPU cards in the GEB instead of the entire system.

As mentioned, the total system power budget sits at 720W with a DC power input from 9 to 48V for ideal GPU computing. To cool this level of performance, both sides of the chassis contain an intelligent cooling system with isolated external smart fan kits.

For military, automotive, industrial, and railway operation and standards, the GP-3000 adheres to the MIL-STD-810G military standard, E-mark, and the EN50155 (EN 50121-3-2 only). Suitable for harsh environments, the product can withstand both extreme hot and cold, -40 to 70℃, temperature operation. The product is also resistant to shock, vibration, and electromagnetic radiation, with an adjustable GPU retainer that fixes the graphics card in place in high-vibration environments.

Getting Started with the Cincoze GP-3000

After selecting the GP-3000 plus GEB combination and a mounting option (desktop, wall, face-up, or 19” rack mounting supported), the platform is ready for use in a variety of edge AI applications.

For more information on the Cincoze GP-3000, visit their website or check out the resources below.

• Cincoze GP-3000 Product Page: www.cincoze.com/goods_info.php?id=348
• Cincoze GP-3000 Solution Brief: www.cincoze.com/events/Industrial-AI-Machine-Vision-Computer-GP-3000
• Cincoze GP-3000 Data Sheet: www.cincoze.com/data/files/202105/Datasheet_GP-3000-Series.pdf
• Contact: [email protected]