How to Install and Configure JetPack SDK

Depending on your Jetson device, there are multiple ways to install JetPack.

SD Card Image

For NVIDIA Jetson Orin Nano developer kit users, the simplest JetPack installation method is to follow the steps at the Getting Started web page to download and write an image to your microSD card, then use it to boot the developer kit.

Note

If you were previously using JetPack 5 on your Jetson Orin Nano Developer Kit and want to upgrade to JetPack 6 without using any host machine, please follow the instructions provided in Jetson Linux documentation.

NVIDIA SDK Manager

NVIDIA SDK Manager supports JetPack installation on these Jetson products:

  • NVIDIA Jetson AGX Orin Developer Kit

  • NVIDIA Jetson Orin Nano Developer Kit

  • NVIDIA Jetson AGX Orin 32GB module on a Jetson AGX Orin Developer Kit carrier board

  • NVIDIA Jetson AGX Xavier series modules on a Jetson AGX Xavier Developer Kit carrier board

  • NVIDIA Jetson Xavier NX modules on a Jetson Xavier NX Developer Kit carrier board

A Linux host computer running Ubuntu Linux x64 version 20.04 or 18.04 is required to run SDK Manager. Detailed instructions can be found here:

Package Management Tool

NVIDIA offers JetPack components as Debian packages. The Debian package management server can be accessed at https://repo.download.nvidia.com/jetson/.

Install JetPack Components on Jetson Linux

This step assumes your Jetson developer kit has been flashed with and is running L4T 36.3. The following commands will install all other JetPack components that correspond to your version of Jetson Linux L4T:

sudo apt update
sudo apt install nvidia-jetpack

To view individual Debian packages which are part of nvidia-jetpack metapackage, enter the command:

sudo apt show nvidia-jetpack

Refer to the NVIDIA Jetson Linux Developer Guide for details about L4T specific Debian packages.

If disk space is limited, use these commands:

sudo apt update
apt depends nvidia-jetpack | awk '{print $2}' | xargs -I {} sudo apt install -y {}

Upgrade JetPack

To upgrade from previous JetPack 6.x releases, first edit etc/apt/sources.list.d/nvidia-l4t-apt-source.list to point to the 36.3 repo (just change the version to r36.3 in both lines). Next, use the following commands, then physically reboot the system:

sudo apt update
sudo apt dist-upgrade
sudo apt install --fix-broken -o Dpkg::Options::="--force-overwrite"

If you see the below error:

ERROR. Procedure for A_kernel-dtb update FAILED.

Use the following three commands, then re-attempt the upgrade. Do not reboot the device after running the below command, before re-attempting the upgrade.

sudo parted /dev/mmcblk0 -s rm 4
sudo parted /dev/mmcblk0 -s resizepart 3 67.9MB
sudo parted /dev/mmcblk0 mkpart A_reserved_on_user 67.9MB 101MB

Select Ignore when prompted.

Upgradable Compute Stack

With introduction of upgradable compute stack supported from JetPack 6.0 production release, developers can upgrade to CUDA-X compute stack form latest JetPack releases on previous JetPack without upgrading the Jetson Linux BSP.

To use CUDA, CUDNN, Tensorrt, DLA and VPI from this JetPack 6.0 production release on JetPack 6.0 Developer Preview release, please follow the steps below:

On Jetson Orin with JetPack 6.0 DP installed:

  1. Add R36.3 repo.

  2. Remove JetPack meta packages if installed.

    sudo apt-get remove nvidia-jetpack nvidia-jetpack-runtime nvidia-jetpack-dev

  3. Install DLA compiler package.

    sudo apt-get install nvidia-l4t-dla-compiler

  4. Install JetPack runtime packages.

    sudo apt-get install nvidia-cuda nvidia-tensorrt nvidia-cudnn8 nvidia-cupva nvidia-opencv nvidia-container nvidia-vpi

  5. Install JetPack development packages if needed.

    sudo apt-get install nvidia-nsight-sys nvidia-cuda-dev nvidia-tensorrt-dev nvidia-vpi-dev nvidia-opencv-dev nvidia-cudnn8-dev

  6. Remove the R36.3 repo to avoid installing nvidia-l4t-gstreamer accidentally later.

    Note

    Don’t use apt-get install nvidia-jetpack or apt-get upgrade because that will upgrade nvidia-l4t-gstreamer, too.

List of Debian Packages

The following is a list of Debian update packages for JetPack components for Jetson devices as of the time of JetPack 6.0. Please refer to the Jetson Linux Developer Guide for the list of Jetson Linux Debian packages.

Jetson Linux Debian Packages

Component Group

Packages

Packages (cont’d.)

CUDA Toolkit for L4T

cuda

cuda-12-2

cuda-cccl-12-2

cuda-command-line-tools-12-2

cuda-compat-12-2

cuda-compiler-12-2

cuda-crt-12-2

cuda-cudart-12-2

cuda-cudart-dev-12-2

cuda-cuobjdump-12-2

cuda-cupti-12-2

cuda-cupti-dev-12-2

cuda-cuxxfilt-12-2

cuda-documentation-12-2

cuda-driver-dev-12-2

cuda-gdb-12-2

cuda-gdb-src-12-2

cuda-libraries-12-2

cuda-libraries-dev-12-2

cuda-minimal-build-12-2

cuda-nsight-compute-12-2

cuda-nvcc-12-2

cuda-nvdisasm-12-2

cuda-nvml-dev-12-2

cuda-nvprune-12-2

cuda-nvrtc-12-2

cuda-nvrtc-dev-12-2

cuda-nvtx-12-2

cuda-nvvm-12-2

cuda-profiler-api-12-2

cuda-runtime-12-2

cuda-sanitizer-12-2

cuda-toolkit

cuda-toolkit-12

cuda-toolkit-12-2

cuda-toolkit-12-2-config-common

cuda-toolkit-12-config-common

cuda-toolkit-config-common

cuda-tools-12-2

cuda-visual-tools-12-2

gds-tools-12-2

libcublas-12-2

libcublas-dev-12-2

libcudla-12-2

libcudla-dev-12-2

libcufft-12-2

libcufft-dev-12-2

libcufile-12-2

libcufile-dev-12-2

libcurand-12-2

libcurand-dev-12-2

libcusolver-12-2

libcusolver-dev-12-2

libcusparse-12-2

libcusparse-dev-12-2

libnpp-12-2

libnpp-dev-12-2

libnvjitlink-12-2

libnvjitlink-dev-12-2

libnvjpeg-12-2

libnvjpeg-dev-12-2

nsight-compute-2023.2.2

nvidia-l4t-cudadebuggingsupport

nvidia-fs

nvidia-fs-dkms

nvidia-gds

nvidia-gds-12-2

cuDNN

libcudnn8

libcudnn8-dev

libcudnn8-samples

TensorRT

graphsurgeon-tf

libnvinfer-bin

libnvinfer-dev

libnvinfer-dispatch-dev

libnvinfer-dispatch8

libnvinfer-headers-dev

libnvinfer-headers-plugin-dev

libnvinfer-lean-dev

libnvinfer-lean8

libnvinfer-plugin-dev

libnvinfer-plugin8

libnvinfer-samples

libnvinfer8

libnvinfer-vc-plugin-dev

libnvinfer-vc-plugin8

libnvinfer8

libnvonnxparsers-dev

libnvonnxparsers8

libnvparsers-dev

libnvparsers8

onnx-graphsurgeon

python3-libnvinfer

python3-libnvinfer-dev

python3-libnvinfer-dispatch

python3-libnvinfer-lean

tensorrt

tensorrt-dev

tensorrt-libs

uff-converter-tf

OpenCV

libopencv

libopencv-dev

libopencv-python

libopencv-samples

opencv-licenses

opencv-samples-data

VPI

libnvvpi3

python3.10-vpi3

vpi3-dev

vpi3-python-src

vpi3-samples

NVIDIA container runtime with Docker integration

libnvidia-container-tools

libnvidia-container1

nvidia-container-toolkit

nvidia-container-toolkit-base

DLA Compiler

nvidia-l4t-dla-compiler

Gstreamer

nvidia-l4t-gstreamer

Multimedia API

nvidia-l4t-jetson-multimedia-api

CUPVA

cupva-2.3-l4t

pva-allow-1

The following is a list of meta-packages that are available to easily install on Jetson. At a higher level, the nvidia-jetpack meta-package includes nvidia-jetpack-runtime meta-package and nvidia-jetpack-dev meta-package. nvidia-jetpack-runtime includes runtime only parts of JetPack components and does not include samples, documentation, etc. Meanwhile, the nvidia-jetpack-dev meta-package includes everything required for development.

You can install either the higher level meta-packages using apt install, OR install individual component meta-packages depending on your requirements. These meta-packages can be installed either on top of Jetson Linux, or in a container running on Jetson Linux.

../_images/jetpack-metapackage.png

JetPack Debian Packages on Host

NVIDIA provides a group of Debian packages that add or update JetPack components on the host computer.

To prepare the host computer to install JetPack components, do the following steps:

  1. Enter the following command to install the public key of the x86_64 repository of the public APT server:

    $ sudo apt-key adv --fetch-key http://repo.download.nvidia.com/jetson/jetson-ota-public.asc

  2. Add the following x86_64 repository to the host system’s source list. - For an Ubuntu 20.04 host:

    deb http://repo.download.nvidia.com/jetson/x86_64/focal 36.3 main

    • For an Ubuntu 22.04 host:

    deb http://repo.download.nvidia.com/jetson/x86_64/jammy r36.3 main

  3. Enter the following command:

    $ sudo apt update

  4. Use apt to download and install the required packages.

$ sudo apt-get install cuda-toolkit-12-2 cuda-cross-aarch64-12-2 nvsci libnvvpi3 vpi3-dev vpi3-cross-aarch64-l4t python3.9-vpi3 vpi3-samples vpi3-python-src nsight-systems-2023.4.3 nsight-graphics-for-embeddedlinux-2023.3.0.0

The following table lists JetPack components that you can install with apt, and the packages that contain them.

Jetson Linux Debian Packages

Component Group

Packages

Packages (cont’d)

CUDA Toolkit for L4T

cuda

cuda-12-2

cuda-cccl-12-2

cuda-command-line-tools-12-2

cuda-compat-12-2

cuda-compiler-12-2

cuda-crt-12-2

cuda-cudart-12-2

cuda-cudart-dev-12-2

cuda-cuxxfilt-12-2

cuda-demo-suite-12-2

cuda-documentation-12-2

cuda-driver-dev-12-2

cuda-drivers

cuda-drivers-535

cuda-gdb-12-2

cuda-gdb-src-12-2

cuda-libraries-12-2

cuda-libraries-dev-12-2

cuda-minimal-build-12-2

cuda-nsight-12-2

cuda-nsight-compute-12-2

cuda-nsight-systems-12-2

cuda-nvcc-12-2

cuda-nvdisasm-12-2

cuda-nvml-dev-12-2

cuda-nvprof-12-2

cuda-nvprune-12-2

cuda-nvrtc-12-2

cuda-nvrtc-dev-12-2

cuda-nvtx-12-2

cuda-nvvm-12-2

cuda-nvvp-12-2

cuda-opencl-12-2

cuda-opencl-dev-12-2

cuda-profiler-api-12-2

cuda-runtime-12-2

cuda-sanitizer-12-2

cuda-toolkit

cuda-toolkit-12

cuda-toolkit-12-2

cuda-toolkit-12-2-config-common

cuda-toolkit-config-common

cuda-tools-12-2

cuda-visual-tools-12-2

gds-tools-12-2

libcublas-12-2

libcublas-dev-12-2

libcufft-12-2

libcufft-dev-12-2

libcufile-12-2

libcufile-dev-12-2

libcurand-12-2

libcurand-dev-12-2

libcusparse-12-2

libcusparse-dev-12-2

libnpp-12-2

libnpp-dev-12-2

libnvidia-cfg1-430

libnvidia-cfg1-535

libnvidia-common-430

libnvidia-common-535

libnvidia-compute-430

libnvidia-compute-535

libnvidia-decode-430

libnvidia-decode-535

libnvidia-encode-430

libnvidia-encode-535

libnvidia-extra-535

libnvidia-fbc1-430

libnvidia-fbc1-535

libnvidia-gl-430

libnvidia-gl-535

libnvjitlink-12-2

libnvjitlink-dev-12-2

libnvjpeg-12-2

libnvjpeg-dev-12-2

libxnvctrl-dev

libxnvctrl0

nsight-compute-2023.2.2

nsight-systems-2023.2.3

nvidia-compute-utils-430

nvidia-compute-utils-535

nvidia-dkms-430

nvidia-dkms-535

nvidia-driver-430

nvidia-driver-535

nvidia-fs

nvidia-fs-dkms

nvidia-gds

nvidia-gds-12-2

nvidia-headless-430

nvidia-headless-535

nvidia-headless-no-dkms-430

nvidia-headless-no-dkms-535

nvidia-kernel-common-430

nvidia-kernel-common-535

nvidia-kernel-source-430

nvidia-kernel-source-535

nvidia-modprobe

nvidia-settings

nvidia-utils-430

nvidia-utils-535

xserver-xorg-video-nvidia-430

xserver-xorg-video-nvidia-535

CUDA cross-compile package (host)

cuda-cccl-cross-aarch64-12-2

cuda-cross-aarch64

cuda-cross-aarch64-12-2

cuda-crt-cross-aarch64-12-2

cuda-cudart-cross-aarch64-12-2

cuda-cupti-cross-aarch64-12-2

cuda-driver-cross-aarch64-12-2

cuda-nsight-compute-addon-l4t-12-2

cuda-nvcc-cross-aarch64-12-2

cuda-nvml-cross-aarch64-12-2

cuda-nvrtc-cross-aarch64-12-2

cuda-profiler-api-cross-aarch64-12-2

libcublas-cross-aarch64-12-2

libcudla-cross-aarch64-12-2

libcufft-cross-aarch64-12-2

libcurand-cross-aarch64-12-2

libcusolver-cross-aarch64-12-2

libcusparse-cross-aarch64-12-2

libnpp-cross-aarch64-12-2

libnvjitlink-cross-aarch64-12-2

libnvjpeg-cross-aarch64-12-2

nsight-compute-addon-l4t-2023.2.2

Computer Vision – VPI (host)

libnvvpi3

vpi3-dev

vpi3-cross-aarch64-l4t

python3.10-vpi3

vpi3-samples

vpi3-python-src

Developer Tools

nsight-systems-2024.2.2

nvsci

nsight-graphics-for-embeddedlinux-2024.1.0