Software installation

Installation of NVIDIA Cuda accellerated Software

The EdgeKit will be sent out without any CUDA related software, since it need around 5GB of disk-space. Install all needed Libraries for interacting with the GPU:

sudo apt update
sudo apt install nvidia-jetpack

You could verify the installed libraries with the command jetson_release. If the programm isn't installed you have to install the package "jetson-stats" via pip3: pip3 install -U jetson-stats

The output should look like the following, when nothing is installed:

root@edgekit-desktop:/home/edgekit# jetson_release
Software part of jetson-stats 4.2.1 - (c) 2023, Raffaello Bonghi
Model: EdgeKit nano by PCB Arts - Jetpack 4.6.3 [L4T 32.7.3]
NV Power Mode[0]: MAXN
Serial Number: [XXX Show with: jetson_release -s XXX]
Hardware:
 - P-Number: p3448-0002
 - Module: NVIDIA Jetson Nano module (16Gb eMMC)
Platform:
 - Distribution: Ubuntu 18.04 Bionic Beaver
 - Release: 4.9.299-tegra
jtop:
 - Version: 4.2.1
 - Service: Active
Libraries:
 - CUDA: Not installed
 - cuDNN: Not installed
 - TensorRT: Not installed
 - VPI: Not installed
 - Vulkan: 1.2.70
 - OpenCV: Not installed

After running the nvidia-jetpack installation you should have version numbers behind all Libraries.

VSCodium

For installing VSCodium on an EdgeKit just go to the download-page (via EdgeKit GUI or CLI): https://github.com/VSCodium/vscodium/releases and choose the latest deb package for arm64 plattforms. Download it via GUI or via the CLI command wget.

Install the VSCodium on the EdgeKit with sudo dpkg -i codium_1.80.1.23194_arm64.deb.

Start VSCodium in a shell with the command: codium In case the GUI does not open run: codium --no-sandbox

Deepstream

In this catalog you will find the latest deepstream installation for NVIDIA Jetson. Please download the arm64.deb package: https://catalog.ngc.nvidia.com/orgs/nvidia/resources/deepstream

Installation

sudo apt-get install ./deepstream-6.3_6.3.0-1_arm64.deb

Tensorflow

Install TensorFlow and its dependencies by following the instructions provided here. Please do not using the sudo command to install packages within your (virtual) environment.

Verify installation with:

$ python3
>>> import tensorflow

This command should execute without error.

To test the GPU utilization you can use this script test script and verify the usage by the jtop command.