+27 (0)11 315 8316 info@coredeeplearning.ai
ASIC Logo Vector Fromat 2
Microchip Logo

AI at the edge: Implementing low-Power CNNs in FPGAs

CNNs are typically implemented in GPUs that consume up to 30 Watts, are relatively heavy and often require active cooling. What if you need less accuracy, but are limited in weight, Power-consumption and PCB-real-estate?

This workshop is demonstrating how to implementing low-Power CNNs in FPGAs. In this workshop we use Microchip’s PolarFire FPGAs that consume 50% less Power than comparable SRAM-FPGAs and provide package sizes as low as 11 mm x 11 mm.

If you never designed with FPGAs before we use a very simple example just to demonstrate the design flow.

You will walk out this course and understand exactly what steps are involved to implementing low-Power CNNs in FPGAs and how ASIC Design Services and Arrow can help you to speed-up your development process.


  • 08:55 a.m. – 09:00 a.m.  Welcome
  • 09:00 a.m. – 10:30 a.m.  FPGA Primer: implementing a simple design using Libero SoC and SMF2000 (Hands-On)
  • 10:30 a.m. – 10:45 a.m.  Networking Break
  • 10:45 a.m. – 12:15 a.m.  AI in a nutshell
  • 12:15 a.m. – 01:00 p.m.  Lunch Break
  • 01:00 p.m. – 02:30 p.m.  Core Deep Learning Development Framework & Demos
  • 02:30 p.m. – 02:45 p.m.  Networking Break
  • 02:45 p.m. – 04:15 p.m.  Digit recognition on SMF2000 using Laptop Camera + UART to transmit digits to CNN (Hands-On)
  • 04:15 p.m. – open end    Open discussion

Free of Charge

Requires SMF2000

5th May 2020

Virtual event based on Teams

CDL & Arrow AI on "FPGA Workshop" Registration

CNNRNNOther (please specify)


We are glad that you want to register. Please fill our short form and you will receive your demo link via email.

I accept the Terms of Use & Privacy Policy