Parallel Processing for Embedded Systems

image courtesy GreenArrays, Inc.
GreenArrays have announced the GA144, a low-energy processor with 144 cores. It's not cheap, as the minimum order is for 10 $20 chips, and the development board is $450 dollars, but it's of great interest to me; next year I will need some real processing power for a neural-network robot I am working on, and this looks as if it would be ideal.

Robotics enthusiasts will probably be familiar with the Parallax propeller, an inexpensive multi-cored processor that is desgined for hobbyist use. The propeller is an affordable introduction to parallel processing, but the GA144 offers a couple of orders of magnitude more processing power and is very energy-efficient.

The chairman of GreenArrays is Chuck Moore, the inventor of Forth. The GA144 supports arrayForth, a powerful language for parallel computing.


  1. I have no knowledge in programming for embedded device but in comparison to Intel chip, GA need to route the message manually to activate several computers, how much overhead will this create?

    What are the performance result if this is use as an embedded server?

  2. Performance depends on the application and the application software. For my Neural Network application, each processor will be performing similar (but not identical) operations on local data. I expect a 100x improvement in processing speed over a single processor.


Post a Comment

Popular posts from this blog

Controlling a Raspberry Pi Pico remotely using PySerial

Five steps to connect Jetson Nano and Arduino

Raspberry Pi Pico project 2 - MCP3008