Published on

Introduction To Nvidia Jetson Nano

Authors

Nvidia Jetson Nano is a tiny yet powerful ARM based single board computer (SBC) module by Nvidia. It is powered by Nvidia's 128 core Maxwell GPU and quad-core 64-bit ARM processor.

Nvidia Jetson Nano 2GB Single Board Computer

The Jetson Nano comes in two flavors:

  1. Nvidia Jetson Nano 4GB (Having 4GB RAM)
  2. Nvidia Jetson Nano 2GB (Having 2GB RAM)

I've recently got hold of an Nvidia Jetson Nano 2GB platform, and hence I intend to write a set of articles related to it as I work around this hardware platform. In this way, I will be able to come back to them in the future to recollect everything I've learnt about this platform.

Nvidia Jetson Nano 2GB features

Some of the features of Nvidia Jetson Nano 2GB are as follows:

  1. Runs on a 64-bit quad-core ARM processor
  2. Integrated with 128-core Maxwell GPU
  3. 2GB LPDDR4 RAM
  4. One MIPI CSI-2 interface to connect to external camera module
  5. Powered by 5V 2.5A using USB-C connector
  6. MicroSD Card for flash memory to store OS and applications image.
  7. HDMI Display interface
  8. 3-USB interfaces
  9. Gigabit Eternet
  10. 40-pin header (Supports 3 I2C, 2 SPI, UART, I2S and GPIOs)

Nvidia Jetson Nano Software

The board has been primarily designed for Artificial Intelligence applications on the Edge side (i.e. running AI applications without needing to connect to cloud services for operations). As a result, primary focus has been on its GPU processors.

The camera interface also enables the board to be used for running computer vision and machine learning algorithms on the edge.

The official OS image provided by Nvidia for it's Jetson Nano board is built on Linux 4.9 kernel and has customized library support for Machine learning tools such as TensorFlow, OpenGL, OpenCV and Nvidia's own CUDA toolkit.

All in all, this is currently the platform of choice if you want to run any AI or ML algorithms on the Edge due to Nvidia having a strong hold on their GPU hardware and their developer friendly CUDA toolkit.