Jetson TX2

The Jetson TX2 family of devices are single-board computers from the NVIDIA corporation. According to NVIDIA, “Jetson TX2 is the fastest, most power-efficient embedded AI computing device”. Each of the Jetson TX2 boards has an NVIDIA GPU and an ARMv8-A CPU. These boards all support the Ubuntu Linux operating system.

Devices similar to the NVIDIA Jetson TX2 may be able to run Xnor Bundles built for Jetson TX2 if their processor and library support is equivalent. However, such platforms are not officially supported by Xnor.

Language-independent setup

NVIDIA maintains instructions for setting up your Jetson TX2 here: Getting-Started-Jetson.

Xnor Bundles have no dependencies of their own and can be used in your C or Python application without any additional setup. However, the Xnor developer SDK needs a few libraries and tools to be present in order to run the sample applications. Instructions are available for installing these for C and Python development below.

Setting up for C development

Ubuntu 16.04 and later come with GCC installed by default, so no further installation is necessary. You should be able to download the SDK to your Jetson TX2 and build the samples in xnor-sdk-jetson-tx2/samples/c as follows:

cd ~/Downloads/xnor-sdk-jetson-tx2/samples/c
make

Next Steps for C development

Setting up for Python development

Ubuntu 16.04 comes with Python 3.5.1 installed by default, but you will need to install some packages on the Jetson TX2 to run the Xnor developer SDK samples. Download the SDK and run:

cd ~/Downloads/xnor-sdk-jetson-tx2
python3 -m pip install -r samples/python/requirements.txt

This will install all the dependencies necessary to run the samples in samples/python.

Next Steps for Python development