ubuntu 3060显卡驱动+cuda+cudnn+pytorch+pycharm+vscode

运行环境

ubuntu18.04 melodic
宏基暗影骑士笔记本

适用

ubuntu18.04 / ubuntu20.04

思路

3060显卡+CUDA11.1+cudnn8.1.0+pytorch1.8.0+pycharm2021+Anaconda+vscode

1.1 3060显卡驱动自动安装

1换清华源
2安装nvidia-driver-470-server
3重启电脑
4nvidia-smi (检查已显卡驱动版本)

CUDA Version: 11.4 (显卡驱动API 11.4)

2.1 CUDA11.1.1

CUDA下载地址https://developer.nvidia.com/cuda-toolkit-archive

1)下载CUDA Toolkit 11.1 Update 1 Downloads

运行API,小于等于上面显卡驱动API 11.4就可以

复制到浏览器下载, 不需要墙30分钟

 https://developer.download.nvidia.com/compute/cuda/11.1.1/local_installers/cuda_11.1.1_455.32.00_linux.run

cd 到下载路径

sudo sh cuda_11.1.1_455.32.00_linux.run

2)contunue , 然后accept

3)回车取消Driver安装然后install

4)添加环境变量

gedit ~/.bashrc

在文件结尾添加如下语句修改成自己的路径

export PATH="/usr/local/cuda-11.1/bin:$PATH"
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-11.1/lib64
export LIBRARY_PATH=$LIBRARY_PATH:/usr/local/cuda-11.1/lib

source ~/.bashrc

nvcc -V 

只要驱动API高于运行时API就没关系
nvidia-smi显示的是驱动APInvcc -V显示的是运行时API

5)确认是否安装成功

进入deviceQuery/目录

cd usr/local/cuda-11.1/samples/1_Utilities/deviceQuery/

sudo make

执行检查版本

./deviceQuery

3.1 cudnn 8.1.0

https://developer.nvidia.com/rdp/cudnn-archive
密码首位设置是大写

1)下载版本

Download cuDNN v8.1.0 (January 26th, 2021), for CUDA 11.0,11.1 and 11.2

2)解压

tar -xvzf cudnn-11.4-linux-x64-v8.2.2.26.tgz

sudo nautilus

打开超级复制复制include和lib64文件夹内容复制到cuda11.1对应文件夹中

3)添加环境变量

参考网址

gedit ~/.bashrc

export PATH=/usr/local/cuda-11.1/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-11.1/lib64\${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-11.1/lib\${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

source ./bashrc

4)查看cudnn版本

ls /usr/local/cuda/include/cudnn.h

如果输出 No such file or directory则cuDNN头文件未安装或者安装位置不正确,
如果输出 /usr/local/cuda/include/cudnn.h则cuDNN头文件已经安装成功。

ls /usr/local/cuda/lib64/libcudnn*

4.1 pycharm2021.3

1)社区版免费

https://www.jetbrains.com/pycharm/download/other.html

2)安装并创建pycharm图标

cd /home/duduzai/pycharm2021/bin
bash pycharm.sh

cd /usr/share/applications
sudo gedit pycharm.desktop

[Desktop Entry]
Version=1.0
Type=Application
Name=Pycharm
Icon=/home/duduzai/pycharm2021/bin/pycharm.png
Exec=sh /home/duduzai/pycharm2021/bin/pycharm.sh
MimeType=application/x-py;
Name[en_US]=pycharm

5.1 Anaconda 5.3.0(python3.7.0)

1)下载

https://mirrors.tuna.tsinghua.edu.cn/help/anaconda/
https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/

下载Anaconda3-5.3.0-Linux-x86_64.sh

2)安装

bash Anaconda3-2023.03-Linux-x86_64.sh

一直按回车 或 输入yes再回车(自动添加环境变量)

ROS自带的3.6.9被替换成anaconda自带的3.7.0

gedit ~/.bashrc
如果导致ROS包编译错误建议注释掉再重新编译ROS包(如相机雷达标定包)
source ~/.bashrc

# added by Anaconda3 5.3.0 installer
# >>> conda init >>>
# !! Contents within this block are managed by 'conda init' !!
__conda_setup="$(CONDA_REPORT_ERRORS=false '/home/duduzai/anaconda3/bin/conda' shell.bash hook 2> /dev/null)"
if [ $? -eq 0 ]; then
    \eval "$__conda_setup"
else
    if [ -f "/home/duduzai/anaconda3/etc/profile.d/conda.sh" ]; then
        . "/home/duduzai/anaconda3/etc/profile.d/conda.sh"
        CONDA_CHANGEPS1=false conda activate base
    else
        \export PATH="/home/duduzai/anaconda3/bin:$PATH"
    fi
fi
unset __conda_setup
# <<< conda init <<<

5.2 Anaconda 环境创建

1)创建名为 lidar_camera 的环境

conda create --name lidar_camera python=3.8

2)激活环境

conda activate  lidar_camera 

3)显示环境的安装包

conda list 

4)查看所有环境

conda env list

6.1 pytorch1.8

https://pytorch.org/get-started/previous-versions/

conda activate lidar_camera

conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=11.1 -c pytorch -c conda-forge

7.1 Anaconda和vscode联合使用

1)安装vscode环境

2)设置conda路径

查看conda环境路径

/home/duduzai/anaconda3/envs/lidar_camera/bin/python

⭐⭐⭐ 嘟嘟崽 ⭐⭐⭐ ⭐⭐⭐ 祝你成功 ⭐⭐⭐