Anaconda related

torch, torchvision, python The version corresponds to

Anaconda Prompt Creating a virtual environment

conda env list  #  View the created virtual environment 
conda create -n  mytorch python=3.9  #  Create a new virtual environment 
conda activate mytorch  #  Enter the virtual environment 
conda deactivate  #  Exit virtual environment 
conda remove -n mytorch --all  #  Delete virtual environment 

test Anaconda Is the installation successful ？

stay CMD Type in python Check to see if there is Python Environmental Science , stay CMD Type in conda -V Check to see if there is conda Environmental Science .

install pytorch

Install from the official website pytorch Too slow , So we configure the Tsinghua image source .
stay Anaconda Prompt In the virtual environment created ：

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --set show_channel_urls yes
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
# conda  install  pytorch
conda install pytorch torchvision torchaudio cpuonly -c pytorch  #  Get rid of the back -c pytorch, It means downloading from the official website , Instead of downloading from Tsinghua mirror source  

AMD Graphics card installation CPU edition

• stay Device manager View in Display adapter
• Ctrl+shift+esc see performance
• AMD Update the driver on the official website → Drive and support → Choose your own graphics card model AMD Radeon (TM) R9 M375→ Submit → download The driver → Installation
• Check the driver of the graphics card ： Control panel → Hardware and sound → Device manager → Display adapter → attribute → The driver
• Use conda from Installation of Tsinghua image CPU Version of pytorch

AMD Official website

NVIDIA Graphics card installation GPU edition

cuda install & cudnn install & pytorch install

Check pytorch Is the installation successful ？

 type  python  Get into  python  Interpreter 
>>>import torch
>>>torch.__version__  #  see torch Current version number 
'1.11.0'

# pytorch GPU edition 
>>>torch.version.cuda  #  Compile the current version of torch The use of cuda Version number 
'11.3'
>>>torch.cuda.is_available()  #  View the current cuda Whether it can be used in the current version of Torch, If the output True, Is available 
True

uninstall pytorch

Use conda uninstall pytorch

conda uninstall pytorch
conda uninstall libtorch

Use pip uninstall Pytorch

pip uninstall torch

CUDA & CUDNN

CUDA（ComputeUnified Device Architecture） It's the video card manufacturer NVIDIA The new computing platform . CUDA It's a kind of NVIDIA General parallel computing architecture , The architecture enables GPU Able to solve complex computing problems .

NVIDIA cuDNN It is used for deep neural network GPU Acceleration Library , Want to be in CUDA Run deep neural network on , Just install cuDNN.

Check GPU Do you support CUDA？

Ctrl+Alt+Delete open Task manager > performance >GPU>intel(R)HDGraphics 5500+AMD +AMD Radeon (TM) R9 M375 I won't support it CUDA