Tensorflow+Tensorlfow_gpu environment install in Windows 10
What is tf2.x and tf1.x different??
compare GPU and CPU package:
tf2.x is in the same package
tf1.x will need to install them separately
I will create the environment like list below step by step
tensorflow=2.3.0
cuda =10.1
cudnn =7.6
my system info:
GPU:GEFORCE 940MX
OS:windows10(x64)
Step 1.google type in “GPU model+SPECIFICATIONS”,check in the GPU official website if it support CUDA.
Step 2. Check tensorflow_gpu version
https://www.tensorflow.org/install/source_windows
Step 3.
Check OS version
Step 4.
Install CUDA and set up environment
(1.)Install CUDA ,download from below
https://developer.nvidia.com/cuda-toolkit-archive
(2.) Cudnn to replace CUDA file,download from below
https://developer.nvidia.com/rdp/cudnn-download
(3.) go to CUDA path to replace file with cudnn
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1
(4.)set up the system environment variables
go to search type
Edit the system environment variables
Ahead to environment variables
edit “system variables”->“Path”
then add two lines below:
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\lib\x64
afterwards open anaconda’s terminal (in any environment) type :nvcc -V
Step 5.Create virtual environment in ANACONDA
(1). install anaconda
https://www.anaconda.com/products/individual
(2.)Create TENSORFLOW environment name “MaskRCNN” with python version 3.7 in terminal
conda create -n MaskRCNN pip python=3.7
(3.) Activate the environment
activate MaskRCNN
(4.) install tensorflow
Program excecute by CPU
pip install tensorflow==2.1.0
check if tensorflow is installed
pip show tensorflow
(5.) double the environment all correct
pip list
(6.)tensorflow_gpu check and test
check if tensorflow_gpu is installed
test if tensorflow_gpu is working
python
from tensorflow.python.client import device_lib print(device_lib.list_local_devices())