window laptop install unbuntu and tensorflow-gpu

1. download ubuntu: https://cn.ubuntu.com/download/desktop

2. follow instructions to install ubuntu:Install Ubuntu desktop | Ubuntu

3.install docker engine: Install Docker Engine on Ubuntu | Docker Docs

4.pull tensorflow docker iamges: Docker  |  TensorFlow

5.install nvidia-docker-toolkit:

GitHub - NVIDIA/nvidia-docker: Build and run Docker containers leveraging NVIDIA GPUsInstalling the NVIDIA Container Toolkit — NVIDIA Container Toolkit 1.14.2 documentation

6.test your tensorflow-gpu-jupyter

1. test on bash:
docker run --gpus all -it --rm tensorflow/tensorflow:latest-gpu \
   python -c "import tensorflow as tf;import numpy as np;import time;value =np.random.randn(5000,1000);a = tf.constant(value);tic = time.time();b = [tf.reduce_sum(a*a) for i in range(1000)];toc = time.time();t_cost = toc - tic;print(t_cost);"




2. test on notebook
2.1 gpu version
docker run --gpus all -p 8888:8888 tensorflow/tensorflow:latest-gpu-jupyte

2.2 cpu version
docker run -it -p 8888:8888 tensorflow/tensorflow:latest-gpu-jupyter

2.3 test time used
import tensorflow as tf;
import numpy as np;
import time;
value = np.random.randn(5000,1000);
a = tf.constant(value);
tic = time.time();
b = [tf.reduce_sum(a*a) for i in range(1000)];
toc = time.time();
t_cost = toc - tic;
print(t_cost);