-
Notifications
You must be signed in to change notification settings - Fork 74k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Build Pip Package Failed on Windows v1.8.0 with GPU #19103
Comments
I have similar issue as well, except my errors are in english
|
@gunan could you take a look. |
It is possible we have missed a bug with |
I try to build v1.8 branch, but same error is occured. |
What is generating this file? I can see if that part is working or not. |
In v.1.7.0 case. generated file is "tensorflow_gpu-1.7.0-cp36-cp36m-win_amd64.whl" |
Is this also the reason why there are no new windows builds for tf-nightly right now? |
No, on windows our CI has another issue we are working to solve right now.
…On Sun, May 13, 2018 at 12:49 AM Julian Niedermeier < ***@***.***> wrote:
Is this also the reason why there are no new windows builds for tf-nightly
right now?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#19103 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AHlCOXkkXPDmhvewa5QKxXmEOHnBIGemks5tx-WNgaJpZM4TzfMK>
.
|
Same error here. Windows 10. VS 17. Master branch.
|
Could you share the full log using pastebin? |
Looks like it is too big for pastebin. So I attach the log here. Thanks! @gunan |
What you are seeing is different. In your case, you are running into this failure. You may want to try the "Release" configuration. |
Thank you @gunan . Using 'release' did remove all C2471. However C1083 is still there. Please find the attached log. Anyway, since I just need the header files to build an user_opr under windows. I use branch r1.7 instead. |
Hi AshimaChawla
I download cudnn-9.1-windows10-x64-v7.1.zip, and extract this zip file.
You can find cuda/bin/cudnn64_7.dll, So copy this file to CUDA bin folder.
In my case, copy it to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v9.1\bin
This works fine.
Please try it.
Best Regards.
From: AshimaChawla <notifications@github.com>
Sent: Thursday, June 7, 2018 3:48 AM
To: tensorflow/tensorflow <tensorflow@noreply.github.com>
Cc: Yoshihiro Yoshimura <yoshimura@yqs.jp>; Mention <mention@noreply.github.com>
Subject: Re: [tensorflow/tensorflow] Build Pip Package Failed on Windows v1.8.0 with GPU (#19103)
hi @yoshimur <https://github.com/yoshimur> , were u able to use tf after installing 1.7 version??
I am getting the below error:
Could not find 'cudnn64_7.1.4.dll'. TensorFlow requires that this DLL be installed in a directory that is named in your %PATH% environment variable. Note that installing cuDNN is a separate step from installing CUDA, and this DLL is often found in a different directory from the CUDA DLLs. You may install the necessary DLL by downloading cuDNN 7.1.4 from this URL: https://developer.nvidia.com/cudnn
Could you please guide.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub <#19103 (comment)> , or mute the thread <https://github.com/notifications/unsubscribe-auth/ABBUB87Z0c36FNr2ypStCapPLnSp7lwcks5t6CPsgaJpZM4TzfMK> . <https://github.com/notifications/beacon/ABBUB2MBOEs6nEAIfCozR-hRVd_WiHTsks5t6CPsgaJpZM4TzfMK.gif>
|
Hi @yoshimur, Thanks very much.. Everything is working fine now.. The only issue I am seeing is when I execute my code, it seem to be bit slower than I used to execute on GPU. Is it something because of Windows ?? Is Ubuntu much faster than that? /Ashima |
Nagging Assignees @gunan, @shivaniag: It has been 15 days with no activity and this issue has an assignee. Please update the label and/or status accordingly. |
I had the same problem (Win10, VS2015, CUDA 9.1, AVX2, with MKL, without gRPC). I checked tensorflow/python/framework and saw that the cpp_shape_inference.pb.h file was in there. So I just ran the MSBuild command again and the build completed (WHL file produced)! Lmk if I should include more info. This is my first post to a Github issue, plz be nice. :) |
Unfortunately, I don't have the build output from the failed build anymore, I didn't save it anywhere and the command line only saves so much history. Many apologies. It definitely failed on tf_c_python_api.vcxproj because of "no such file or directory" of file cpp_shape_inference.pb.h, which clearly existed at the indicated location. |
System information
git pull
git checkout refs/tags/v1.8.0
cd tensorflow\contrib\cmake
mkdir build
cd build
cmake .. -A x64 -DCMAKE_BUILD_TYPE=Release ^
-DSWIG_EXECUTABLE=C:\swigwin-3.0.12\swig.exe ^
-DPYTHON_EXECUTABLE=C:\Anaconda3\python.exe ^
-DPYTHON_LIBRARIES=C:\Anaconda3\libs\python36.lib ^
-Dtensorflow_ENABLE_GPU=ON ^
-DCUDNN_HOME="%CUDA_PATH%" ^
-Dtensorflow_CUDA_VERSION=9.1 ^
-Dtensorflow_WIN_CPU_SIMD_OPTIONS=/arch:AVX2
MSBuild /p:Configuration=Release tf_tutorials_example_trainer.vcxproj
MSBuild /p:Configuration=Release tf_python_build_pip_package.vcxproj
After this command, following error was occured. Sorry, this is Japanese System message.
But, problem is seems missing the "tensorflow/python/framework/cpp_shape_inference.pb.h"
"C:\repo\tensorflow\tensorflow\contrib\cmake\build\tf_python_build_pip_package.
vcxproj" (既定のターゲット) (1) ->
"C:\repo\tensorflow\tensorflow\contrib\cmake\build\pywrap_tensorflow_internal.v
cxproj" (既定のターゲット) (2) ->
"C:\repo\tensorflow\tensorflow\contrib\cmake\build\pywrap_tensorflow_internal_s
tatic.vcxproj" (既定のターゲット) (3) ->
"C:\repo\tensorflow\tensorflow\contrib\cmake\build\tf_c_python_api.vcxproj" (既 定
のターゲット) (134) ->
(ClCompile ターゲット) ->
C:\repo\tensorflow\tensorflow\c\python_api.cc(19): fatal error C1083: include
ファイルを開けません。'tensorflow/python/framework/cpp_shape_inference.pb.h':No such file
or directory [C:\repo\tensorflow\tensorflow\contrib\cmake\build\tf_c_python_ap
i.vcxproj]
Best regards.
The text was updated successfully, but these errors were encountered: