cudnn-10.1-windows10-x64-v7.6.5.32.zip适合环境为： win10 x64 cuda10.1

cudnn官网下载需要注册，比较麻烦，现在直接提供资源下载。window10下安装tensorflow1.1或以上所需要的cudnn，该cudnn版本对应CUDA8.0版本，文件解压后分别copy至CUDA\v8.0 中三个目录下：bin，include，lib即可，亲测可用

a debugger use vt technology 1.support windows 10 x64 only(recommend version:19042.1526 or higher) 2.you must download the pdb file (ntkrnlmp.pdb) corresponding to your windows version in specific path(C:\symbols) ,you can use ida or windbg to download pdb file automally. 3.your system must support vt technology 4.sign the drivers by yourself

共两个压缩包（解压后145M），此为其二 环境： windows10（X64） vs2015 Anaconda3-4.4.0-Windows-x86_64（python3.6） cuda_8.0.61_win10(basic) cuda_8.0.61.2_windows(Patch 2) cudnn-8.0 tensorflow1.30 pip install tensorflow-gpu后 Anaconda prompt中输入：import tensorflow时，报错：ImportError，需要用到这个库文件，分别copy至CUDA\v8.0 中三个目录下：bin，include，lib 具体错误一般如下： File "C:\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 18, in swig_import_helper return importlib.import_module(mname) File "C:\Anaconda3\lib\importlib\__init__.py", line 126, in import_module return _bootstrap._gcd_import(name[level:], package, level) File "", line 986, in _gcd_import File "", line 969, in _find_and_load File "", line 958, in _find_and_load_unlocked File "", line 666, in _load_unlocked File "", line 577, in module_from_spec File "", line 906, in create_module File "", line 222, in _call_with_frames_removed ImportError: DLL load failed: The specified module could not be found. During handling of the above exception, another exception occurred: Traceback (most recent call last): File "C:\Anaconda3\lib\site-packages\tensorflow\python\__init__.py", line 66, in from tensorflow.python import pywrap_tensorflow File "C:\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 21, in _pywrap_tensorflow = swig_import_helper() File "C:\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 20, in swig_import_helper return importlib.import_module('_pywrap_tensorflow') File "C:\Anaconda3\lib\importlib\__init__.py", line 126, in import_module return _bootstrap._gcd_import(name[level:], package, level) ImportError: No module named '_pywrap_tensorflow' During handling of the above exception, another exception occurr

共两个压缩包（解压后145M），此为其一 环境： windows10（X64） vs2015 Anaconda3-4.4.0-Windows-x86_64（python3.6） cuda_8.0.61_win10(basic) cuda_8.0.61.2_windows(Patch 2) cudnn-8.0 tensorflow1.30 pip install tensorflow-gpu后 Anaconda prompt中输入：import tensorflow时，报错：ImportError，需要用到这个库文件，分别copy至CUDA\v8.0 中三个目录下：bin，include，lib 具体错误一般如下： File "C:\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 18, in swig_import_helper return importlib.import_module(mname) File "C:\Anaconda3\lib\importlib\__init__.py", line 126, in import_module return _bootstrap._gcd_import(name[level:], package, level) File "", line 986, in _gcd_import File "", line 969, in _find_and_load File "", line 958, in _find_and_load_unlocked File "", line 666, in _load_unlocked File "", line 577, in module_from_spec File "", line 906, in create_module File "", line 222, in _call_with_frames_removed ImportError: DLL load failed: The specified module could not be found. During handling of the above exception, another exception occurred: Traceback (most recent call last): File "C:\Anaconda3\lib\site-packages\tensorflow\python\__init__.py", line 66, in from tensorflow.python import pywrap_tensorflow File "C:\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 21, in _pywrap_tensorflow = swig_import_helper() File "C:\Anaconda3\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 20, in swig_import_helper return importlib.import_module('_pywrap_tensorflow') File "C:\Anaconda3\lib\importlib\__init__.py", line 126, in import_module return _bootstrap._gcd_import(name[level:], package, level) ImportError: No module named '_pywrap_tensorflow' During handling of the above exception, another exception occurr

cuDNN官网下载较慢时，可使用该资源作为替代下载。 NVIDIA cuDNN是用于深度神经网络的GPU加速库。它强调性能、易用性和低内存开销。NVIDIA cuDNN可以集成到更高级别的机器学习框架中，如加州大学伯克利分校的流行CAFFE软件。简单的，插入式设计可以让开发人员专注于设计和实现神经网络模型，而不是调整性能，同时还可以在GPU上实现高性能现代并行计算。

cudnn-10.1-windows10-x64-v8.0.3.33.zip，适用于windows10下的CUDA10.1，其他版本可能不适配

运行GPU版本的tensorflow时cudnn升级报错

cudnn7.6

cudnn-10.1-windows10-x64-v7.6.5.32.7z