Use Anaconda to configure the tensorflow of GPU Version (less than 30 series graphics cards)

Use anaconda To configure gpu Version of tensorflow（30 Series following graphics cards ）

Environmental content

One 、Anaconda Environment configuration

1、Anaconda The download （ You can skip ）

2、Anaconda Installation

Two 、Cudnn and CUDA Download and install  

1、Cudnn and CUDA The download （ Given ） 

2、Cudnn and CUDA Installation

  3、 ... and 、 To configure Tensorflow Environmental Science

1、Tensorflow Creation and activation of environment

2、 Installation of related libraries

Four 、 Use tf2-gpu Environmental Science ( But look at the last step of my previous tutorial ）

---

Environmental content

tensorflow-gpu：2.2.0

No installation required keras,tensorflow And keras Amalgamated , In short ,tensorflow Bring their own keras

All installation files to be used next —— Online disk download （ Don't download it yourself ）：

link ：https://pan.baidu.com/s/1cSwNCRKY0syOt-NBEdr94Q 
Extraction code ：p002   

One 、Anaconda Environment configuration

1、Anaconda The download （ You can skip ）

————————————————
The new version anaconda The download ：
         Install the latest Anaconda, Log in first Anaconda Its official website ：Anaconda | Individual Edition. Directly download the corresponding installation package .

2、Anaconda Installation

Open the download package .

         Choose where to install , It can not be installed in C disc .  I chose Add Anaconda to my PATH environment variable, This will automatically anaconda Put it into the environment variable of the system , Configuration will be more convenient .

Wait until the installation is complete ,Anaconda The installation of is over . 

Two 、Cudnn and CUDA Download and install  

         What I use here is tensorflow-gpu=2.2.0, Officially recommended Cuda The version is 10.1, So it's going to use cuda10.1, And cuda10.0 Corresponding cudnn yes 7.6.5.（ Of course, there is no problem with higher versions , But try not to be lower than ）

1、Cudnn and CUDA The download （ Given ） 

The specific download process includes cumbersome steps such as registration , Now that you have given the installation package , In order to keep it simple, I won't repeat it here .

2、Cudnn and CUDA Installation

After downloading, you can open exe File to install .

Select custom here .

Then just click the next step . 

          After installation C The root directory can be found at this location .C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0 Then we put Cudnn To extract the contents of .  Copy the contents of this directly to C Just under the root directory of the disk .

  3、 ... and 、 To configure Tensorflow Environmental Science

1、Tensorflow Creation and activation of environment

  open anaconda prompt：

conda create -n tf-gpu python=3.7 -y

activate tf2-gpu

There are two instructions here ：
The previous instruction is used to create a file named tf2-gpu Environment , Of the environment python Version is 3.7.
The latter instruction is used to activate a program named tf2-gpu Environment . 

Because all our operations must be carried out in the corresponding environment , Therefore, you need to activate the environment before installing the library .

2、 Installation of related libraries

Overview of libraries to be installed ：

scipy==1.4.1
numpy==1.18.4
matplotlib==3.2.1
opencv_python==4.2.0.34
tensorflow_gpu==2.2.0
tqdm==4.46.1
Pillow==8.2.0
h5py==2.10.0

If you want a more convenient installation, you can create one on the desktop or elsewhere requirements.txt file , Copy the above to txt In file .

  Enter the command ：

Tips ： Just move the file to anaconda/cmd The absolute path of the file can be automatically obtained in the window .

pip install -r C:\Users\33232\Desktop\requirements.txt -i https://pypi.mirrors.ustc.edu.cn/simple/

Four 、 Use tf2-gpu Environmental Science ( But look at the last step of my previous tutorial ）

(5 Bar message ) How to use without a graphics card anaconda To configure tensorflow Deep learning environment _ Dance rain . The blog of -CSDN Blog _tensorflow-cpu

complete ！

If you think this article is helpful to you , Please like it + Collect it , thank you ！