windows10 tensorflow2 环境搭建

首先下载 python3

安装时带上pip，自动添加PATH环境变量，一开始想把python装在E盘，但在后续步骤中一直报权限错误，只好选择了默认安装……

官方的源更新要慢死，先更换taobao源，在自己的HOME目录（C:\Users\mycti）中新建一个pip目录，进入后新建pip.ini文件，将下列内容复制到文件。

[global] 
index-url = https://mirrors.aliyun.com/pypi/simple/ 
[install] 
trusted-host=mirrors.aliyun.com

新建并进入目录tensorflow

pip3 install -U pip virtualenv

很奇怪，第一次会报权限错误，运行第二次就不出错了

创建虚拟环境

virtualenv --system-site-packages -p python3 ./venv

进入虚拟环境

.\venv\Scripts\activate

如果你跟我一样用的powershell，这里报错 无法加载文件 E:\tensorflow\venv\Scripts\activate.ps1
那就要在开发者选项中设置，让PowerShell可以运行未签名脚本

然后开始安装tensorflow

pip install --upgrade tensorflow

安装完成后运行一下

python -c "import tensorflow as tf;print(tf.reduce_sum(tf.random.normal([1000, 1000])))"

输出如下，似乎是在说没有安装cuda驱动，不过这也说明tensorflow安装成功啦！

2020-04-12 12:30:37.383456: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cudart64_101.dll'; dlerror: cudart64_101.dll not found
2020-04-12 12:30:37.390595: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
2020-04-12 12:30:39.496176: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library nvcuda.dll
2020-04-12 12:30:39.561478: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1555] Found device 0 with properties:
pciBusID: 0000:01:00.0 name: GeForce GTX 1060 6GB computeCapability: 6.1
coreClock: 1.759GHz coreCount: 10 deviceMemorySize: 6.00GiB deviceMemoryBandwidth: 178.99GiB/s
2020-04-12 12:30:39.579631: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cudart64_101.dll'; dlerror: cudart64_101.dll not found
2020-04-12 12:30:39.595518: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cublas64_10.dll'; dlerror: cublas64_10.dll not found
2020-04-12 12:30:39.611450: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cufft64_10.dll'; dlerror: cufft64_10.dll not found
2020-04-12 12:30:39.628313: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'curand64_10.dll'; dlerror: curand64_10.dll not found
2020-04-12 12:30:39.643642: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cusolver64_10.dll'; dlerror: cusolver64_10.dll not found
2020-04-12 12:30:39.660441: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cusparse64_10.dll'; dlerror: cusparse64_10.dll not found
2020-04-12 12:30:39.676294: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'cudnn64_7.dll'; dlerror: cudnn64_7.dll not found
2020-04-12 12:30:39.684541: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1592] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
2020-04-12 12:30:39.703769: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1096] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-04-12 12:30:39.710747: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102]
tf.Tensor(-748.5011, shape=(), dtype=float32)

安装CUDA
看提示，我这里需要的是10.1版本的cuda，在官网下载安装 https://developer.download.nvidia.cn/compute/cuda/10.1/Prod/local_installers/cuda_10.1.243_426.00_win10.exe

选的默认安装，安装在 C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1

继续安装CUDNN
官网有安装步骤说明
到 https://developer.nvidia.com/rdp/cudnn-download 这里选择合适的版本进行下载，我选择的是 https://developer.nvidia.com/compute/machine-learning/cudnn/secure/7.6.5.32/Production/10.1_20191031/cudnn-10.1-windows10-x64-v7.6.5.32.zip
中间可能会需要注册用户，随便写一下就完了

上面的zip下载下来后，将里面的内容，连带所有子目录一起复制到CUDA的安装目录 即 C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1

在系统PATH环境变量检查一下有没有
CUDA_PATH C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1
安装了CUDA一般都会自动设置好的

至此，环境安装完毕