ubuntu 18.04 Nvidia Rtx 2060 Driver+cudnn+cuda

1.显卡驱动470

参考:https://blog.csdn.net/weixin_41281151/article/details/121590003

2.安装cuda10.0

(下载链接https://developer.nvidia.com/cuda-toolkit-archive，对比关系图https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html)

1）sudo chmod a+x cuda_xxxxx_linux.run

2）sudo sh cuda_xxxxx_linux.run

中间再安装显卡驱动时，要NO，其他都是YES或者default

3）成功后，修改配置文件sudo vim ~/.bashrc

4）添加语句并保存：

export PATH=/usr/local/cuda/binKaTeX parse error: Expected '}', got 'EOF' at end of input: {PATH:+:{PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/lib64KaTeX parse error: Expected '}', got 'EOF' at end of input: …LIBRARY_PATH:+:{LD_LIBRARY_PATH}}

或者可以为：

export PATH=/usr/local/cuda-10.0/binKaTeX parse error: Expected '}', got 'EOF' at end of input: {PATH:+:{PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64KaTeX parse error: Expected '}', got 'EOF' at end of input: …LIBRARY_PATH:+:{LD_LIBRARY_PATH}}

5）source ~/.bashrc

测试：

cd /usr/local/cuda-8.0/samples/1_Utilities/deviceQuery
sudo make
sudo ./deviceQuery

或者：nvcc–version

连接：sudo ln -s /usr/local/cuda-10.0 /usr/local/cuda

3.cudnn7.6.1

(下载链接https://developer.nvidia.com/rdp/cudnn-download)

1）解压tar zxvf FileName.tgz

2）将文件倒入相关内容中

sudo cp cudnn.h /usr/local/cuda/include/

sudo cp lib* /usr/local/cuda/lib64/

3）动态文件进行链接

cd /usr/local/cuda/lib64/
sudo rm -rf libcudnn.so libcudnn.so.7 #删除原有动态文件
sudo ln -s libcudnn.so.7.6.1 libcudnn.so.7 #生成软衔接
sudo ln -s libcudnn.so.7 libcudnn.so #生成软链接

4.安装pip，进行安装其他依赖库

sudo apt-get install python3-pip

之后pip install opencv-python numpy scipy sklearn

5.安装tensorflow-gpu

首先确认自己安装的版本，因为相互之间的对应关系https://tensorflow.google.cn/install/source#linux

pip install tensorflow-gpu或者pip install tensorflow-gpu==1.31.1 (高版本貌似会报错)

pip install keras