本文以Ubuntu 18.04 + CUDA 10.0 + cuDNN 7.4 + TensorFlow 1.13.1为例,其他版本安装时版本匹配即可
CUDA安装
安装前
在安装CUDA工具包和驱动程序之前,必须执行一些操作
1. Verify You Have a CUDA-Capable GPU
lspci | grep -i nvidia
2. 验证你有一个受支持的Linux版本
uname -m && cat /etc/*release
你应该看到输出类似如下,修改为您的特定系统:
x86_64
Red Hat Enterprise Linux Workstation release 6.0 (Santiago)
x86_64行表示您在64位系统上运行,其余部分给出了关于版本的信息。
3. 验证系统已安装gcc
gcc --version
# 若未安装运行:
sudo apt-get install build-essential
4. 验证系统已正确安装Kernel Headers和Development Packages
uname -r
# 对应当前运行内核的内核头文件和开发包安装:
sudo apt-get install linux-headers-$(uname -r)
5. 下载NVIDIA CUDA Toolkit
CUDA官方下载(deb版本)
MD5检验
md5sum cuda-repo-ubuntu1804-10-0-local-10.0.130-410.48_1.0-1_amd64.deb
安装
1. 执行安装前操作
2. Install repository meta-data
sudo dpkg -i cuda-repo-ubuntu1804-10-0-local-10.0.130-410.48_1.0-1_amd64.deb
3. Installing the CUDA public GPG key
sudo apt-key add /var/cuda-repo-10-0-local-10.0.130-410.48/7fa2af80.pub
4. Update the Apt repository cache
sudo apt-get update
5. 安装CUDA
sudo apt-get install cuda
# 若报错,则运行以下两条
sudo apt-get install aptitude
sudo aptitude install cuda
6. 执行安装后操作
安装后
重启电脑,检查Device Node Verification
ls /dev/nvidia*
设置环境变量
sudo gedit /etc/profile
# 打开后添加:
export PATH=/usr/local/cuda-10.0/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64\ ${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
# 环境变量:在~/.bashrc的最后添加
export PATH=/usr/local/cuda/bin${PATH:+:${PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
export CUDA_HOME=/usr/local/cuda
保存并重启
验证驱动版本
cat /proc/driver/nvidia/version
验证CUDA Toolkit
nvcc -V
cuDNN安装
cnDNN官方下载
cp cudnn-10.0-linux-x64-v7.4.2.24.solitairetheme8 cudnn-10.0-linux-x64-v7.4.2.24.tgz
tar -xzvf cudnn-10.0-linux-x64-v7.4.2.24.tgz
sudo cp cuda/include/cudnn.h /usr/local/cuda/include
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*
TensorFlow安装
python --version
pip --version
virtualenv --version
sudo apt update
sudo apt install python-dev python-pip
sudo pip install -U virtualenv # system-wide install
virtualenv --system-site-packages -p python2.7 ./venv
source ./venv/bin/activate # sh, bash, ksh, or zsh
pip install --upgrade pip
pip list # show packages installed within the virtual environment
pip install --upgrade tensorflow-gpu
deactivate # don't exit until you're done using TensorFlow
卸载
# cuda卸载
sudo apt-get --purge remove <package_name>
# cudnn卸载
sudo rm -rf /usr/local/cuda/include/cudnn.h
sudo rm -rf /usr/local/cuda/lib64/libcudnn*
# tensorflow卸载
pip uninstall tensorflow-gpu