Installation tutorial under tensorflow-GPU acceleration-win10 (NVIDIA GeForce 940MX)

1. Install Anaconda

Due to xx (compiler or something, I didn't see it clearly), the complete tensorflow only supports python3.5, (except when using Docker). So Anaconda generally downloads the python3.5 version. To add the path path

2. Install CUDA Drivers

The resources are here, you can choose the system corresponding to your computer. After downloading, click to open the installation, follow the installation instructions step by step, it is best to choose the default path C drive, you can save unnecessary trouble. Download the 1.2G one

3.下载cuDNN - CUDA for Deep Neural Networks

This is actually the deep neural network support of CUDA. After downloading, extract a cuda folder, put it in the location you want, and then put the bin directory address under the cuda folder into the PATH. (must be logged in first)

4. Install tensorflow via Pip

# GPU version:pip install --upgrade tensorflow-gpu

# CPU version:pip install --upgrade tensorflow

a little slow

5. Test it with a simple matrix multiplication,

import tensorflow as tf

a = tf.random_normal((100, 100))
b = tf.random_normal((100, 500))
c = tf.matmul(a, b)
sess = tf.InteractiveSession()
sess.run(c)

Reference: https://blog.csdn.net/u010099080/article/details/53418159

https://blog.csdn.net/joker316701882/article/details/54426004