I wanna run tf in win7-x64 platform. My env is win7-64, vs2015, python3.5.4,tensorflow 1.12.0(?). Now I have installed cuda10 and cudnn7.3 successfully. But when I run tensorflow(import tensorflow as tf), the error show the tensorflow can not be imported(not indicated DLL). Which Tensorflow-gpu version can be supported by CUDA10.0?
error message:
用户代码未处理 ImportError
Message: Traceback (most recent call last):
File “D:\Program Files\Python35\lib\site-packages\tensorflow\python\pywrap_tensorflow.py”, line 58, in
from tensorflow.python.pywrap_tensorflow_internal import *
File “D:\Program Files\Python35\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py”, line 28, in
_pywrap_tensorflow_internal = swig_import_helper()
File “D:\Program Files\Python35\lib\site-packages\tensorflow\python\pywrap_tensorflow_internal.py”, line 24, in swig_import_helper
_mod = imp.load_module(‘_pywrap_tensorflow_internal’, fp, pathname, description)
File “D:\Program Files\Python35\lib\imp.py”, line 243, in load_module
return load_dynamic(name, filename, file)
File “D:\Program Files\Python35\lib\imp.py”, line 343, in load_dynamic
return _load(spec)
ImportError: DLL load failed: 找不到指定的模块。
Failed to load the native TensorFlow runtime.
See Build and install error messages | TensorFlow
for some common reasons and solutions. Include the entire stack trace
above this error message when asking for help.
I don’t believe there are any pre-built TF versions (from official TF sources) that support CUDA 10 at the moment. That will probably change in the future.
You can build TF from sources to address this issue, but building TF from sources on windows may be challenging.
Sure I also think so. But yesterday I try to install tf on CUDA9.2, it is also failed, the same error. It seems I have to try to make/build to install it. Good luck for myself!
The current pre-built versions of TF from official sources require/use CUDA 9.0 only.