首先 cuda 装的是 9.1.85_win10_64 位，是在百度网盘上下载的 local 可执行文件（因为官网的下载太慢而且联网版安装包也很慢），用 vs2015 测试自带 Samples 成功。
然后 cudnn 是下载官网的 9.0 版本，但是按照网上做法测试报错：

#include <iostream>
#include <cuda_runtime.h>
#include <cudnn.h>
using namespace std;

void main() {
	cudnnHandle_t handle;
	cudnnStatus_t t = cudnnCreate(&handle);
	cout << cudnnGetErrorString(t);
	getchar();
}

错误 MSB3721 命令“"D:\CUDA9.2\Development\bin\nvcc.exe" -gencode=arch=compute_30,code="sm_30,compute_30" --use-local-env --cl-version 2015 -ccbin "C:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\bin" -x cu -ID:\CUDA9.2\Development\include -ID:\CUDA9.2\Development\include -G --keep-dir Debug -maxrregcount=0 --machine 32 --compile -cudart static -g -D_MBCS -Xcompiler "/EHsc /W3 /nologo /Od /FS /Zi /RTC1 /MDd " -o Debug\test.cu.obj "C:\Users\24346\Documents\C++\testCUDA\test_cudnn\test_cudnn\test.cu"”已退出，返回代码为 1。

Microsoft Visual C++ 2017 Redistributable(X64)已安装
python 是 3.6.8 版本的 64 位，pip install tensorflow-gpu 显示安装成功，但 import tensorflow 报错：

ImportError: Traceback (most recent call last):
  File "D:\anaconda\envs\tensorflow-gpu\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 18, in swig_import_helper
    return importlib.import_module(mname)
  File "D:\anaconda\envs\tensorflow-gpu\lib\importlib\__init__.py", line 126, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
  File "<frozen importlib._bootstrap>", line 978, in _gcd_import
  File "<frozen importlib._bootstrap>", line 961, in _find_and_load
  File "<frozen importlib._bootstrap>", line 950, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 648, in _load_unlocked
  File "<frozen importlib._bootstrap>", line 560, in module_from_spec
  File "<frozen importlib._bootstrap_external>", line 922, in create_module
  File "<frozen importlib._bootstrap>", line 205, in _call_with_frames_removed
ImportError: DLL load failed: 找不到指定的模块。

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "D:\anaconda\envs\tensorflow-gpu\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 41, in <module>
    from tensorflow.python.pywrap_tensorflow_internal import *
  File "D:\anaconda\envs\tensorflow-gpu\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 21, in <module>
    _pywrap_tensorflow_internal = swig_import_helper()
  File "D:\anaconda\envs\tensorflow-gpu\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py", line 20, in swig_import_helper
    return importlib.import_module('_pywrap_tensorflow_internal')
  File "D:\anaconda\envs\tensorflow-gpu\lib\importlib\__init__.py", line 126, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
ModuleNotFoundError: No module named '_pywrap_tensorflow_internal'


Failed to load the native TensorFlow runtime.

See https://www.tensorflow.org/install/install_sources#common_installation_problems

for some common reasons and solutions.  Include the entire stack trace
above this error message when asking for help.

使用 anaconda 安装 tensorflow-gpu 也出现这个 ImportError
现在我安装了 tensorflow 的 cpu 版就没有这个问题了。。。
想问一下 10000 条以内评论的词向量训练需要显卡加速吗