ADLINK Technology Inc, an edge computing company, has launched compact GPU-enabled deep learning acceleration platforms on the market with its latest DLAP x86 series. The DLAP x86 series targets the deployment of deep learning in volume, at the edge where data is generated and actions are taken. It is optimised to deliver AI performance in various industry applications by accelerating compute intensive, memory-hungry AI inferencing and learning tasks.

According to ADLINK, the DLAP x86 series features:

• Heterogeneous architecture for high performance - featuring Intel processors and NVIDIA Turing GPU architecture delivering higher GPU-accelerated computation than others and returning optimised performance per watt and per dollar.
  
• A compact size that starts at 3.2 liters; it is optimal within mobility devices or instruments where physical space is limited, such as mobile medical imaging equipment.
  
• Sustaining temperatures up to 50 degrees celsius or 240 watts of heat dissipation, strong vibration (up to 2 Grms) and shock protection (up to 30 Grms), for reliability in industrial, manufacturing and healthcare environments.

Delivering a mix of SWaP and AI performance in edge AI applications, the DLAP x86 operates in healthcare, manufacturing, transportation and other sectors. Examples of use include:

• Mobile medical imaging equipment: C-arm, endoscopy systems, surgical navigation systems
  
• Manufacturing operations: object recognition, robotic pick and place, quality inspection
  
• Edge AI servers for knowledge transfer: combining pre-trained AI models with local data sets

“The value of ADLINK’s DLAP series is the flexibility it provides for deep learning applications; architects can choose the optimal combination of CPU and GPU processors based on the demands of an application’s neural networks and AI inferencing speed, yielding a high performance per dollar,” said Zane Tsai, Director ADLINK’s Embedded Platforms and Modules Product Center.

ADLINK Technology Inc is an edge computing company including in robust boards, real-time data acquisition solutions, and application enablement for AIoT.