Skip to content

About

Environment setup for Nvidia Jetson series -> Xavier AGX, Xavier NX. Nvidia邊緣裝置環境建置

Resources

Readme
Activity

Stars

2 stars

Watchers

1 watching

Forks

0 forks
Report repository
BranchesTags

Latest commit

 

History

44 Commits

Folders and files

NameName
Last commit message
Last commit date
README.md
README.md
 
 

Repository files navigation

MyJetsonNote

NVIDIA® Jetson AGX Xavier™ | NVIDIA® Jetson Xavier™ NX

目錄

Flash JetPack on Jetson Xavier

NX

AGX - JetPack 4.6.1

Host電腦端: 準備一台系統為linux的電腦

  1. 在Host電腦端下載並安裝SDK Manager

  2. 硬體連接

  • 使用盒內Type-c線連接Host電腦端與AGX, 將AGX連接上電腦螢幕.
  • 進入Recovery Mode: AGX接上電源並保持關機狀態下, 按住中間的按鍵（Force Recovery）不放開, 再按下左邊的電源（Power）不放開, 等待約兩秒後同時放開.
  • 在Host電腦端上輸入lsusb查看是否連線上AGX（NVIDIA Corp.）

  1. 開啟SDK Manager, 登入NVIDIA 帳戶 login

  2. 如下圖設定 Jetson AGX , JetPack 4.6.1 , 可勾可不勾DeepStream, 點擊CONTINUE step1

  3. 點擊左下 I accept t..., 點擊 CONTINE step2

  4. 跳出一個視窗要求輸入系統密碼, 輸入系統密碼後以繼續安裝 4

  5. 進入 STEP 03, 等待下載及安裝至出現出視窗如下, 點選Manual Setup並按照提示操作後, 點擊Flash ss

  6. 等待出現視窗如下後, 移至AGX, 可以看到系統登入畫面, 請設定系統名稱、密碼等, 自動重開機後, 畫面停留至桌面 bfl

  7. 回到Host電腦端輸入剛才在TX2的設定的使用者名稱與密碼, 點擊Install bfl2

  8. 安裝結束, 點擊Finish and exit, 安裝完成 8

Check Installed

  1. CUDA
sudo find / -name  cuda

Or

## set an environment variable for cuda
export PATH=$PATH:/usr/local/cuda/bin
nvcc -V
  1. cuDNN
sudo find / -name libcudnn*
  1. TensorRT
sudo find / -name tensorrt

Change the Python3 default version

使用alternatives管理多版本軟體 將python版本指定為python2.7

sudo update-alternatives --install /usr/bin/python python /usr/bin/python2.7 100

將python版本指定為python3.6

sudo update-alternatives --install /usr/bin/python python /usr/bin/python3.6 150

如需切換python版本

sudo update-alternatives --config python

Install new versions of software

sudo apt update && sudo apt upgrade

Install python3-pip 

sudo apt install python3-pip

Jetson stats

jetson-stats is a package for monitoring and control your NVIDIA Jetson [Xavier NX, Nano, AGX Xavier, TX1, TX2] Works with all NVIDIA Jetson ecosystem

sudo pip3 install jetson-stats

Tensorflow

  1. Install JetPack on your Jetson device.

  2. Install system packages required by TensorFlow:

    sudo apt install libhdf5-serial-dev hdf5-tools libhdf5-dev zlib1g-dev zip libjpeg8-dev liblapack-dev libblas-dev gfortran

  3. Install and upgrade pip3

    sudo apt-get install python3-pip
    sudo pip3 install -U pip testresources setuptools==49.6.0

  4. Install the Python package dependencies

    sudo pip3 install -U --no-deps numpy==1.19.4 future==0.18.2 mock==3.0.5 \
keras_preprocessing==1.1.2 keras_applications==1.0.8 gast==0.4.0 \
protobuf pybind11 cython pkgconfig
    sudo env H5PY_SETUP_REQUIRES=0 pip3 install -U h5py==3.1.0

  5. Install TensorFlow using the pip3 command.

  • 對應 Jetpack 版本 - TENSORFLOW VERSION

  • [安裝範例1] 在 Jetpack 4.6 安裝最新版 TensorFlow

    sudo pip3 install --pre --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v46 tensorflow

  • [安裝範例2] 在 Jetpack 4.6 安裝 TensorFlow 版本小於2

    sudo pip3 install --pre --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v46 'tensorflow<2'

  • [安裝範例3] 安裝指定版本的 TensorFlow

    sudo pip3 install --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v$JP_VERSION tensorflow==$TF_VERSION+nv$NV_VERSION```

Opencv 4.5 

sudo apt install \
python3-dev python3-pip python3-tk \
build-essential make cmake cmake-curses-gui \
g++ libavformat-dev libavutil-dev \
libswscale-dev libv4l-dev libeigen3-dev \
libglew-dev libgtk2.0-dev libdc1394-22-dev libxine2-dev \
libgstreamer1.0-dev libgstreamer-plugins-base1.0-dev \
libjpeg8-dev libjpeg-turbo8-dev libtiff5-dev \
libavcodec-dev libxvidcore-dev libx264-dev libgtk-3-dev \
libatlas-base-dev gfortran libopenblas-dev liblapack-dev liblapacke-dev qt5-default      
sudo apt purge libopencv*
wget https://github.com/opencv/opencv/archive/4.5.0.zip -O opencv-4.5.0.zip \
&& unzip opencv-4.5.0.zip && cd opencv-4.5.0
wget https://github.com/opencv/opencv_contrib/archive/4.5.0.zip -O opencv_contrib-4.5.0.zip \
&& unzip opencv_contrib-4.5.0.zip
mkdir build && cd build
cmake \
-D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local/ \
-D OPENCV_EXTRA_MODULES_PATH=../opencv_contrib-4.5.0/modules \
-D CUDA_ARCH_BIN='7.2' \
-D WITH_CUDA=1 \
-D WITH_V4L=ON \
-D WITH_QT=ON \
-D WITH_FFMPEG=ON \
-D WITH_OPENGL=ON \
-D CUDA_FAST_MATH=1 \
-D WITH_CUBLAS=1 \
-D OPENCV_GENERATE_PKGCONFIG=1 \
-D BUILD_opencv_python2=1 \
-D BUILD_opencv_python3=1 \
-D CUDA_TOOLKIT_ROOT_DIR=/usr/local/cuda \
-D WITH_GTK_2_X=ON ..
sudo make -j$(nproc) && sudo make install

Opencv 3.4 (No opencv_contrib)

  • 由於AGX系統刷完後空間剩下10GB, 沒有足夠空間安裝opencv額外套件(opencv_contrib)
  • 參考 JK Jung's blog
  • 參考 Jetsonhacks
sudo apt purge libopencv*
sudo apt install --only-upgrade g++-5 cpp-5 gcc-5
sudo apt install \
python3-dev python3-pip python3-tk \
build-essential make cmake cmake-curses-gui \
g++ libavformat-dev libavutil-dev \
libswscale-dev libv4l-dev libeigen3-dev \
libglew-dev libgtk2.0-dev libdc1394-22-dev libxine2-dev \
libgstreamer1.0-dev libgstreamer-plugins-base1.0-dev \
libjpeg8-dev libjpeg-turbo8-dev libtiff5-dev \
libavcodec-dev libxvidcore-dev libx264-dev libgtk-3-dev \
libatlas-base-dev gfortran libopenblas-dev liblapack-dev liblapacke-dev qt5-default      
sudo pip3 install matplotlib==3.3.4  # 注意： 會自動安裝其他matplotlib所需套件, 如果要指定套件版本請自行個別安裝

Modify matplotlibrc (line #41) as 'backend: TkAgg'

sudo vim /usr/local/lib/python3.6/dist-packages/matplotlib/mpl-data/matplotlibrc

Modify /usr/local/cuda/include/cuda_gl_interop.h and fix the symbolic link of libGL.so.

sudo vim /usr/local/cuda/include/cuda_gl_interop.h

Here’s how the relevant lines (line #62~68) of cuda_gl_interop.h look like after the modification.

//#if defined(__arm__) || defined(__aarch64__)
//#ifndef GL_VERSION
//#error Please include the appropriate gl headers before including cuda_gl_interop.h
//#endif
//#else
 #include <GL/gl.h>
//#endif

Next, download opencv-3.4.0 source code, cmake and compile. Note that opencv_contrib modules (cnn/dnn stuffs) would cause problem on pycaffe, so after some experiments I decided not to include those modules at all.

wget https://github.com/opencv/opencv/archive/3.4.0.zip -O opencv-3.4.0.zip
unzip opencv-3.4.0.zip && cd opencv-3.4.0
mkdir build && cd build

Build opencv (CUDA_ARCH_BIN="6.2" for TX2, or "5.3" for TX1)

cmake -D CMAKE_BUILD_TYPE=RELEASE  \
      -D CMAKE_INSTALL_PREFIX=/usr/local \
      -D WITH_CUDA=ON  \
      -D CUDA_ARCH_BIN="7.2"  \
      -D CUDA_ARCH_PTX="" \
      -D WITH_CUBLAS=ON  \
      -D ENABLE_FAST_MATH=ON  \
      -D CUDA_FAST_MATH=ON \
      -D ENABLE_NEON=ON  \
      -D WITH_LIBV4L=ON  \
      -D BUILD_TESTS=OFF \
      -D BUILD_PERF_TESTS=OFF  \
      -D BUILD_EXAMPLES=OFF \
      -D WITH_QT=ON  \
      -D WITH_OPENGL=ON ..

sudo make -j$(nproc) && sudo make install

To verify the installation:

ls /usr/local/lib/python3.6/dist-packages/cv2.*
ls /usr/local/lib/python2.7/dist-packages/cv2.*
python3 -c 'import cv2; print(cv2.__version__)'
python2 -c 'import cv2; print(cv2.__version__)'

PyAudio & Sounddevice 

sudo apt install python-all-dev portaudio19-dev libffi-dev
sudo pip3 install pyaudio sounddevice

# pip install pipwin
# pipwin install pyaudio

Librosa 0.6.3

llvmlite Compatibility

llvmlite versions compatible LLVM versions
0.34.0 - ... 10.0.x (9.0.x for aarch64 only)
0.33.0 9.0.x
0.29.0 - 0.32.0 7.0.x, 7.1.x, 8.0.x
0.27.0 - 0.28.0 7.0.x
0.23.0 - 0.26.0 6.0.x
0.21.0 - 0.22.0 5.0.x
0.17.0 - 0.20.0 4.0.x
0.16.0 - 0.17.0 3.9.x
0.13.0 - 0.15.0 3.8.x
0.9.0 - 0.12.1 3.7.x
0.6.0 - 0.8.0 3.6.x
0.1.0 - 0.5.1 3.5.x

Upgrade the SETUP tools:

sudo pip3 install -U setuptools 
sudo pip3 install cython

Install LLVM & LLVMLITE:

wget http://releases.llvm.org/7.0.1/llvm-7.0.1.src.tar.xz \
tar -xvf llvm-7.0.1.src.tar.xz \
cd llvm-7.0.1.src && mkdir llvm_build_dir && cd llvm_build_dir/ \
cmake ../ -DCMAKE_BUILD_TYPE=Release -DLLVM_TARGETS_TO_BUILD="ARM;X86;AArch64" \
sudo make -j$(nproc) && sudo make install
cd bin/ \
echo "export LLVM_CONFIG=\""`pwd`"/llvm-config\"" >> ~/.bashrc \
echo "alias llvm='"`pwd`"/llvm-lit'" >> ~/.bashrc \
source ~/.bashrc
Finally
sudo pip3 install llvmlite==0.32.1

Install Numba, Scipy, Joblib, Scikit-learn

sudo apt install libblas-dev liblapack-dev libatlas-base-dev gfortran
sudo pip3 install numba==0.38.0  # numba==0.48.0
sudo pip3 install scipy==1.1.0  # scipy==1.4.1
sudo pip3 install joblib==0.12  # joblib==0.14
sudo pip3 install scikit-learn==0.21.1

Install Librosa

sudo pip3 install librosa==0.6.3  # librosa==0.7.2
  • 如果 joblib 有問題 可以嘗試更新 joblib

PyTorch 

sudo apt install python3-pip libopenblas-base libopenmpi-dev
sudo apt install libjpeg-dev zlib1g-dev
sudo pip3 install Cython

PyTorch 1.7.0 + Torchvision 0.8.1

wget https://nvidia.box.com/shared/static/cs3xn3td6sfgtene6jdvsxlr366m2dhq.whl -O torch-1.7.0-cp36-cp36m-linux_aarch64.whl
sudo pip3 install torch-1.7.0-cp36-cp36m-linux_aarch64.whl
git clone --branch v0.8.1 https://github.com/pytorch/vision torchvision
cd torchvision && sudo python setup.py install

PyTorch 1.8.0 + Torchvision 0.9.0

wget https://nvidia.box.com/shared/static/p57jwntv436lfrd78inwl7iml6p13fzh.whl -O torch-1.8.0-cp36-cp36m-linux_aarch64.whl
sudo pip3 install torch-1.8.0-cp36-cp36m-linux_aarch64.whl
git clone --branch v0.9.0 https://github.com/pytorch/vision torchvision
cd torchvision && sudo python setup.py install

PyTorch 1.10.0 + Torchvision 0.11.3

wget https://nvidia.box.com/shared/static/fjtbno0vpo676a25cgvuqc1wty0fkkg6.whl -O torch-1.10.0-cp36-cp36m-linux_aarch64.whl
sudo pip3 install torch-1.10.0-cp36-cp36m-linux_aarch64.whl
git clone --branch v0.11.3 https://github.com/pytorch/vision torchvision
cd torchvision && sudo python setup.py install

Torchaudio 

git clone https://github.com/pytorch/audio.git torchaudio --branch v0.7.2
sudo apt install sox libsox-dev libsox-fmt-all
cd torchaudio && sudo python setup.py install

PyQt5

Method 1

If build wheel error, try Method 2

sudo apt install qt5-default python3-pyqt5 pyqt5-dev-tools qttools5-dev-tools
sudo pip3 install pyqt5==5.15.6 
# 測試中
# pyqt5-sip==12.9.1
# sudo pip3 install sip==5.0.1

Method 2

Install SIP

If you have an old version of SIP

sudo apt remove python3-sip
wget https://www.riverbankcomputing.com/static/Downloads/sip/4.19.25/sip-4.19.25.tar.gz
tar -xvf sip-4.19.25.tar.gz
cd sip-4.19.25
python3 configure.py
make -j$(nproc) && sudo make install
Install PyQt5
使用 版本 PyQt5-5.15.4.tar.gz, 下載後並解壓
python3 configure.py
make -j$(nproc) && sudo make install

Install ie multimedia

sudo apt install python3-pyqt5.qtmultimedia

Install kudio 

pip install kudio
  • 如果遇到ubuntu20.04系統無法安裝PyAudio
sudo apt -y install portaudio19-dev
# pip install PyAudio-0.2.11-cp38-cp38-linux_x86_64.whl

Others

sudo pip install pyusb click xlsxwriter tqdm imutils qdarkstyle
sudo pip install pandas==1.1.4 PyYAML==5.3.1 --ignore-installed
sudo pip install seaborn==0.11.0 # 注意, 會自動安裝相關套件!!
sudo dpkg-reconfigure dash
sudo apt install filezilla curl

Matplotlib

  • 不推薦, 但是如果你真的裝不起來再試試這個 
    sudo apt install python3-matplotlib

pyqtgraph

sudo pip install pyqtgraph==0.11.1

FTP server 

sudo apt install vsftpd
# systemctl status vsftpd
# netstat -tupln | grep ftp
sudo vi /etc/vsftpd.conf
# Allow anonymous FTP? (Disabled by default).
anonymous_enable=YES
# Uncomment this to enable any form of FTP write command.
write_enable=YES

# Uncomment this to allow the anonymous FTP user to upload files. This only
# has an effect if the above global write enable is activated. Also, you will
# obviously need to create a directory writable by the FTP user.
anon_upload_enable=YES
sudo systemctl restart vsftpd

Firefox 

sudo apt install firefox

Kolourpaint4 

sudo apt install kolourpaint4

exFAT driver 

sudo add-apt-repository universe && sudo apt install exfat-fuse exfat-utils

Java 

sudo apt install default-jre # sudo apt install default-jdk

Clone SD card

find your SD card
sudo fdisk -l
clone SD card to img
sudo dd if=/dev/sda conv=sync,noerror bs=1M status=progress | gzip -c > ./backup.img.gz
clone img to SD card
sudo gunzip -c ./backup.img.gz | sudo dd of=/dev/sda bs=1M status=progress

AGX Fast Backup

  • AGX快速備份與平展, 不須透過 NVIDIA SDK Manager, 平均20分鐘內可安裝完成一台
Coming soon

About

Environment setup for Nvidia Jetson series -> Xavier AGX, Xavier NX. Nvidia邊緣裝置環境建置

Resources

Readme
Activity

Stars

2 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

Packages

Contributors