1. First check if there is any problem with your graphics card driver.
If you right-click the computer and there is (nvidia control panel), do not do the following
If not, you need to perform the following operations
(Right-click this computer, find Management and open it)
Find device manager
Find the display adapter (your graphics card model will be here)
Based on the above information, we can go to the NVIDIA official website to find the corresponding graphics card driver update or download for our graphics card.
Graphics card driver download address
After installing the graphics card driver, we press the win+R key combination to open the cmd command window and enter the following command
nvidia-smi
You can see that the driver version is 527.56; the highest supported CUDA version is version 12.0. We can install the environment based on this information.
2. Installation of anaconda
Go to the official website to download the anaconda installation package Anaconda | Anaconda Distribution
Install
Remember to select this step (add)
In this way, our anaconda is installed
3. Installation of pytorch
Enter the anaconda prompt
Execute code
conda env list
After executing the code, you can see the virtual environment you created under anaconda (this is what I have created including paddle, pytorch, etc.)
If you install it for the first time, you will only have a base environment.
At this time we create our own pytorch environment (here we can specify the python version number, for example, mine is python3.9)
conda create -n pytorch python=3.9
It will ask you to configure the corresponding environment package and enter y (y represents yes)
At this time, our pytorch environment has been established, but we need to go in and configure the corresponding packages for deep learning.
First we have to go into the pytorch environment
conda activate pytorch
Downloading related environment packages is relatively slow, so we change the source of the environment. In the pytorch environment, execute the following naming to change the environment to Tsinghua source.
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 --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
conda config --set show_channel_urls yes
At this time we open the official website of pytorch ( PyTorch ) and select our version
Copy it and execute it in our pytorch environment
conda install pytorch torchvision torchaudio pytorch-cuda=11.7
因为我们刚刚进行了换源,不需要把后面的-c pytorch -c nvidia复制过来,那样下载速度会很慢,然后慢慢下载就好啦!