Reference document: (official) https://developer.nvidia.com/embedded/learn/get-started-jetson-agx-orin-devkit
- Check your L4T version first to see if you have a unit flashed with older version of the BSP.
cat /etc/nv_tegra_release
You may get something like this, # R34 (release), REVISION: 1.0, GCID: 30102743, BOARD: t186ref, EABI: aarch64, DATE: Wed Apr 6 19:11:41 UTC 2022, and this shows that you have L4T for JetPack 5.0 Developer Preview.
- My version number is R35.1, which is different from the official website
- If you have an earlier version of L4T, issue the following command to manually put the apt repository entries using commands below.
- Open it
/etc/apt/sources.list.d/nvidia-l4t-apt-source.listand find that the version is correct, that is, R35.1, and R35.1 does not need to execute the content of the official website. In other words, the version numbers of deb https://repo.download.nvidia.com/jetson/common r35.1 main and deb https://repo.download.nvidia.com/jetson/t234 r35.1 main should be It is the same as the version number in step 1, if it is the same, it is not needed.
sudo bash -c 'echo "deb https://repo.download.nvidia.com/jetson/common r34.1 main" >> /etc/apt/sources.list.d/nvidia-l4t-apt-source.list'
sudo bash -c 'echo "deb https://repo.download.nvidia.com/jetson/t234 r34.1 main" >> /etc/apt/sources.list.d/nvidia-l4t-apt-source.list'
If you see R34 (release), REVISION: 1.0 or newer, then your apt sources lists are already up to date and you can proceed
- Issue the following commands to install JetPack components.
sudo apt update
sudo apt dist-upgrade
sudo reboot
sudo apt install nvidia-jetpack
- If you specify the first one to encounter the following situation, you need to execute the following:
swift@ubuntu:~/Desktop$ sudo apt update
[sudo] password for swift:
Reading package lists... Done
E: Could not get lock /var/lib/apt/lists/lock. It is held by process 10336 (apt-get)
N: Be aware that removing the lock file is not a solution and may break your system.
E: Unable to lock directory /var/lib/apt/lists/
- solution:
sudo rm /var/lib/apt/lists/* -vf - Start the installation, be sure to pay attention to restart, and because the mirror image has not been replaced, the installation is very slow. It took me maybe 6+ hours and 73% of the downloads were done. If the network speed is relatively good, one hour is enough (2M network), it is recommended not to use the computer to share the network (taking 2M network speed as an example, it may actually be between 50kb->600kb.), the network speed is unstable, and wifi is also Not necessarily stable, but more than enough to download this. It is best to use a network cable.
- Execute after installation.
(1) CUDA: Execute the following instructions
nvcc -V
If the following information appears, it means correct.
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_May__4_00:02:26_PDT_2022
Cuda compilation tools, release 11.4, V11.4.239
Build cuda_11.4.r11.4/compiler.31294910_0
- If there is no result, you can refer to this article: http://www.cppblog.com/jack-wang/archive/2022/07/27/229377.aspx
vim .bashrc
export PATH=/usr/local/cuda/bin:$PATH
source .bashrc
nvcc -VIt is displayed, you can view the version information of CUDA
(2) cuDNN: Execute the following instructions
dpkg -l libcudnn8
If the following information appears, it means correct.
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Architecture Description
+++-==============-===================-============-===========================>
ii libcudnn8 8.4.1.50-1+cuda11.4 arm64 cuDNN runtime libraries
（3）TensorRT：
dpkg -l tensorrt
If the following information appears, it means correct.
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Architecture Description
+++-==============-==================-============-============================>
ii tensorrt 8.4.1.5-1+cuda11.4 arm64 Meta package for TensorRT
（4）OpenCV：
dpkg -l libopencv
If the following information appears, it means correct.
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Architecture Description
+++-==============-===================-============-===========================>
ii libopencv 4.5.4-8-g3e4c170df4 arm64 Open Computer Vision Library
4. Install the jtop system monitoring tool.
This is a very famous monitoring tool of the Jetson series. Please execute the following instructions to install it:
sudo apt install python3-pip
sudo -H pip3 install -U pip
sudo -H pip install jetson-stats==4.0.0rc1
Then execute the following startup command to start the monitoring tool:
jtop
If there is an error in starting, please restart Jetson Orin to fully start the service.
sudo reboot
image.png
5. Chinese input method stepping on the pit
If you install Chinese input hair later, you must install the arm architecture, and note that the mirror source cannot be randomly selected.
A tutorial for installing Chinese input method. (Baidu and Sogou cannot be installed)
https://blog.csdn.net/weixin_39589455/article/details/122328395
sudo apt-get install fcitx fcitx-googlepinyin -y
- Set up fcitx, then reboot
sudo reboot
- Select input configuration, select Google pinyin
- Then you can use Google Pinyin, although it is not very easy to use, but it is better than nothing.