Anaconda = > pycharm=> CUDA=> cudnn=> pytorch environment configuration

Tips ： When the article is finished , Directories can be generated automatically , How to generate it, please refer to the help document on the right

---

Preface

Anaconda => PyCharm => CUDA => cudnn => PyTorch Environment configuration

Tips ： The following is the main body of this article , The following cases can be used for reference

---

One 、Anaconda install

1、 Go to the official website to download the version , Just download the latest version , The following is the official website link .

https://www.anaconda.com/products/distribution

2、 After downloading, it has been run as an administrator

3、 For personal use Just Me, A bunch of people use it All Users, Will do , Not much difference .

4、 Change the path by yourself , You need to remember the corresponding path ！

5、 Check the first , Otherwise, you need to add environment variables later .

6、 then Install, Wait patiently for installation .

Uncheck both , Otherwise, the browser will open , Pop up the page . Click on Finish Complete the installation .

Anaconda Prompt: Command line window .
Spyder: Can be done Python Code writing .

7、 To check whether this computer has been installed Anaconda, May press win+R Call up the command line window , Input cmd command

8、 Input conda -V You can check whether it is installed conda, If it is not installed, it displays conda Not internal or system commands .

---

Two 、Anaconda Configuration needs Python A virtual environment

1、 open Anaconda Prompt .

2、 Input python You can view it Anaconda Installed Python edition （ Pre installation was also mentioned in the previous installation process .）

3、 Enter the following command to create a virtual environment .
conda create -n python3.7 python=3.7
conda create -n : Prefix , establish
python3.7 : The name of the folder created （ It can be changed to a name convenient for you ）
python=3.7 ：python Corresponding version , Here to install 3.7 edition

4、 Create success ！

5、 Activate the environment
conda activate python3.7

6、 After entering the environment, you can view the installed python edition ,
as well as You need to install the corresponding CUDA Version of Pytorch.

7、 Then create a folder , stay pycharm Open as project in ,
Set up python The interpreter is created in the virtual environment python.exe. that will do
View the created python See the following blog for the location of virtual environment ：

https://blog.csdn.net/LiDe2000/article/details/125946216?spm=1001.2014.3001.5501

---

3、 ... and 、Pycharm install

Download it on the official website , The following is the official website link .

https://www.jetbrains.com/zh-cn/pycharm/

The professional edition costs money （ You can find the cracked version by yourself ）, Community free （ Basically enough for daily use ）
Download the latest Community Edition （ You can also download the previous version , No big problem ）

It can be installed normally （ Fill in the pictures later ..）

---

Four 、CUDA install

1、 open NVIDA Control panel , Click the system information in the lower left corner , Check the maximum supported by the graphics card CUDA edition .

2、 Go to Pytorch Check the official website for supported CUDA edition .

https://pytorch.org/get-started/locally/

3、 Find Pytorch Supported by CUDA After version （CUDA Version needs to be associated with Pytorch The installation command corresponds to , Otherwise, you may not be able to call GPU）, Go to CUDA Download the corresponding version on the official website .

Official website ：https://developer.nvidia.cn/zh-cn

I choose to download 11.3 （11.3.0 And 11.3.1 All possible ）.

4、 Download it CUDA after , Choose to run as Administrator
Compare new 、 current version , If the new version is low, uncheck , Not to install .

Choose the installation location, then you can create a new location by yourself （ Forget the screenshot , There is no picture ）

It can be installed normally .

---

5、 ... and 、cudnn To configure

1、 Go to the official website to download CUDA Corresponding cudnn edition , The corresponding version will be displayed when downloading .

Official website ：https://developer.nvidia.cn/zh-cn

You may have to join to download , Then log in , Download after joining .

2、 Download after decompression , Copy the three folders 、 Paste into the previously installed CUDA Under the installation way .


complete ！

---

6、 ... and 、PyTorch install

1、 open Anaconda Prompt , Using the previous command, you need to install Pytorch Virtual environment for .

You can view the installed CUDA edition .

2、 stay Pytorch Find the corresponding installation command on the official website , Enter into the panel above , Ready for installation （ The installation uses American sources , Therefore, the installation may fail or be slow , If the installation fails, install it during the day or after installing the domestic source .）
Previously installed by the author CUDA 11.3, So choose the installation command here .

3、 Wait for installation

---

7、 ... and 、 Verify that GPU

1、 stay Pycharm Choose to install Pytorch Of Python Environmental Science , Above installation Anaconda When writing in detail .

import torch
print(torch.__version__)
flag = torch.cuda.is_available()
if flag:
    print("CUDA is available !")
else:
    print("CUDA is not available !")

gpu_num = 1
# Device which device we want to run on
device = torch.device("cuda:0" if (torch.cuda.is_available() and gpu_num > 0) else "cpu")
print("Drive is: ", device)
print("GPU model: ", torch.cuda.get_device_name(0))

GPU Can successfully call .

---

summary

The configuration process is a little long , It may not be detailed enough （ It is a record post of the author ）, Other problems encountered during installation will be updated later .