Foreword:
CUDA and cuDNN installation (linux) Install graphics driver and CUDA program under Linux
System parameters:
System: ubuntu20.04.2
Graphics card: GeForce GTX 1050
Version to be installed: CUDA 11.4
1. Install the graphics card driver
(1) Detect the model of your NVIDIA graphics card and the recommended driver
ubuntu-drivers devices
(2) Choose to install all recommended drivers
sudo ubuntu-drivers autoinstall
(3) nvidia-smi checks whether the driver installation is successful
The results show that:
NVIDIA-SMI 470.74 Driver Version: 470.74 CUDA Version: 11.4
Second, install the CUDA program
(1) Download cuda:
Enter NVIDIA to download the installation package CUDA Toolkit 11.7 Downloads | NVIDIA Developer
Select the corresponding version, and enter the relevant commands in the terminal (note: enter your own installation version)
wget http://developer.download.nvidia.com/compute/cuda/11.0.2/local_installers/cuda_11.0.2_450.51.05_linux.run
(2) cuda installation:
sudo sh cuda_11.0.2_450.51.05_linux.run
(3) cuda environment variable configuration:
.1. Install vim package
sudo apt install vim # version 2:8.1.2269-1ubuntu5.4, or
sudo apt install vim-tiny # version 2:8.1.2269-1ubuntu5.4
sudo apt install neovim # version 0.4.3-3
sudo apt install vim-athena # version 2:8.1.2269-1ubuntu5.4
sudo apt install vim-gtk3 # version 2:8.1.2269-1ubuntu5.4
sudo apt install vim-nox # version 2:8.1.2269-1ubuntu5.4
.2. Setting environment variables gedit ~/.bashrc
推荐用这个,打开是文件好操作
vim ~/.bashrc
Open the file and enter the following: (note: i
插入 :
wq
退出
)
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64
export PATH=$PATH:/usr/local/cuda/bin
export CUDA_HOME=$CUDA_HOME:/usr/local/cuda
Finally, enter the following to update the environment variables
source ~/.bashrc
(4) Check whether the installation is successful
nvcc -V
3. Install the cudnn program
(1) cudnn download
cuDNN is a library for GPU-accelerated computing of deep neural networks. First go to the official website cuDNN Archive | NVIDIA Developer (2) Installation
解压缩->在压缩后的
cuda文件夹,打开终端
#Note, the unzipped folder will copy the corresponding file to cuda in /usr/local
sudo cp cuda/include/cudnn.h /usr/local/cuda-11.0/include/
sudo cp cuda/lib64/libcudnn* /usr/local/cuda-11.0/lib64/
sudo chmod a+r /usr/local/cuda-11.0/include/cudnn.h
sudo chmod a+r /usr/local/cuda-11.0/lib64/libcudnn*