常见问题

NVIDIA JetPack SDK

2022-04-27 14:38:09 admin 425

1. How to Install JetPack

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


1.1. SD Card Image

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

  • Getting Started with Jetson Nano Developer Kit

  • Getting Started with Jetson Nano 2GB Developer Kit

  • Getting Started with Jetson Xavier NX Developer Kit




1.2. NVIDIA SDK Manager

NVIDIA SDK Manager supports JetPack installation on these Jetson products:

  • 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

  • NVIDIA Jetson Nano module on a Jetson Nano Developer Kit carrier board

  • NVIDIA Jetson TX2 series modules on a Jetson TX2 Developer Kit carrier board

  • NVIDIA Jetson TX1 on a Jetson TX1 or TX2 Developer Kit carrier board

A Linux host computer running Ubuntu Linux x64 version 18.04 or 16.04 is required to run SDK Manager. Detailed instructions can be found here:

https://docs.nvidia.com/sdk-manager/index.html




1.3. Package Management Tool

NVIDIA offers JetPack components as Debian packages.


1.3.1. Install JetPack

Assuming your Jetson developer kit has been flashed with and is running L4T 32.3.1 or higher, the following commands will install all other JetPack components that correspond to your version of 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 (for example, when using a 16GB microSD card with a Jetson Nano or Jetson Xavier NX developer kit), use these commands:

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




1.3.2. Upgrade JetPack

Starting with JetPack 4.4, upgrading to the next JetPack release can be achieved using a package management tool like apt. Follow the below steps to perform the upgrade:

  1. First, upgrade L4T.

    Please refer to the section "To upgrade to a new point release" in the NVIDIA Jetson Linux Developer Guide if upgrading to a new point release; for example, upgrading from JetPack 4.4 to JetPack 4.4.1.

    Please refer to the section "To upgrade to a new minor release" in the NVIDIA Jetson Linux Developer Guide if upgrading to a new minor release; for example, upgrading from JetPack 4.5 to JetPack 4.6.

  2. Next, upgrade the other JetPack components.

    sudo apt install nvidia-jetpack

If disk space is limited (for example, when using a 16GB microSD card with a Jetson Nano or Jetson Xavier NX developer kit), follow these steps:

  1. Remove all JetPack compute components.

    If you are running JetPack 4.4, use the following command:

    sudo apt autoremove --purge nvidia-container-csv-cuda libopencv-python libvisionworks-sfm-dev libvisionworks-dev libvisionworks-samples libnvparsers6 libcudnn7-doc libcudnn7-dev libnvinfer-samples libnvinfer-bin nvidia-container-csv-cudnn libvisionworks-tracking-dev vpi-samples tensorrt libopencv libnvinfer-doc libnvparsers-dev libcudnn7 libnvidia-container0 cuda-toolkit-10-0 nvidia-container-csv-visionworks graphsurgeon-tf libopencv-samples python-libnvinfer-dev libnvinfer-plugin-dev libnvinfer-plugin6 nvidia-container-toolkit libnvinfer-dev libvisionworks libopencv-dev nvidia-l4t-jetson-multimedia-api vpi-dev vpi python3-libnvinfer python3-libnvinfer-dev opencv-licenses nvidia-container-csv-tensorrt libnvinfer6 libnvonnxparsers-dev libnvonnxparsers6 uff-converter-tf nvidia-docker2 libvisionworks-sfm libnvidia-container-tools nvidia-container-runtime python-libnvinfer libvisionworks-tracking

    If you are running JetPack 4.5, use the following command:

    sudo apt autoremove --purge nvidia-container-csv-cuda libopencv-python libvisionworks-sfm-dev libvisionworks-dev libvisionworks-samples libnvparsers7 libcudnn8-doc libcudnn8-dev libnvinfer-samples libnvinfer-bin nvidia-container-csv-cudnn libvisionworks-tracking-dev vpi1-samples tensorrt libopencv libnvinfer-doc libnvparsers-dev libcudnn8 libnvidia-container0 cuda-toolkit-10-2 nvidia-container-csv-visionworks graphsurgeon-tf libopencv-samples python-libnvinfer-dev libnvinfer-plugin-dev libnvinfer-plugin7 nvidia-container-toolkit libnvinfer-dev libvisionworks libopencv-dev nvidia-l4t-jetson-multimedia-api vpi1-dev vpi1-demos python3-libnvinfer python3-libnvinfer-dev opencv-licenses nvidia-container-csv-tensorrt libnvinfer7 libnvonnxparsers-dev libnvonnxparsers7 uff-converter-tf nvidia-docker2 libvisionworks-sfm libnvidia-container-tools nvidia-container-runtime python-libnvinfer libvisionworks-tracking
  2. If the current version of JetPack was installed using SDK Manager, remove the local repo. (The example shown below assumes you will upgrade from JetPack 4.3.)

    sudo apt purge cuda-repo-l4t-10-0-local-10.0.326 libvisionworks-repo libvisionworks-sfm-repo libvisionworks-tracking-repo
  3. Free up additional disk space.

    sudo apt clean
  4. Upgrade L4T by referring to the OTA section of the NVIDIA Jetson Linux Developer Guide.

  5. Install the new JetPack components.

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




1.4. List of JetPack OTA Packages

Following is a list of JetPack OTA update packages for Jetson devices as of the time of JetPack 4.6. Please refer to the Jetson Linux Developer Guide for the list of L4T OTA packages.

Component GroupPackages

CUDA Toolkit for L4T

cuda-command-line-tools-10-2

cuda-compiler-10-2

cuda-cudart-10-2

cuda-cudart-dev-10-2

cuda-cuobjdump-10-2

cuda-cupti-10-2

cuda-cupti-dev-10-2

cuda-documentation-10-2

cuda-driver-dev-10-2

cuda-gdb-10-2

cuda-gdb-src-10-2

cuda-libraries-10-2

cuda-libraries-dev-10-2

cuda-memcheck-10-2

cuda-minimal-build-10-2

cuda-nvcc-10-2

cuda-nvdisasm-10-2

cuda-nvgraph-10-2

cuda-nvgraph-dev-10-2

cuda-nvml-dev-10-2

cuda-nvprof-10-2

cuda-nvprune-10-2

cuda-nvrtc-10-2

cuda-nvrtc-dev-10-2

cuda-nvtx-10-2

cuda-samples-10-2

cuda-toolkit-10-2

cuda-tools-10-2

cuda-visual-tools-10-2

libcublas-dev

libcublas10

libcufft-10-2

libcufft-dev-10-2

libcurand-10-2

libcurand-dev-10-2

libcusolver-10-2

libcusolver-dev-10-2

libcusparse-10-2

libcusparse-dev-10-2

libnpp-10-2

libnpp-dev-10-2

cuDNN

libcudnn8

libcudnn8-dev

libcudnn8-samples

TensorRT

libnvinfer8

libnvinfer-dev

libnvinfer-plugin8

libnvinfer-plugin-dev

libnvonnxparsers8

libnvonnxparsers-dev

libnvparsers8

libnvparsers-dev

libnvinfer-bin

libnvinfer-doc

libnvinfer-samples

tensorrt

python3-libnvinfer

python3-libnvinfer-dev

graphsurgeon-tf

uff-converter-tf

OpenCV

libopencv

libopencv-dev

opencv-licenses

libopencv-python

libopencv-samples

VisionWorks

libvisionworks

libvisionworks-dev

libvisionworks-samples

libvisionworks-sfm

libvisionworks-sfm-dev

libvisionworks-tracking

libvisionworks-tracking-dev  

VPI

libnvvpi1

vpi1-dev

python-vpi1

python3-vpi1

vpi1-samples

vpi1-demos

NVIDIA container runtime with Docker integration

nvidia-container-csv-cuda

nvidia-container-csv-cudnn

nvidia-container-csv-tensorrt

nvidia-container-csv-opencv

nvidia-container-csv-visionworks  

libnvidia-container0

libnvidia-container-tools

nvidia-container-toolkit

nvidia-container-runtime

nvidia-docker2

Multimedia API

nvidia-l4t-jetson-multimedia-api




1.5. 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 16.04 host:

      deb http://repo.download.nvidia.com/jetson/x86_64 xenial r32.4
    • For an Ubuntu 18.04 host:

      deb http://repo.download.nvidia.com/jetson/x86_64 bionic r32.4
  3. Enter the following command:

    $ sudo apt update
  4. Use apt to download and install the required packages.

    $ sudo apt-get install libnvvpi1 vpi1-dev vpi1-cross-aarch64-l4t python-vpi1 python3-vpi1 vpi1-samples libvisionworks-samples libvisionworks-dev libvisionworks libvisionworks-sfm-dev libvisionworks-sfm libvisionworks-tracking libvisionworks-tracking-dev cuda-toolkit-10-2 cuda-cross-aarch64-10-2 nsight-systems-2021.2.3 nsight-graphics-for-embeddedlinux-2021.2.1


首页
产品
案例
联系