Contact
Support

Industrial AI Accelerator

Product overview

The Industrial AI accelerator is using the nearbAI FPGA IP core from easics, and REFLEX CES Achilles Arria 10® SoC module (Lite, Turbo or Indus version).

 

This Industrial AI accelerator will help you deploy your AI model inside your embedded application. 

Your model and the constraints of your application like speed, latency, power consumption and cost defines the right FPGA SoM as inference engine. The easics software tools will convert the model and the weights into an FPGA build file that is ready to deploy on the chosen FPGA hardware.

 

This solution demonstrates AI Model recognition :

  • Real-time FPGA-based solution
  • Demo running Yolo v3 DNN
  • Recognizes up to 80 different objects

Benefits & features

Benefits and features of the nearbAI as AI accelerator:

  • nearbAI is optimized for the best performance on the FPGA of your choice
  • easics' software offers a flexible approach to program the FPGA and a fast-time to market.
  • The FPGA logic can be shaped to match any neural network architecture.
  • High performance per Watt and low latency make it suitable for real-time embedded applications.
  • Performance, cost and power will define the nearbAI IP.
  • Future proof and scalable solution as the FPGA architecture can be re-configured for future neural networks.
  • The deep learning core can be easily integrated within the top level of your application

 

Benefits and features of the Achilles SoM as industrial AI engine:

  • FPGAs have product lifecycles of 15 years.
  • Industrial temperature ranges: -45 to 85 degrees C
  • High performance per Watt and low latency make it suitable for real-time embedded applications.
  • Low memory footprint because of fixed point (16 bit, 12 bit, 8 bit, 6 bit) data types
  • The FPGA logic can be shaped to match any neural network architecture.
  • Performance, cost and power will define the FPGA of choice.
  • Future proof and scalable solution as the FPGA architecture can be re-configured for future neural networks.
  • Easics' framework offers a flexible approach to program the FPGA and a fast-time to market.
  • The deep learning core can be easily integrated with other CPU’s, vision functionality and connectivity.

Diagrams

Applications

Supported sensors:

  • visual RGB
  • infrared: thermal IR, near-IR
  • X-ray
  • hyperspectral
  • Time-of-Flight (ToF)
  • 3D stereo
  • LiDAR, radar
  • ultrasound
  • non-image sensors
  • multiple sensors, sensor fusion

Performances

 

Network model  Input image resolution FPGA FPS
Resnet 50 224x224 ARRIA 10 GX 480 55.6
YoloV3 224x224 ARRIA 10 GX 480 25.9

Keep me informed

If you want to know more about REFLEX CES, sign up for our newsletter to be updated on our initiatives, sectorial news and upcoming events.

Reflex CES Reflex CES