Discussion:
[theano-users] cannot compile with cuDNN error when importing theano
Qinpeng Wang
2018-03-30 17:27:22 UTC
Permalink
Hi, all,


I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0 (Windows7).
I created a new virtual environment with conda and installed required
libraries like this:

conda install numpy scipy mkl-service libpython m2w64-toolchain
conda install theano pygpu

Here are versions of major relavant libraries:

theano 1.0.1
python 3.5.5
numpy 1.14.2
scipy 1.0.0
m2w64-toolchain 5.3.0
mkl-service 1.1.2
pygpu 0.7.5
libgpuarray 0.7.5
libpython 2.1
mkl 2018.0.2

The way I installed cuDNN is to copy files into corresponding folders
inside CUDA folder. Originally I installed CUDA at C:\Program Files\NVIDIA
GPU Computing Toolkit, then I copied the CUDA folder to C root.

When I run:
python -c "import theano"

I get this message:

ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last):
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 227, in <module>
use(config.device)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 214, in use
init_dev(device, preallocate=preallocate)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 121, in init_dev
dnn.dnn_available.msg)
RuntimeError: You enabled cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got this error:
b'C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program: No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status'

My theanoc file:

[global]
floatX = float32
device = cuda

[dnn]
enabled = True
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64

[blas]
ldflags = C:/Program Files/Anaconda3/envs/t/Library/bin mkl_rt

Can somebody help please?


Thanks!
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Pascal Lamblin
2018-03-31 02:37:55 UTC
Permalink
Your "ldflags" option in the "[blas]" section is not correct, it should
at least start with -I or -l, and there should be quotation marks around
the path since there is a space in "Program Files".

What happens if you do not include this line in the .theanorc?

Are you able to run `gcc -v` from inside the conda env, and what is the
output?

Also, it may seem stupid, but have you restarted your computer, or at
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0 (Windows7).
I created a new virtual environment with conda and installed required
|conda install numpy scipy mkl-service libpython m2w64-toolchain conda
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0 m2w64-toolchain 5.3.0
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython 2.1 mkl 2018.0.2 |
The way I installed cuDNN is to copy files into corresponding folders
inside CUDA folder. Originally I installed CUDA at C:\Program
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA folder to C root.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit status' |
|[global] floatX = float32 device = cuda [dnn] enabled = True
include_path=C:/CUDA/v7.5/include library_path=C:/CUDA/v7.5/lib/x64
[blas] ldflags = C:/Program Files/Anaconda3/envs/t/Library/bin mkl_rt |
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Qinpeng Wang
2018-03-31 03:16:24 UTC
Permalink
Hi Pascal,

Thanks for your reply!! Indeed I wasn't really sure how to set the ldflags
parameter, but if I don't have this line, I get this warning message:

WARNING (theano.tensor.blas): Using NumPy C-API based implementation for
BLAS functions.

After inspecting theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>,
I found the following:

if not config.blas.ldflags:
logger.warning('Using NumPy C-API based implementation for BLAS functions.')

So I realize I may have to have ldflags in the theaorc file to get rid of
this warning message. *How to set ldflags correctly?*

I was asking around what the correct output of "gcc -v" would look like,
but have not got any reply yet, here it is:

Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\ Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Target: x86_64-w64-mingw32
Configured with: ../gcc-5.3.0/configure --prefix=/mingw64 --with-local-prefix=/mingw64/local --build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32 --target=x86_64-w64-mingw32 --with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include --libexecdir=/mingw64/lib --with-gxx-include-dir=/mingw64/include/c++/5.3.0 --enable-bootstrap --with-arch=x86-64 --with-tune=generic --enable-languages=c,lto,c++,objc,obj-c++,fortran,ada --enable-shared --enable-static --enable-libatomic --enable-threads=posix --enable-graphite --enable-fully-dynamic-string --enable-libstdcxx-time=yes --disable-libstdcxx-pch --disable-libstdcxx-debug --enable-version-specific-runtime-libs --disable-isl-version-check --enable-lto --enable-libgomp --disable-multilib --enable-checking=release --disable-rpath --disable-win32-registry --disable-nls --disable-werror --disable-symvers --with-libiconv --with-system-zlib --with-gmp=/mingw64 --with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64 --with-pkgversion='Rev5, Built by MSYS2 project' --with-bugurl=https://sourceforge.net/projects/msys2 --with-gnu-as --with-gnu-ld
Thread model: posix
gcc version 5.3.0 (Rev5, Built by MSYS2 project)


Lastly, I have restarted the computer, the error message stayed the same.

Thank you so much!
Post by Pascal Lamblin
Your "ldflags" option in the "[blas]" section is not correct, it should
at least start with -I or -l, and there should be quotation marks around
the path since there is a space in "Program Files".
What happens if you do not include this line in the .theanorc?
Are you able to run `gcc -v` from inside the conda env, and what is the
output?
Also, it may seem stupid, but have you restarted your computer, or at
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0 (Windows7).
I created a new virtual environment with conda and installed required
|conda install numpy scipy mkl-service libpython m2w64-toolchain conda
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0 m2w64-toolchain
5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython 2.1 mkl
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into corresponding folders
inside CUDA folder. Originally I installed CUDA at C:\Program
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA folder to C
root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled = True
include_path=C:/CUDA/v7.5/include library_path=C:/CUDA/v7.5/lib/x64
[blas] ldflags = C:/Program Files/Anaconda3/envs/t/Library/bin mkl_rt |
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Pascal Lamblin
2018-04-03 14:28:28 UTC
Permalink
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set the
ldflags parameter, but if I don't have this line, I get this warning
WARNING (theano.tensor.blas): Using NumPy C-API based implementation for
BLAS functions.
This is just a warning, the worse issue would be a slight slow down in
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>,
|if not config.blas.ldflags: logger.warning('Using NumPy C-API based
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to get rid
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is, when
it is not found automatically.
If you do not have an optimized BLAS library installed, then it should
stay empty.

Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that will
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Target: x86_64-w64-mingw32 Configured with: ../gcc-5.3.0/configure
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0 --enable-bootstrap
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada --enable-shared
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib --with-gmp=/mingw64
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2 --with-gnu-as
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5, Built by
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed the same.
Thank you so much!
Your "ldflags" option in the "[blas]" section is not correct, it should
at least start with -I or -l, and there should be quotation marks around
the path since there is a space in "Program Files".
What happens if you do not include this line in the .theanorc?
Are you able to run `gcc -v` from inside the conda env, and what is the
output?
Also, it may seem stupid, but have you restarted your computer, or at
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and installed
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython m2w64-toolchain
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython 2.1 mkl
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into corresponding
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at C:\Program
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA folder
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with cuDNN.
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled = True
include_path=C:/CUDA/v7.5/include library_path=C:/CUDA/v7.5/lib/x64
[blas] ldflags = C:/Program Files/Anaconda3/envs/t/Library/bin
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Qinpeng Wang
2018-04-03 14:46:35 UTC
Permalink
Hi Pascal,

Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable to
locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me figure
out how to get rid of the error message in my original post? I attach it
here again for your convenience.

When I run:
python -c "import theano"

I get this message:

ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last):
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 227, in <module>
use(config.device)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 214, in use
init_dev(device, preallocate=preallocate)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 121, in init_dev
dnn.dnn_available.msg)
RuntimeError: You enabled cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got this error:
b'C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program: No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status'


Thanks!
Post by Pascal Lamblin
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set the
ldflags parameter, but if I don't have this line, I get this warning
WARNING (theano.tensor.blas): Using NumPy C-API based implementation for
BLAS functions.
This is just a warning, the worse issue would be a slight slow down in
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>,
Post by Qinpeng Wang
|if not config.blas.ldflags: logger.warning('Using NumPy C-API based
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to get rid
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is, when
it is not found automatically.
If you do not have an optimized BLAS library installed, then it should
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that will
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Post by Qinpeng Wang
Target: x86_64-w64-mingw32 Configured with: ../gcc-5.3.0/configure
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0 --enable-bootstrap
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada --enable-shared
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib --with-gmp=/mingw64
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2 --with-gnu-as
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5, Built by
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed the
same.
Post by Qinpeng Wang
Thank you so much!
Your "ldflags" option in the "[blas]" section is not correct, it
should
Post by Qinpeng Wang
at least start with -I or -l, and there should be quotation marks around
the path since there is a space in "Program Files".
What happens if you do not include this line in the .theanorc?
Are you able to run `gcc -v` from inside the conda env, and what is
the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted your computer, or
at
Post by Qinpeng Wang
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and installed
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython m2w64-toolchain
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython 2.1 mkl
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into corresponding
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at C:\Program
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA folder
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with cuDNN.
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled = True
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program Files/Anaconda3/envs/t/Library/bin
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
<javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Pascal Lamblin
2018-04-03 14:59:32 UTC
Permalink
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.

Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable
to locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me
figure out how to get rid of the error message in my original post? I
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit status' |
Thanks!
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set the
ldflags parameter, but if I don't have this line, I get this warning
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a slight slow down in
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>,
Post by Qinpeng Wang
|if not config.blas.ldflags: logger.warning('Using NumPy C-API based
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to
get rid
Post by Qinpeng Wang
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is, when
it is not found automatically.
If you do not have an optimized BLAS library installed, then it should
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that will
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Post by Qinpeng Wang
Target: x86_64-w64-mingw32 Configured with: ../gcc-5.3.0/configure
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0 --enable-bootstrap
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2> --with-gnu-as
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5, Built by
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed
the same.
Post by Qinpeng Wang
Thank you so much!
    Your "ldflags" option in the "[blas]" section is not correct,
it should
Post by Qinpeng Wang
    at least start with -I or -l, and there should be quotation
marks
Post by Qinpeng Wang
    around
    the path since there is a space in "Program Files".
    What happens if you do not include this line in the .theanorc?
    Are you able to run `gcc -v` from inside the conda env, and
what is the
Post by Qinpeng Wang
    output?
    Also, it may seem stupid, but have you restarted your
computer, or at
Post by Qinpeng Wang
    least the terminal, since you installed m2w64-toolchain?
     > Hi, all,
     >
     >
     > I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0
    (Windows7).
     > I created a new virtual environment with conda and installed
    required
     >
     > |conda install numpy scipy mkl-service libpython
m2w64-toolchain
Post by Qinpeng Wang
    conda
     > install theano pygpu |
     >
     >
     > |theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
    m2w64-toolchain 5.3.0
     > mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython
2.1 mkl
Post by Qinpeng Wang
    2018.0.2 |
     >
     > The way I installed cuDNN is to copy files into corresponding
    folders
     > inside CUDA folder. Originally I installed CUDA at C:\Program
     > Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA
folder
Post by Qinpeng Wang
    to C root.
     >
     > |python -c "import theano"|
     >
     >
     > |ERROR (theano.gpuarray): Could not initialize pygpu, support
    disabled
     > Traceback (most recent call last): File "C:\Program
     >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
     > line 227, in <module> use(config.device) File "C:\Program
     >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
     > line 214, in use init_dev(device, preallocate=preallocate)
File
Post by Qinpeng Wang
     > "C:\Program
     >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
You
Post by Qinpeng Wang
    enabled
     > cuDNN, but we aren't able to use it: cannot compile with
cuDNN.
Post by Qinpeng Wang
    We got
     > this error: b'C:/Program
     >
     > cannot find C:/Program: No such file or directory C:/Program
     >
     > cannot find
     >
     > No such file or directory collect2.exe: error: ld returned
1 exit
Post by Qinpeng Wang
    status' |
     >
     >
     > |[global] floatX = float32 device = cuda [dnn] enabled = True
     > include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
     > [blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
    mkl_rt |
     >
     > Can somebody help please?
     >
     >
     > Thanks!
     >
     > --
     >
     > ---
     > You received this message because you are subscribed to
the Google
Post by Qinpeng Wang
     > Groups "theano-users" group.
     > To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
    send
<javascript:> <javascript:>>.
Post by Qinpeng Wang
     > For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
    <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
    --
    Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Qinpeng Wang
2018-04-03 16:01:26 UTC
Permalink
Hi Pascal,

I have CUDA installed a while ago when I was still using the old GPU
backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then I copied the
CUDA folder to C root.

Then I installed cuDNN by copying unzipped files into corresponding folders
inside CUDA folder. I'm using CUDA 7.5 and cuDNN 6.0 (Windows7). See my
theanoc file in my original post to make sure I have pointed theano to the
right location.
Yes this error is not related to Idflags.

Thanks!
Post by Pascal Lamblin
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable
to locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me
figure out how to get rid of the error message in my original post? I
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status' |
Post by Qinpeng Wang
Thanks!
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set
the
Post by Qinpeng Wang
Post by Qinpeng Wang
ldflags parameter, but if I don't have this line, I get this
warning
Post by Qinpeng Wang
Post by Qinpeng Wang
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a slight slow down
in
Post by Qinpeng Wang
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>,
Post by Qinpeng Wang
Post by Qinpeng Wang
|if not config.blas.ldflags: logger.warning('Using NumPy C-API
based
Post by Qinpeng Wang
Post by Qinpeng Wang
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to
get rid
Post by Qinpeng Wang
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is,
when
Post by Qinpeng Wang
it is not found automatically.
If you do not have an optimized BLAS library installed, then it
should
Post by Qinpeng Wang
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that
will
Post by Qinpeng Wang
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
../gcc-5.3.0/configure
Post by Qinpeng Wang
Post by Qinpeng Wang
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
Post by Qinpeng Wang
Post by Qinpeng Wang
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0
--enable-bootstrap
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2> --with-gnu-as
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5, Built
by
Post by Qinpeng Wang
Post by Qinpeng Wang
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed
the same.
Post by Qinpeng Wang
Thank you so much!
On Friday, March 30, 2018 at 9:38:11 PM UTC-5, Pascal Lamblin
Your "ldflags" option in the "[blas]" section is not correct,
it should
Post by Qinpeng Wang
at least start with -I or -l, and there should be quotation
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program Files".
What happens if you do not include this line in the
.theanorc?
Post by Qinpeng Wang
Post by Qinpeng Wang
Are you able to run `gcc -v` from inside the conda env, and
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted your
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN
6.0
Post by Qinpeng Wang
Post by Qinpeng Wang
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and
installed
Post by Qinpeng Wang
Post by Qinpeng Wang
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into
corresponding
Post by Qinpeng Wang
Post by Qinpeng Wang
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at
C:\Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu,
support
Post by Qinpeng Wang
Post by Qinpeng Wang
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate)
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
You
Post by Qinpeng Wang
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory
C:/Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
cannot find
No such file or directory collect2.exe: error: ld returned
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled =
True
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
send
<javascript:>
<javascript:>
Post by Qinpeng Wang
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
<javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Arnaud Bergeron
2018-04-10 17:45:46 UTC
Permalink
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the old GPU backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA folder to C root.
This most likely broke DLL registrations. Please don't do that.

If you reinstall cuda in its original location, put the cudnn files in the cuda directory and remove these two parameters in your theanorc file:
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
It should work again.

---

I you really want to try with the current setup, you can try to set this
base_path=C:/CUDA/v7.5
Instead of the two above, but I'm pretty sure that will still break on the cuda DLLs.
Post by Qinpeng Wang
Then I installed cuDNN by copying unzipped files into corresponding folders inside CUDA folder. I'm using CUDA 7.5 and cuDNN 6.0 (Windows7). See my theanoc file in my original post to make sure I have pointed theano to the right location.
Yes this error is not related to Idflags.
Thanks!
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable
to locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me
figure out how to get rid of the error message in my original post? I
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit status' |
Thanks!
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set the
ldflags parameter, but if I don't have this line, I get this warning
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a slight slow down in
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py <https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py <https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>>,
Post by Qinpeng Wang
|if not config.blas.ldflags: logger.warning('Using NumPy C-API based
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to
get rid
Post by Qinpeng Wang
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is, when
it is not found automatically.
If you do not have an optimized BLAS library installed, then it should
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that will
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Post by Qinpeng Wang
Target: x86_64-w64-mingw32 Configured with: ../gcc-5.3.0/configure
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0 --enable-bootstrap
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2 <https://sourceforge.net/projects/msys2>
<https://sourceforge.net/projects/msys2 <https://sourceforge.net/projects/msys2>> --with-gnu-as
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5, Built by
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed
the same.
Post by Qinpeng Wang
Thank you so much!
Your "ldflags" option in the "[blas]" section is not correct,
it should
Post by Qinpeng Wang
at least start with -I or -l, and there should be quotation
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program Files".
What happens if you do not include this line in the .theanorc?
Are you able to run `gcc -v` from inside the conda env, and
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted your
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and installed
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into corresponding
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at C:\Program
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate)
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
You
Post by Qinpeng Wang
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled = True
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
send
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>
Post by Qinpeng Wang
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>>.
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Qinpeng Wang
2018-04-10 21:19:26 UTC
Permalink
Hi, Arnaud,

Thanks for your reply! I reinstalled CUDA, copied cudnn files in the cuda
directory, and removed the two parameters. However, I got the same errors
from my first post:

ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last):
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 227, in <module>
use(config.device)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 214, in use
init_dev(device, preallocate=preallocate)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 121, in init_dev
dnn.dnn_available.msg)
RuntimeError: You enabled cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got this error:
b'C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program: No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status'


Any other suggestion you have?

Thanks!
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the old GPU
backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then I copied
the CUDA folder to C root.
This most likely broke DLL registrations. Please don't do that.
If you reinstall cuda in its original location, put the cudnn files in the
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
It should work again.
---
I you really want to try with the current setup, you can try to set this
base_path=C:/CUDA/v7.5
Instead of the two above, but I'm pretty sure that will still break on the cuda DLLs.
Then I installed cuDNN by copying unzipped files into corresponding
folders inside CUDA folder. I'm using CUDA 7.5 and cuDNN 6.0 (Windows7).
See my theanoc file in my original post to make sure I have pointed theano
to the right location.
Yes this error is not related to Idflags.
Thanks!
Post by Pascal Lamblin
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable
to locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me
figure out how to get rid of the error message in my original post? I
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status' |
Post by Qinpeng Wang
Thanks!
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set
the
Post by Qinpeng Wang
Post by Qinpeng Wang
ldflags parameter, but if I don't have this line, I get this
warning
Post by Qinpeng Wang
Post by Qinpeng Wang
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a slight slow down
in
Post by Qinpeng Wang
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
Post by Qinpeng Wang
,
|if not config.blas.ldflags: logger.warning('Using NumPy C-API
based
Post by Qinpeng Wang
Post by Qinpeng Wang
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to
get rid
Post by Qinpeng Wang
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is,
when
Post by Qinpeng Wang
it is not found automatically.
If you do not have an optimized BLAS library installed, then it
should
Post by Qinpeng Wang
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that
will
Post by Qinpeng Wang
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
../gcc-5.3.0/configure
Post by Qinpeng Wang
Post by Qinpeng Wang
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
Post by Qinpeng Wang
Post by Qinpeng Wang
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0
--enable-bootstrap
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2> --with-gnu-as
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5, Built
by
Post by Qinpeng Wang
Post by Qinpeng Wang
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed
the same.
Post by Qinpeng Wang
Thank you so much!
On Friday, March 30, 2018 at 9:38:11 PM UTC-5, Pascal Lamblin
Your "ldflags" option in the "[blas]" section is not correct,
it should
Post by Qinpeng Wang
at least start with -I or -l, and there should be quotation
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program Files".
What happens if you do not include this line in the
.theanorc?
Post by Qinpeng Wang
Post by Qinpeng Wang
Are you able to run `gcc -v` from inside the conda env, and
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted your
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN
6.0
Post by Qinpeng Wang
Post by Qinpeng Wang
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and
installed
Post by Qinpeng Wang
Post by Qinpeng Wang
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into
corresponding
Post by Qinpeng Wang
Post by Qinpeng Wang
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at
C:\Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu,
support
Post by Qinpeng Wang
Post by Qinpeng Wang
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate)
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
You
Post by Qinpeng Wang
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory
C:/Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
cannot find
No such file or directory collect2.exe: error: ld returned
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled =
True
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
send
<javascript:>
Post by Qinpeng Wang
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit
https://groups.google.com/d/optout
Post by Qinpeng Wang
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
<javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Pascal Lamblin
2018-04-10 21:58:23 UTC
Permalink
You _also_ need to remove the incorrect "blas.ldflags" line from .theanorc.
Post by Qinpeng Wang
Hi, Arnaud,
Thanks for your reply! I reinstalled CUDA, copied cudnn files in the
cuda directory, and removed the two parameters. However, I got the same
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit status'|
Any other suggestion you have?
Thanks!
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the old
GPU backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then
I copied the CUDA folder to C root.
This most likely broke DLL registrations.  Please don't do that.
If you reinstall cuda in its original location, put the cudnn files
in the cuda directory and remove these two parameters in your
|include_path=C:/CUDA/v7.5/include library_path=C:/CUDA/v7.5/lib/x64|
It should work again.
---
I you really want to try with the current setup, you can try to set this
|base_path=C:/CUDA/v7.5 |
Instead of the two above, but I'm pretty sure that will still break
on the cuda DLLs.
Post by Qinpeng Wang
Then I installed cuDNN by copying unzipped files into
corresponding folders inside CUDA folder. I'm using CUDA 7.5 and
cuDNN 6.0 (Windows7). See my theanoc file in my original post to
make sure I have pointed theano to the right location.
Yes this error is not related to Idflags.
Thanks!
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have
versions of
Post by Qinpeng Wang
installed libraries in my original post), but somehow theano
is unable
Post by Qinpeng Wang
to locate the BLAS library automatically, hence the warning
message.
Post by Qinpeng Wang
Since the "gcc-v" output seems reasonable, could you please
help me
Post by Qinpeng Wang
figure out how to get rid of the error message in my
original post? I
Post by Qinpeng Wang
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu,
support disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
You enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN. We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1
exit status' |
Post by Qinpeng Wang
Thanks!
On Tuesday, April 3, 2018 at 9:28:37 AM UTC-5, Pascal
     > Hi Pascal,
     >
     > Thanks for your reply!! Indeed I wasn't really sure
how to set the
Post by Qinpeng Wang
     > ldflags parameter, but if I don't have this line, I
get this warning
Post by Qinpeng Wang
     >
     > WARNING (theano.tensor.blas): Using NumPy C-API based
    implementation for
     > BLAS functions.
    This is just a warning, the worse issue would be a
slight slow down in
Post by Qinpeng Wang
    some CPU operations.
     > After inspecting theano/tensor/blas_headers.py
     >
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>>,
Post by Qinpeng Wang
     >
     > |if not config.blas.ldflags: logger.warning('Using
NumPy C-API based
Post by Qinpeng Wang
     > implementation for BLAS functions.')|
     >
     > So I realize I may have to have ldflags in the
theaorc file to
Post by Qinpeng Wang
    get rid
     > of this warning message. *How to set ldflags correctly?*
    These flags are for specifying where an optimized BLAS
library is, when
Post by Qinpeng Wang
    it is not found automatically.
    If you do not have an optimized BLAS library installed,
then it should
Post by Qinpeng Wang
    stay empty.
    Did you install the "mkl-service" package in conda?
    This is usually the easiest way to install an optimized
BLAS that will
Post by Qinpeng Wang
    be automatically found.
     > I was asking around what the correct output of "gcc
-v" would
Post by Qinpeng Wang
    look like,
    This seems reasonable.
     >
     > |Using built-in specs. COLLECT_GCC=gcc
    COLLECT_LTO_WRAPPER=C:/Program\
     >
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
../gcc-5.3.0/configure
Post by Qinpeng Wang
     > --prefix=/mingw64 --with-local-prefix=/mingw64/local
     > --build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
     > --target=x86_64-w64-mingw32
     >
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
Post by Qinpeng Wang
     > --libexecdir=/mingw64/lib
     > --with-gxx-include-dir=/mingw64/include/c++/5.3.0
--enable-bootstrap
Post by Qinpeng Wang
     > --with-arch=x86-64 --with-tune=generic
     > --enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
    --enable-shared
     > --enable-static --enable-libatomic --enable-threads=posix
     > --enable-graphite --enable-fully-dynamic-string
     > --enable-libstdcxx-time=yes --disable-libstdcxx-pch
     > --disable-libstdcxx-debug
--enable-version-specific-runtime-libs
Post by Qinpeng Wang
     > --disable-isl-version-check --enable-lto --enable-libgomp
     > --disable-multilib --enable-checking=release
--disable-rpath
Post by Qinpeng Wang
     > --disable-win32-registry --disable-nls --disable-werror
     > --disable-symvers --with-libiconv --with-system-zlib
    --with-gmp=/mingw64
     > --with-mpfr=/mingw64 --with-mpc=/mingw64
--with-isl=/mingw64
Post by Qinpeng Wang
     > --with-pkgversion='Rev5, Built by MSYS2 project'
     > --with-bugurl=https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>
Post by Qinpeng Wang
    <https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>> --with-gnu-as
Post by Qinpeng Wang
     > --with-gnu-ld Thread model: posix gcc version 5.3.0
(Rev5, Built by
Post by Qinpeng Wang
     > MSYS2 project)|
     >
     >
     > Lastly, I have restarted the computer, the error
message stayed
Post by Qinpeng Wang
    the same.
     >
     > Thank you so much!
     >
     > On Friday, March 30, 2018 at 9:38:11 PM UTC-5, Pascal
     >
     >     Your "ldflags" option in the "[blas]" section is
not correct,
Post by Qinpeng Wang
    it should
     >     at least start with -I or -l, and there should be
quotation
Post by Qinpeng Wang
    marks
     >     around
     >     the path since there is a space in "Program Files".
     >
     >     What happens if you do not include this line in
the .theanorc?
Post by Qinpeng Wang
     >
     >     Are you able to run `gcc -v` from inside the
conda env, and
Post by Qinpeng Wang
    what is the
     >     output?
     >
     >     Also, it may seem stupid, but have you restarted your
    computer, or at
     >     least the terminal, since you installed
m2w64-toolchain?
Post by Qinpeng Wang
     >
     >      > Hi, all,
     >      >
     >      >
     >      > I'm trying to setup theano 1.0.1 with CUDA 7.5
and cuDNN 6.0
Post by Qinpeng Wang
     >     (Windows7).
     >      > I created a new virtual environment with conda
and installed
Post by Qinpeng Wang
     >     required
     >      >
     >      > |conda install numpy scipy mkl-service libpython
    m2w64-toolchain
     >     conda
     >      > install theano pygpu |
     >      >
     >      >
     >      > |theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy
1.0.0
Post by Qinpeng Wang
     >     m2w64-toolchain 5.3.0
     >      > mkl-service 1.1.2 pygpu 0.7.5 libgpuarray
0.7.5 libpython
Post by Qinpeng Wang
    2.1 mkl
     >     2018.0.2 |
     >      >
     >      > The way I installed cuDNN is to copy files
into corresponding
Post by Qinpeng Wang
     >     folders
     >      > inside CUDA folder. Originally I installed
CUDA at C:\Program
Post by Qinpeng Wang
     >      > Files\NVIDIA GPU Computing Toolkit, then I
copied the CUDA
Post by Qinpeng Wang
    folder
     >     to C root.
     >      >
     >      > |python -c "import theano"|
     >      >
     >      >
     >      > |ERROR (theano.gpuarray): Could not initialize
pygpu, support
Post by Qinpeng Wang
     >     disabled
     >      > Traceback (most recent call last): File
"C:\Program
Post by Qinpeng Wang
     >      >
     >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
     >      > line 227, in <module> use(config.device) File
"C:\Program
Post by Qinpeng Wang
     >      >
     >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
     >      > line 214, in use init_dev(device,
preallocate=preallocate)
Post by Qinpeng Wang
    File
     >      > "C:\Program
     >      >
     >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
     >      > line 121, in init_dev dnn.dnn_available.msg)
    You
     >     enabled
     >      > cuDNN, but we aren't able to use it: cannot
compile with
Post by Qinpeng Wang
    cuDNN.
     >     We got
     >      > this error: b'C:/Program
     >      >
     >
     >
     >      > cannot find C:/Program: No such file or
directory C:/Program
Post by Qinpeng Wang
     >      >
     >
     >
     >      > cannot find
     >      >
     >
     >
ld returned
Post by Qinpeng Wang
    1 exit
     >     status' |
     >      >
     >      >
     >      > |[global] floatX = float32 device = cuda [dnn]
enabled = True
Post by Qinpeng Wang
     >      > include_path=C:/CUDA/v7.5/include
    library_path=C:/CUDA/v7.5/lib/x64
     >      > [blas] ldflags = C:/Program
    Files/Anaconda3/envs/t/Library/bin
     >     mkl_rt |
     >      >
     >      > Can somebody help please?
     >      >
     >      >
     >      > Thanks!
     >      >
     >      > --
     >      >
     >      > ---
     >      > You received this message because you are
subscribed to
Post by Qinpeng Wang
    the Google
     >      > Groups "theano-users" group.
     >      > To unsubscribe from this group and stop
receiving emails
Post by Qinpeng Wang
    from it,
     >     send
     >      > an email
    <javascript:> <javascript:>>.
     >      > For more options,
visithttps://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
    <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>
Post by Qinpeng Wang
     >     <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
    <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>>.
Post by Qinpeng Wang
     >
     >     --
     >     Pascal Lamblin
     >
     > --
     >
     > ---
     > You received this message because you are subscribed
to the Google
Post by Qinpeng Wang
     > Groups "theano-users" group.
     > To unsubscribe from this group and stop receiving
emails from it,
Post by Qinpeng Wang
    send
     >
     > For more options,
visithttps://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
    <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
    --
    Pascal Lamblin
--
---
You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it, send
<http://googlegroups.com/>
Post by Qinpeng Wang
For more options, visithttps://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
For more options, visithttps://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Qinpeng Wang
2018-04-11 12:56:53 UTC
Permalink
Hi Pascal,

I have got rid of that line a while ago following your advice, here is my
new theanoc file after Arnaud's advice:

[global]
floatX = float32
device = cuda

[dnn]
enabled = True

Thanks!
Post by Pascal Lamblin
You _also_ need to remove the incorrect "blas.ldflags" line from .theanorc.
Post by Qinpeng Wang
Hi, Arnaud,
Thanks for your reply! I reinstalled CUDA, copied cudnn files in the
cuda directory, and removed the two parameters. However, I got the same
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status'|
Post by Qinpeng Wang
Any other suggestion you have?
Thanks!
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the old
GPU backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then
I copied the CUDA folder to C root.
This most likely broke DLL registrations. Please don't do that.
If you reinstall cuda in its original location, put the cudnn files
in the cuda directory and remove these two parameters in your
|include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64|
Post by Qinpeng Wang
It should work again.
---
I you really want to try with the current setup, you can try to set
this
Post by Qinpeng Wang
|base_path=C:/CUDA/v7.5 |
Instead of the two above, but I'm pretty sure that will still break
on the cuda DLLs.
Post by Qinpeng Wang
Then I installed cuDNN by copying unzipped files into
corresponding folders inside CUDA folder. I'm using CUDA 7.5 and
cuDNN 6.0 (Windows7). See my theanoc file in my original post to
make sure I have pointed theano to the right location.
Yes this error is not related to Idflags.
Thanks!
On Tuesday, April 3, 2018 at 9:59:38 AM UTC-5, Pascal Lamblin
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have
versions of
Post by Qinpeng Wang
installed libraries in my original post), but somehow theano
is unable
Post by Qinpeng Wang
to locate the BLAS library automatically, hence the warning
message.
Post by Qinpeng Wang
Since the "gcc-v" output seems reasonable, could you please
help me
Post by Qinpeng Wang
figure out how to get rid of the error message in my
original post? I
Post by Qinpeng Wang
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu,
support disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate)
File
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
You enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN. We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1
exit status' |
Post by Qinpeng Wang
Thanks!
On Tuesday, April 3, 2018 at 9:28:37 AM UTC-5, Pascal
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure
how to set the
Post by Qinpeng Wang
Post by Qinpeng Wang
ldflags parameter, but if I don't have this line, I
get this warning
Post by Qinpeng Wang
Post by Qinpeng Wang
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a
slight slow down in
Post by Qinpeng Wang
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
Post by Qinpeng Wang
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>>,
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
|if not config.blas.ldflags: logger.warning('Using
NumPy C-API based
Post by Qinpeng Wang
Post by Qinpeng Wang
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the
theaorc file to
Post by Qinpeng Wang
get rid
Post by Qinpeng Wang
of this warning message. *How to set
ldflags correctly?*
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
These flags are for specifying where an optimized BLAS
library is, when
Post by Qinpeng Wang
it is not found automatically.
If you do not have an optimized BLAS library installed,
then it should
Post by Qinpeng Wang
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized
BLAS that will
Post by Qinpeng Wang
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc
-v" would
Post by Qinpeng Wang
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Post by Qinpeng Wang
Post by Qinpeng Wang
../gcc-5.3.0/configure
Post by Qinpeng Wang
Post by Qinpeng Wang
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0
--enable-bootstrap
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic
--enable-threads=posix
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug
--enable-version-specific-runtime-libs
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-isl-version-check --enable-lto
--enable-libgomp
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-multilib --enable-checking=release
--disable-rpath
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-win32-registry --disable-nls
--disable-werror
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64
--with-isl=/mingw64
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>
Post by Qinpeng Wang
<https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>> --with-gnu-as
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0
(Rev5, Built by
Post by Qinpeng Wang
Post by Qinpeng Wang
MSYS2 project)|
Lastly, I have restarted the computer, the error
message stayed
Post by Qinpeng Wang
the same.
Post by Qinpeng Wang
Thank you so much!
On Friday, March 30, 2018 at 9:38:11 PM UTC-5, Pascal
Your "ldflags" option in the "[blas]" section is
not correct,
Post by Qinpeng Wang
it should
Post by Qinpeng Wang
at least start with -I or -l, and there should be
quotation
Post by Qinpeng Wang
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program
Files".
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
What happens if you do not include this line in
the .theanorc?
Post by Qinpeng Wang
Post by Qinpeng Wang
Are you able to run `gcc -v` from inside the
conda env, and
Post by Qinpeng Wang
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted
your
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed
m2w64-toolchain?
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5
and cuDNN 6.0
Post by Qinpeng Wang
Post by Qinpeng Wang
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda
and installed
Post by Qinpeng Wang
Post by Qinpeng Wang
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service
libpython
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy
1.0.0
Post by Qinpeng Wang
Post by Qinpeng Wang
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray
0.7.5 libpython
Post by Qinpeng Wang
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files
into corresponding
Post by Qinpeng Wang
Post by Qinpeng Wang
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed
CUDA at C:\Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Files\NVIDIA GPU Computing Toolkit, then I
copied the CUDA
Post by Qinpeng Wang
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize
pygpu, support
Post by Qinpeng Wang
Post by Qinpeng Wang
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device,
preallocate=preallocate)
Post by Qinpeng Wang
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 121, in init_dev dnn.dnn_available.msg)
You
Post by Qinpeng Wang
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot
compile with
Post by Qinpeng Wang
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or
directory C:/Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
cannot find
ld returned
Post by Qinpeng Wang
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn]
enabled = True
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are
subscribed to
Post by Qinpeng Wang
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop
receiving emails
Post by Qinpeng Wang
from it,
Post by Qinpeng Wang
send
Post by Qinpeng Wang
an email
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options,
visithttps://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed
to the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving
emails from it,
Post by Qinpeng Wang
send
Post by Qinpeng Wang
For more options,
visithttps://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it, send
<http://googlegroups.com/>
<javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visithttps://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
For more options, visithttps://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Pascal Lamblin
2018-04-11 17:01:41 UTC
Permalink
And you _still_ have the following in your error message?
Post by Qinpeng Wang
cannot find C:/Program: No such file or directory
That is surprising, I don't see where it would get "C:/Program" from.
Post by Qinpeng Wang
Hi Pascal,
I have got rid of that line a while ago following your advice, here is
[global]
floatX = float32
device = cuda
[dnn]
enabled = True
Thanks!
You _also_ need to remove the incorrect "blas.ldflags" line from .theanorc.
Post by Qinpeng Wang
Hi, Arnaud,
Thanks for your reply! I reinstalled CUDA, copied cudnn files in the
cuda directory, and removed the two parameters. However, I got
the same
Post by Qinpeng Wang
|ERROR (theano.gpuarray): Could not initialize pygpu, support
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with cuDNN.
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status'|
Post by Qinpeng Wang
Any other suggestion you have?
Thanks!
On Tuesday, April 10, 2018 at 12:45:59 PM UTC-5, Arnaud Bergeron
    Hi Pascal,
    I have CUDA installed a while ago when I was still using the
old
Post by Qinpeng Wang
    GPU backend at C:\Program Files\NVIDIA GPU Computing
Toolkit, then
Post by Qinpeng Wang
    I copied the CUDA folder to C root.
    This most likely broke DLL registrations.  Please don't do that.
    If you reinstall cuda in its original location, put the cudnn
files
Post by Qinpeng Wang
    in the cuda directory and remove these two parameters in your
    |include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64|
Post by Qinpeng Wang
    It should work again.
    ---
    I you really want to try with the current setup, you can try
to set this
Post by Qinpeng Wang
    |base_path=C:/CUDA/v7.5 |
    Instead of the two above, but I'm pretty sure that will still
break
Post by Qinpeng Wang
    on the cuda DLLs.
    Then I installed cuDNN by copying unzipped files into
    corresponding folders inside CUDA folder. I'm using CUDA 7.5
and
Post by Qinpeng Wang
    cuDNN 6.0 (Windows7). See my theanoc file in my original
post to
Post by Qinpeng Wang
    make sure I have pointed theano to the right location.
    Yes this error is not related to Idflags.
    Thanks!
    On Tuesday, April 3, 2018 at 9:59:38 AM UTC-5, Pascal
        This is a completely different error message.
        This one has to do with cuDNN, not BLAS or MKL.
        Where have you installed cuDNN?
        Does _that_ error still happen when you do not ldflags?
        > Hi Pascal,
        >
        > Thanks for your reply!
        > I have "mkl-service" package installed via conda (I have
        versions of
        > installed libraries in my original post), but somehow
theano
Post by Qinpeng Wang
        is unable
        > to locate the BLAS library automatically, hence the
warning
Post by Qinpeng Wang
        message.
        > Since the "gcc-v" output seems reasonable, could you
please
Post by Qinpeng Wang
        help me
        > figure out how to get rid of the error message in my
        original post? I
        > attach it here again for your convenience.
        >
        > |python -c "import theano"|
        >
        >
        > |ERROR (theano.gpuarray): Could not initialize pygpu,
        support disabled
        > Traceback (most recent call last): File "C:\Program
        >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
        > line 227, in <module> use(config.device) File "C:\Program
        >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
        > line 214, in use init_dev(device,
preallocate=preallocate) File
Post by Qinpeng Wang
        > "C:\Program
        >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
        > line 121, in init_dev dnn.dnn_available.msg)
        You enabled
        > cuDNN, but we aren't able to use it: cannot compile with
        cuDNN. We got
        > this error: b'C:/Program
        >
        > cannot find C:/Program: No such file or directory
C:/Program
Post by Qinpeng Wang
        >
        > cannot find
        >
        > No such file or directory collect2.exe: error: ld
returned 1
Post by Qinpeng Wang
        exit status' |
        >
        >
        > Thanks!
        >
        >
        > On Tuesday, April 3, 2018 at 9:28:37 AM UTC-5, Pascal
        >
        >      > Hi Pascal,
        >      >
        >      > Thanks for your reply!! Indeed I wasn't really
sure
Post by Qinpeng Wang
        how to set the
        >      > ldflags parameter, but if I don't have this
line, I
Post by Qinpeng Wang
        get this warning
        >      >
        >      > WARNING (theano.tensor.blas): Using NumPy C-API
based
Post by Qinpeng Wang
        >     implementation for
        >      > BLAS functions.
        >
        >     This is just a warning, the worse issue would be a
        slight slow down in
        >     some CPU operations.
        >
        >      > After inspecting theano/tensor/blas_headers.py
        >      >
        >
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>
Post by Qinpeng Wang
        >
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>>>,
Post by Qinpeng Wang
        >
        >      >
        >      > |if not config.blas.ldflags: logger.warning('Using
        NumPy C-API based
        >      > implementation for BLAS functions.')|
        >      >
        >      > So I realize I may have to have ldflags in the
        theaorc file to
        >     get rid
        >      > of this warning message. *How to set
ldflags correctly?*
Post by Qinpeng Wang
        >
        >     These flags are for specifying where an optimized
BLAS
Post by Qinpeng Wang
        library is, when
        >     it is not found automatically.
        >     If you do not have an optimized BLAS library
installed,
Post by Qinpeng Wang
        then it should
        >     stay empty.
        >
        >     Did you install the "mkl-service" package in conda?
        >     This is usually the easiest way to install an
optimized
Post by Qinpeng Wang
        BLAS that will
        >     be automatically found.
        >
        >      > I was asking around what the correct output of
"gcc
Post by Qinpeng Wang
        -v" would
        >     look like,
        >
        >     This seems reasonable.
        >
        >      >
        >      > |Using built-in specs. COLLECT_GCC=gcc
        >     COLLECT_LTO_WRAPPER=C:/Program\
        >      >
        >
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Post by Qinpeng Wang
        >
        ../gcc-5.3.0/configure
        >      > --prefix=/mingw64
--with-local-prefix=/mingw64/local
Post by Qinpeng Wang
        >      > --build=x86_64-w64-mingw32
--host=x86_64-w64-mingw32
Post by Qinpeng Wang
        >      > --target=x86_64-w64-mingw32
        >      >
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
Post by Qinpeng Wang
        >      > --libexecdir=/mingw64/lib
        >      > --with-gxx-include-dir=/mingw64/include/c++/5.3.0
        --enable-bootstrap
        >      > --with-arch=x86-64 --with-tune=generic
        >      >
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
Post by Qinpeng Wang
        >     --enable-shared
        >      > --enable-static --enable-libatomic
--enable-threads=posix
Post by Qinpeng Wang
        >      > --enable-graphite --enable-fully-dynamic-string
        >      > --enable-libstdcxx-time=yes
--disable-libstdcxx-pch
Post by Qinpeng Wang
        >      > --disable-libstdcxx-debug
        --enable-version-specific-runtime-libs
        >      > --disable-isl-version-check --enable-lto
--enable-libgomp
Post by Qinpeng Wang
        >      > --disable-multilib --enable-checking=release
        --disable-rpath
        >      > --disable-win32-registry --disable-nls
--disable-werror
Post by Qinpeng Wang
        >      > --disable-symvers --with-libiconv
--with-system-zlib
Post by Qinpeng Wang
        >     --with-gmp=/mingw64
        >      > --with-mpfr=/mingw64 --with-mpc=/mingw64
        --with-isl=/mingw64
        >      > --with-pkgversion='Rev5, Built by MSYS2 project'
        >      >
--with-bugurl=https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>
Post by Qinpeng Wang
        <https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>>
Post by Qinpeng Wang
        >     <https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>
Post by Qinpeng Wang
        <https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>>> --with-gnu-as
Post by Qinpeng Wang
        >      > --with-gnu-ld Thread model: posix gcc version
5.3.0
Post by Qinpeng Wang
        (Rev5, Built by
        >      > MSYS2 project)|
        >      >
        >      >
        >      > Lastly, I have restarted the computer, the error
        message stayed
        >     the same.
        >      >
        >      > Thank you so much!
        >      >
        >      > On Friday, March 30, 2018 at 9:38:11 PM UTC-5,
Pascal
Post by Qinpeng Wang
        >      >
        >      >     Your "ldflags" option in the "[blas]"
section is
Post by Qinpeng Wang
        not correct,
        >     it should
        >      >     at least start with -I or -l, and there
should be
Post by Qinpeng Wang
        quotation
        >     marks
        >      >     around
        >      >     the path since there is a space in "Program
Files".
Post by Qinpeng Wang
        >      >
        >      >     What happens if you do not include this
line in
Post by Qinpeng Wang
        the .theanorc?
        >      >
        >      >     Are you able to run `gcc -v` from inside the
        conda env, and
        >     what is the
        >      >     output?
        >      >
        >      >     Also, it may seem stupid, but have you
restarted your
Post by Qinpeng Wang
        >     computer, or at
        >      >     least the terminal, since you installed
        m2w64-toolchain?
        >      >
        >      >      > Hi, all,
        >      >      >
        >      >      >
        >      >      > I'm trying to setup theano 1.0.1 with
CUDA 7.5
Post by Qinpeng Wang
        and cuDNN 6.0
        >      >     (Windows7).
        >      >      > I created a new virtual environment with
conda
Post by Qinpeng Wang
        and installed
        >      >     required
        >      >      >
        >      >      > |conda install numpy scipy mkl-service
libpython
Post by Qinpeng Wang
        >     m2w64-toolchain
        >      >     conda
        >      >      > install theano pygpu |
        >      >      >
        >      >      > Here are versions of major relavant
        >      >      >
        >      >      > |theano 1.0.1 python 3.5.5 numpy 1.14.2
scipy
Post by Qinpeng Wang
        1.0.0
        >      >     m2w64-toolchain 5.3.0
        >      >      > mkl-service 1.1.2 pygpu 0.7.5 libgpuarray
        0.7.5 libpython
        >     2.1 mkl
        >      >     2018.0.2 |
        >      >      >
        >      >      > The way I installed cuDNN is to copy files
        into corresponding
        >      >     folders
        >      >      > inside CUDA folder. Originally I installed
        CUDA at C:\Program
        >      >      > Files\NVIDIA GPU Computing Toolkit, then I
        copied the CUDA
        >     folder
        >      >     to C root.
        >      >      >
        >      >      > |python -c "import theano"|
        >      >      >
        >      >      >
        >      >      > |ERROR (theano.gpuarray): Could not
initialize
Post by Qinpeng Wang
        pygpu, support
        >      >     disabled
        >      >      > Traceback (most recent call last): File
        "C:\Program
        >      >      >
        >      >
        >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
        >      >      > line 227, in <module> use(config.device)
File
Post by Qinpeng Wang
        "C:\Program
        >      >      >
        >      >
        >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
        >      >      > line 214, in use init_dev(device,
        preallocate=preallocate)
        >     File
        >      >      > "C:\Program
        >      >      >
        >      >
        >
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
        >      >      > line 121, in init_dev
dnn.dnn_available.msg)
Post by Qinpeng Wang
        >     You
        >      >     enabled
        >      >      > cuDNN, but we aren't able to use it: cannot
        compile with
        >     cuDNN.
        >      >     We got
        >      >      > this error: b'C:/Program
        >      >      >
        >      >
        >
        >
        >      >
        >      >      > cannot find C:/Program: No such file or
        directory C:/Program
        >      >      >
        >      >
        >
        >
        >      >
        >      >      > cannot find
        >      >      >
        >      >
        >
        >
        >      >
        ld returned
        >     1 exit
        >      >     status' |
        >      >      >
        >      >      >
        >      >      > |[global] floatX = float32 device = cuda
[dnn]
Post by Qinpeng Wang
        enabled = True
        >      >      > include_path=C:/CUDA/v7.5/include
        >     library_path=C:/CUDA/v7.5/lib/x64
        >      >      > [blas] ldflags = C:/Program
        >     Files/Anaconda3/envs/t/Library/bin
        >      >     mkl_rt |
        >      >      >
        >      >      > Can somebody help please?
        >      >      >
        >      >      >
        >      >      > Thanks!
        >      >      >
        >      >      > --
        >      >      >
        >      >      > ---
        >      >      > You received this message because you are
        subscribed to
        >     the Google
        >      >      > Groups "theano-users" group.
        >      >      > To unsubscribe from this group and stop
        receiving emails
        >     from it,
        >      >     send
        >      >      > an email
        >      >      >
        >     <javascript:> <javascript:>>.
        >      >      > For more options,
        visithttps://groups.google.com/d/optout
<http://groups.google.com/d/optout>
Post by Qinpeng Wang
        <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>
Post by Qinpeng Wang
        >     <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
        <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>>
Post by Qinpeng Wang
        >      >     <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
        <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>
Post by Qinpeng Wang
        >     <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
        <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>>>.
Post by Qinpeng Wang
        >      >
        >      >     --
        >      >     Pascal Lamblin
        >      >
        >      > --
        >      >
        >      > ---
        >      > You received this message because you are
subscribed
Post by Qinpeng Wang
        to the Google
        >      > Groups "theano-users" group.
        >      > To unsubscribe from this group and stop receiving
        emails from it,
        >     send
        >      > an email
        >      >
<javascript:><javascript:>>.
Post by Qinpeng Wang
        >      > For more options,
        visithttps://groups.google.com/d/optout
<http://groups.google.com/d/optout>
Post by Qinpeng Wang
        <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>
Post by Qinpeng Wang
        >     <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
        <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>>.
Post by Qinpeng Wang
        >
        >     --
        >     Pascal Lamblin
        >
        > --
        >
        > ---
        > You received this message because you are subscribed
to the
Post by Qinpeng Wang
        Google
        > Groups "theano-users" group.
        > To unsubscribe from this group and stop receiving emails
        from it, send
        <http://googlegroups.com/>
<javascript:>>.
Post by Qinpeng Wang
        > For more options,
visithttps://groups.google.com/d/optout
<http://groups.google.com/d/optout>
Post by Qinpeng Wang
        <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
        --
        Pascal Lamblin
    --
    ---
    You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
    Groups "theano-users" group.
    To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
    For more options, visithttps://groups.google.com/d/optout
<http://groups.google.com/d/optout>
Post by Qinpeng Wang
    <https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Qinpeng Wang
2018-04-11 21:04:06 UTC
Permalink
Hi, Pascal,

Yes, in my error message there is still such a line about "C:/Program", I
attach it here again:

ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last):
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__i
nit__.py", line 227, in <module>
use(config.device)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__i
nit__.py", line 214, in use
init_dev(device, preallocate=preallocate)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__i
nit__.py", line 121, in init_dev
dnn.dnn_available.msg)
RuntimeError: You enabled cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got this error:
b'In file included from C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v7.5\\include/driver_types.h:53:0,
from C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v7.5\\include/cudnn.h:63,
from C:\\Users\\Z003RH0H\\AppData\\Local\\Temp\\try_flags_ss3nfk9h.c:4:
C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v7.5\\include/host_defines.h:8
4:0: warning: "__cdecl" redefined
#define __cdecl
^\
<built-in>: note:
this is the location of the previous definition
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program: No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../
../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status\r\n'


Thanks!
Post by Pascal Lamblin
And you _still_ have the following in your error message?
Post by Qinpeng Wang
cannot find C:/Program: No such file or directory
That is surprising, I don't see where it would get "C:/Program" from.
Post by Qinpeng Wang
Hi Pascal,
I have got rid of that line a while ago following your advice, here is
[global]
floatX = float32
device = cuda
[dnn]
enabled = True
Thanks!
You _also_ need to remove the incorrect "blas.ldflags" line from .theanorc.
Post by Qinpeng Wang
Hi, Arnaud,
Thanks for your reply! I reinstalled CUDA, copied cudnn files in
the
Post by Qinpeng Wang
Post by Qinpeng Wang
cuda directory, and removed the two parameters. However, I got
the same
Post by Qinpeng Wang
|ERROR (theano.gpuarray): Could not initialize pygpu, support
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with cuDNN.
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status'|
Post by Qinpeng Wang
Any other suggestion you have?
Thanks!
On Tuesday, April 10, 2018 at 12:45:59 PM UTC-5, Arnaud Bergeron
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the
old
Post by Qinpeng Wang
Post by Qinpeng Wang
GPU backend at C:\Program Files\NVIDIA GPU Computing
Toolkit, then
Post by Qinpeng Wang
Post by Qinpeng Wang
I copied the CUDA folder to C root.
This most likely broke DLL registrations. Please don't do
that.
Post by Qinpeng Wang
Post by Qinpeng Wang
If you reinstall cuda in its original location, put the cudnn
files
Post by Qinpeng Wang
in the cuda directory and remove these two parameters in your
|include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64|
Post by Qinpeng Wang
It should work again.
---
I you really want to try with the current setup, you can try
to set this
Post by Qinpeng Wang
|base_path=C:/CUDA/v7.5 |
Instead of the two above, but I'm pretty sure that will still
break
Post by Qinpeng Wang
on the cuda DLLs.
Post by Qinpeng Wang
Then I installed cuDNN by copying unzipped files into
corresponding folders inside CUDA folder. I'm using CUDA 7.5
and
Post by Qinpeng Wang
Post by Qinpeng Wang
cuDNN 6.0 (Windows7). See my theanoc file in my original
post to
Post by Qinpeng Wang
Post by Qinpeng Wang
make sure I have pointed theano to the right location.
Yes this error is not related to Idflags.
Thanks!
On Tuesday, April 3, 2018 at 9:59:38 AM UTC-5, Pascal
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I
have
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
versions of
Post by Qinpeng Wang
installed libraries in my original post), but somehow
theano
Post by Qinpeng Wang
Post by Qinpeng Wang
is unable
Post by Qinpeng Wang
to locate the BLAS library automatically, hence the
warning
Post by Qinpeng Wang
Post by Qinpeng Wang
message.
Post by Qinpeng Wang
Since the "gcc-v" output seems reasonable, could you
please
Post by Qinpeng Wang
Post by Qinpeng Wang
help me
Post by Qinpeng Wang
figure out how to get rid of the error message in my
original post? I
Post by Qinpeng Wang
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu,
support disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device,
preallocate=preallocate) File
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 121, in init_dev dnn.dnn_available.msg)
You enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile
with
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
cuDNN. We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory
C:/Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
cannot find
No such file or directory collect2.exe: error: ld
returned 1
Post by Qinpeng Wang
Post by Qinpeng Wang
exit status' |
Post by Qinpeng Wang
Thanks!
On Tuesday, April 3, 2018 at 9:28:37 AM UTC-5, Pascal
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really
sure
Post by Qinpeng Wang
Post by Qinpeng Wang
how to set the
Post by Qinpeng Wang
Post by Qinpeng Wang
ldflags parameter, but if I don't have this
line, I
Post by Qinpeng Wang
Post by Qinpeng Wang
get this warning
Post by Qinpeng Wang
Post by Qinpeng Wang
WARNING (theano.tensor.blas): Using NumPy C-API
based
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a
slight slow down in
Post by Qinpeng Wang
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>>>,
logger.warning('Using
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
NumPy C-API based
Post by Qinpeng Wang
Post by Qinpeng Wang
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the
theaorc file to
Post by Qinpeng Wang
get rid
Post by Qinpeng Wang
of this warning message. *How to set
ldflags correctly?*
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
These flags are for specifying where an optimized
BLAS
Post by Qinpeng Wang
Post by Qinpeng Wang
library is, when
Post by Qinpeng Wang
it is not found automatically.
If you do not have an optimized BLAS library
installed,
Post by Qinpeng Wang
Post by Qinpeng Wang
then it should
Post by Qinpeng Wang
stay empty.
Did you install the "mkl-service" package in
conda?
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
This is usually the easiest way to install an
optimized
Post by Qinpeng Wang
Post by Qinpeng Wang
BLAS that will
Post by Qinpeng Wang
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of
"gcc
Post by Qinpeng Wang
Post by Qinpeng Wang
-v" would
Post by Qinpeng Wang
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
../gcc-5.3.0/configure
Post by Qinpeng Wang
Post by Qinpeng Wang
--prefix=/mingw64
--with-local-prefix=/mingw64/local
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--build=x86_64-w64-mingw32
--host=x86_64-w64-mingw32
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--enable-bootstrap
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic
--enable-threads=posix
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes
--disable-libstdcxx-pch
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-libstdcxx-debug
--enable-version-specific-runtime-libs
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-isl-version-check --enable-lto
--enable-libgomp
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-multilib --enable-checking=release
--disable-rpath
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-win32-registry --disable-nls
--disable-werror
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-symvers --with-libiconv
--with-system-zlib
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64
--with-isl=/mingw64
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-pkgversion='Rev5, Built by MSYS2
project'
Post by Qinpeng Wang
--with-bugurl=https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>>
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2>>> --with-gnu-as
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version
5.3.0
Post by Qinpeng Wang
Post by Qinpeng Wang
(Rev5, Built by
Post by Qinpeng Wang
Post by Qinpeng Wang
MSYS2 project)|
Lastly, I have restarted the computer, the
error
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
message stayed
Post by Qinpeng Wang
the same.
Post by Qinpeng Wang
Thank you so much!
On Friday, March 30, 2018 at 9:38:11 PM UTC-5,
Pascal
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Your "ldflags" option in the "[blas]"
section is
Post by Qinpeng Wang
Post by Qinpeng Wang
not correct,
Post by Qinpeng Wang
it should
Post by Qinpeng Wang
at least start with -I or -l, and there
should be
Post by Qinpeng Wang
Post by Qinpeng Wang
quotation
Post by Qinpeng Wang
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program
Files".
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
What happens if you do not include this
line in
Post by Qinpeng Wang
Post by Qinpeng Wang
the .theanorc?
Post by Qinpeng Wang
Post by Qinpeng Wang
Are you able to run `gcc -v` from inside
the
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
conda env, and
Post by Qinpeng Wang
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you
restarted your
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed
m2w64-toolchain?
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with
CUDA 7.5
Post by Qinpeng Wang
Post by Qinpeng Wang
and cuDNN 6.0
Post by Qinpeng Wang
Post by Qinpeng Wang
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with
conda
Post by Qinpeng Wang
Post by Qinpeng Wang
and installed
Post by Qinpeng Wang
Post by Qinpeng Wang
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service
libpython
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
Here are versions of major relavant
|theano 1.0.1 python 3.5.5 numpy 1.14.2
scipy
Post by Qinpeng Wang
Post by Qinpeng Wang
1.0.0
Post by Qinpeng Wang
Post by Qinpeng Wang
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5
libgpuarray
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
0.7.5 libpython
Post by Qinpeng Wang
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy
files
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
into corresponding
Post by Qinpeng Wang
Post by Qinpeng Wang
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I
installed
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
CUDA at C:\Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Files\NVIDIA GPU Computing Toolkit, then
I
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
copied the CUDA
Post by Qinpeng Wang
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not
initialize
Post by Qinpeng Wang
Post by Qinpeng Wang
pygpu, support
Post by Qinpeng Wang
Post by Qinpeng Wang
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device)
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device,
preallocate=preallocate)
Post by Qinpeng Wang
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 121, in init_dev
dnn.dnn_available.msg)
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
You
Post by Qinpeng Wang
enabled
cannot
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
compile with
Post by Qinpeng Wang
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or
directory C:/Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
cannot find
ld returned
Post by Qinpeng Wang
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda
[dnn]
Post by Qinpeng Wang
Post by Qinpeng Wang
enabled = True
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you
are
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
subscribed to
Post by Qinpeng Wang
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop
receiving emails
Post by Qinpeng Wang
from it,
Post by Qinpeng Wang
send
Post by Qinpeng Wang
an email
<javascript:>
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options,
visithttps://groups.google.com/d/optout
<http://groups.google.com/d/optout>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>>
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are
subscribed
Post by Qinpeng Wang
Post by Qinpeng Wang
to the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop
receiving
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
emails from it,
Post by Qinpeng Wang
send
Post by Qinpeng Wang
an email
<javascript:>
Post by Qinpeng Wang
<javascript:><javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options,
visithttps://groups.google.com/d/optout
<http://groups.google.com/d/optout>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed
to the
Post by Qinpeng Wang
Post by Qinpeng Wang
Google
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving
emails
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
from it, send
<http://googlegroups.com/>
<javascript:>
Post by Qinpeng Wang
<javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options,
visithttps://groups.google.com/d/optout
<http://groups.google.com/d/optout>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it,
<javascript:>.
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visithttps://groups.google.com/d/optout
<http://groups.google.com/d/optout>
Post by Qinpeng Wang
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
--
---
You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
<javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Arnaud Bergeron
2018-04-11 17:18:47 UTC
Permalink
It seems like the gcc in anaconda has trouble when installed in a path with spaces:

Can you copy this to 'test.c':

int main() {
return 1;
}

and try to compile with:

g++ test.c -o test
Post by Qinpeng Wang
Hi, Arnaud,
ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 227, in <module>
use(config.device)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 214, in use
init_dev(device, preallocate=preallocate)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 121, in init_dev
dnn.dnn_available.msg)
b'C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program: No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status'
Any other suggestion you have?
Thanks!
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the old GPU backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA folder to C root.
This most likely broke DLL registrations. Please don't do that.
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
It should work again.
---
I you really want to try with the current setup, you can try to set this
base_path=C:/CUDA/v7.5
Instead of the two above, but I'm pretty sure that will still break on the cuda DLLs.
Post by Qinpeng Wang
Then I installed cuDNN by copying unzipped files into corresponding folders inside CUDA folder. I'm using CUDA 7.5 and cuDNN 6.0 (Windows7). See my theanoc file in my original post to make sure I have pointed theano to the right location.
Yes this error is not related to Idflags.
Thanks!
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable
to locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me
figure out how to get rid of the error message in my original post? I
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit status' |
Thanks!
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set the
ldflags parameter, but if I don't have this line, I get this warning
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a slight slow down in
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py <https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py <https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>>,
Post by Qinpeng Wang
|if not config.blas.ldflags: logger.warning('Using NumPy C-API based
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to
get rid
Post by Qinpeng Wang
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is, when
it is not found automatically.
If you do not have an optimized BLAS library installed, then it should
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that will
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Post by Qinpeng Wang
Target: x86_64-w64-mingw32 Configured with: ../gcc-5.3.0/configure
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0 --enable-bootstrap
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2 <https://sourceforge.net/projects/msys2>
<https://sourceforge.net/projects/msys2 <https://sourceforge.net/projects/msys2>> --with-gnu-as
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5, Built by
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed
the same.
Post by Qinpeng Wang
Thank you so much!
Your "ldflags" option in the "[blas]" section is not correct,
it should
Post by Qinpeng Wang
at least start with -I or -l, and there should be quotation
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program Files".
What happens if you do not include this line in the .theanorc?
Are you able to run `gcc -v` from inside the conda env, and
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted your
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and installed
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into corresponding
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at C:\Program
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate)
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
You
Post by Qinpeng Wang
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled = True
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
send
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>
Post by Qinpeng Wang
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>>.
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Qinpeng Wang
2018-04-11 21:07:23 UTC
Permalink
Hi, Arnaud,

Thanks for your reply! I believe currently I rely on the gcc from
m2w64-toolchain package, installed via conda.
Where can I find this file "test.c", and how can I compile with your code?

Thanks!
Post by Arnaud Bergeron
int main() {
return 1;
}
g++ test.c -o test
Hi, Arnaud,
Thanks for your reply! I reinstalled CUDA, copied cudnn files in the cuda
directory, and removed the two parameters. However, I got the same errors
ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 227, in <module>
use(config.device)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 214, in use
init_dev(device, preallocate=preallocate)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 121, in init_dev
dnn.dnn_available.msg)
b'C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program: No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status'
Any other suggestion you have?
Thanks!
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the old GPU
backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then I copied
the CUDA folder to C root.
This most likely broke DLL registrations. Please don't do that.
If you reinstall cuda in its original location, put the cudnn files in
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
It should work again.
---
I you really want to try with the current setup, you can try to set this
base_path=C:/CUDA/v7.5
Instead of the two above, but I'm pretty sure that will still break on the cuda DLLs.
Then I installed cuDNN by copying unzipped files into corresponding
folders inside CUDA folder. I'm using CUDA 7.5 and cuDNN 6.0 (Windows7).
See my theanoc file in my original post to make sure I have pointed theano
to the right location.
Yes this error is not related to Idflags.
Thanks!
Post by Pascal Lamblin
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable
to locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me
figure out how to get rid of the error message in my original post? I
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status' |
Post by Qinpeng Wang
Thanks!
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set
the
Post by Qinpeng Wang
Post by Qinpeng Wang
ldflags parameter, but if I don't have this line, I get this
warning
Post by Qinpeng Wang
Post by Qinpeng Wang
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a slight slow
down in
Post by Qinpeng Wang
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
Post by Qinpeng Wang
,
|if not config.blas.ldflags: logger.warning('Using NumPy C-API
based
Post by Qinpeng Wang
Post by Qinpeng Wang
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to
get rid
Post by Qinpeng Wang
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is,
when
Post by Qinpeng Wang
it is not found automatically.
If you do not have an optimized BLAS library installed, then it
should
Post by Qinpeng Wang
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that
will
Post by Qinpeng Wang
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
../gcc-5.3.0/configure
Post by Qinpeng Wang
Post by Qinpeng Wang
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
Post by Qinpeng Wang
Post by Qinpeng Wang
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0
--enable-bootstrap
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2> --with-gnu-as
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5,
Built by
Post by Qinpeng Wang
Post by Qinpeng Wang
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed
the same.
Post by Qinpeng Wang
Thank you so much!
On Friday, March 30, 2018 at 9:38:11 PM UTC-5, Pascal Lamblin
Your "ldflags" option in the "[blas]" section is not
correct,
Post by Qinpeng Wang
it should
Post by Qinpeng Wang
at least start with -I or -l, and there should be quotation
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program Files".
What happens if you do not include this line in the
.theanorc?
Post by Qinpeng Wang
Post by Qinpeng Wang
Are you able to run `gcc -v` from inside the conda env, and
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted your
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN
6.0
Post by Qinpeng Wang
Post by Qinpeng Wang
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and
installed
Post by Qinpeng Wang
Post by Qinpeng Wang
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into
corresponding
Post by Qinpeng Wang
Post by Qinpeng Wang
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at
C:\Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Files\NVIDIA GPU Computing Toolkit, then I copied the
CUDA
Post by Qinpeng Wang
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu,
support
Post by Qinpeng Wang
Post by Qinpeng Wang
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device,
preallocate=preallocate)
Post by Qinpeng Wang
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 121, in init_dev dnn.dnn_available.msg)
You
Post by Qinpeng Wang
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory
C:/Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
cannot find
No such file or directory collect2.exe: error: ld
returned
Post by Qinpeng Wang
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled =
True
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
send
<javascript:>
Post by Qinpeng Wang
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit
https://groups.google.com/d/optout
Post by Qinpeng Wang
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from
it,
Post by Qinpeng Wang
send
<javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.
--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Arnaud Bergeron
2018-04-16 17:45:53 UTC
Permalink
Create a new file and call it whatever (like test.c), then try to compile it. This is just to check if the compiler works outside of Theano.

I’ve added sample content so that the file has valid C code.
Post by Qinpeng Wang
Hi, Arnaud,
Thanks for your reply! I believe currently I rely on the gcc from m2w64-toolchain package, installed via conda.
Where can I find this file "test.c", and how can I compile with your code?
Thanks!
int main() {
return 1;
}
g++ test.c -o test
Post by Qinpeng Wang
Hi, Arnaud,
ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 227, in <module>
use(config.device)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 214, in use
init_dev(device, preallocate=preallocate)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 121, in init_dev
dnn.dnn_available.msg)
b'C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program: No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status'
Any other suggestion you have?
Thanks!
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the old GPU backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA folder to C root.
This most likely broke DLL registrations. Please don't do that.
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
It should work again.
---
I you really want to try with the current setup, you can try to set this
base_path=C:/CUDA/v7.5
Instead of the two above, but I'm pretty sure that will still break on the cuda DLLs.
Post by Qinpeng Wang
Then I installed cuDNN by copying unzipped files into corresponding folders inside CUDA folder. I'm using CUDA 7.5 and cuDNN 6.0 (Windows7). See my theanoc file in my original post to make sure I have pointed theano to the right location.
Yes this error is not related to Idflags.
Thanks!
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable
to locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me
figure out how to get rid of the error message in my original post? I
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit status' |
Thanks!
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set the
ldflags parameter, but if I don't have this line, I get this warning
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a slight slow down in
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py <https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py <https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>>,
Post by Qinpeng Wang
|if not config.blas.ldflags: logger.warning('Using NumPy C-API based
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to
get rid
Post by Qinpeng Wang
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is, when
it is not found automatically.
If you do not have an optimized BLAS library installed, then it should
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that will
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Post by Qinpeng Wang
Target: x86_64-w64-mingw32 Configured with: ../gcc-5.3.0/configure
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0 --enable-bootstrap
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2 <https://sourceforge.net/projects/msys2>
<https://sourceforge.net/projects/msys2 <https://sourceforge.net/projects/msys2>> --with-gnu-as
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5, Built by
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed
the same.
Post by Qinpeng Wang
Thank you so much!
Your "ldflags" option in the "[blas]" section is not correct,
it should
Post by Qinpeng Wang
at least start with -I or -l, and there should be quotation
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program Files".
What happens if you do not include this line in the .theanorc?
Are you able to run `gcc -v` from inside the conda env, and
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted your
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and installed
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into corresponding
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at C:\Program
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate)
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
You
Post by Qinpeng Wang
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled = True
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
send
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>
Post by Qinpeng Wang
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>>.
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Qinpeng Wang
2018-04-16 19:26:37 UTC
Permalink
Hi, Arnaud,

I think you have located the issue, here is what I did: I activated the
virtual env in cmd, then ran "g++ test.c -o test", here is the error
message:

C:/Program
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe:
cannot find C:/Program:No such file or directory
C:/Program
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe:
cannot find
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o:
No such file or directory
collect2.exe: error: ld returned 1 exit status

What do you suggest that I do now? Do I have to remove Anaconda and install
at C root?

Thanks!
Post by Arnaud Bergeron
Create a new file and call it whatever (like test.c), then try to compile
it. This is just to check if the compiler works outside of Theano.
I’ve added sample content so that the file has valid C code.
Hi, Arnaud,
Thanks for your reply! I believe currently I rely on the gcc from
m2w64-toolchain package, installed via conda.
Where can I find this file "test.c", and how can I compile with your code?
Thanks!
Post by Arnaud Bergeron
int main() {
return 1;
}
g++ test.c -o test
Hi, Arnaud,
Thanks for your reply! I reinstalled CUDA, copied cudnn files in the cuda
directory, and removed the two parameters. However, I got the same errors
ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 227, in <module>
use(config.device)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 214, in use
init_dev(device, preallocate=preallocate)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 121, in init_dev
dnn.dnn_available.msg)
b'C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program: No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status'
Any other suggestion you have?
Thanks!
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the old GPU
backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then I copied
the CUDA folder to C root.
This most likely broke DLL registrations. Please don't do that.
If you reinstall cuda in its original location, put the cudnn files in
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
It should work again.
---
I you really want to try with the current setup, you can try to set this
base_path=C:/CUDA/v7.5
Instead of the two above, but I'm pretty sure that will still break on the cuda DLLs.
Then I installed cuDNN by copying unzipped files into corresponding
folders inside CUDA folder. I'm using CUDA 7.5 and cuDNN 6.0
(Windows7). See my theanoc file in my original post to make sure I have
pointed theano to the right location.
Yes this error is not related to Idflags.
Thanks!
Post by Pascal Lamblin
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable
to locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me
figure out how to get rid of the error message in my original post? I
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit
status' |
Post by Qinpeng Wang
Thanks!
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set
the
Post by Qinpeng Wang
Post by Qinpeng Wang
ldflags parameter, but if I don't have this line, I get this
warning
Post by Qinpeng Wang
Post by Qinpeng Wang
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a slight slow
down in
Post by Qinpeng Wang
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
<
https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py
Post by Qinpeng Wang
Post by Qinpeng Wang
,
|if not config.blas.ldflags: logger.warning('Using NumPy C-API
based
Post by Qinpeng Wang
Post by Qinpeng Wang
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to
get rid
Post by Qinpeng Wang
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library
is, when
Post by Qinpeng Wang
it is not found automatically.
If you do not have an optimized BLAS library installed, then it
should
Post by Qinpeng Wang
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that
will
Post by Qinpeng Wang
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
../gcc-5.3.0/configure
Post by Qinpeng Wang
Post by Qinpeng Wang
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
Post by Qinpeng Wang
Post by Qinpeng Wang
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0
--enable-bootstrap
Post by Qinpeng Wang
Post by Qinpeng Wang
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug
--enable-version-specific-runtime-libs
Post by Qinpeng Wang
Post by Qinpeng Wang
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2
<https://sourceforge.net/projects/msys2> --with-gnu-as
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5,
Built by
Post by Qinpeng Wang
Post by Qinpeng Wang
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed
the same.
Post by Qinpeng Wang
Thank you so much!
On Friday, March 30, 2018 at 9:38:11 PM UTC-5, Pascal Lamblin
Your "ldflags" option in the "[blas]" section is not
correct,
Post by Qinpeng Wang
it should
Post by Qinpeng Wang
at least start with -I or -l, and there should be quotation
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program Files".
What happens if you do not include this line in the
.theanorc?
Post by Qinpeng Wang
Post by Qinpeng Wang
Are you able to run `gcc -v` from inside the conda env, and
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted your
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and
cuDNN 6.0
Post by Qinpeng Wang
Post by Qinpeng Wang
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and
installed
Post by Qinpeng Wang
Post by Qinpeng Wang
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5
libpython
Post by Qinpeng Wang
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into
corresponding
Post by Qinpeng Wang
Post by Qinpeng Wang
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at
C:\Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
Files\NVIDIA GPU Computing Toolkit, then I copied the
CUDA
Post by Qinpeng Wang
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu,
support
Post by Qinpeng Wang
Post by Qinpeng Wang
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device,
preallocate=preallocate)
Post by Qinpeng Wang
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
line 121, in init_dev dnn.dnn_available.msg)
You
Post by Qinpeng Wang
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory
C:/Program
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
cannot find
No such file or directory collect2.exe: error: ld
returned
Post by Qinpeng Wang
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled
= True
Post by Qinpeng Wang
Post by Qinpeng Wang
Post by Qinpeng Wang
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
send
<javascript:>
Post by Qinpeng Wang
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit
https://groups.google.com/d/optout
Post by Qinpeng Wang
<https://groups.google.com/d/optout>
Post by Qinpeng Wang
<https://groups.google.com/d/optout
<https://groups.google.com/d/optout>>.
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the
Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from
it,
Post by Qinpeng Wang
send
<javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout
<https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout.
--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.
--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an
For more options, visit https://groups.google.com/d/optout.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Arnaud Bergeron
2018-04-17 01:39:14 UTC
Permalink
For a short term fix, you can reinstall anaconda in a path that doesn’t have spaces. It doesn’t have to be at the root though.

For a longer-term fix you can try to report the issue about compilation to continuum and see if they’ll fix it.
Post by Qinpeng Wang
Hi, Arnaud,
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program:No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status
What do you suggest that I do now? Do I have to remove Anaconda and install at C root?
Thanks!
Create a new file and call it whatever (like test.c), then try to compile it. This is just to check if the compiler works outside of Theano.
I’ve added sample content so that the file has valid C code.
Post by Qinpeng Wang
Hi, Arnaud,
Thanks for your reply! I believe currently I rely on the gcc from m2w64-toolchain package, installed via conda.
Where can I find this file "test.c", and how can I compile with your code?
Thanks!
int main() {
return 1;
}
g++ test.c -o test
Post by Qinpeng Wang
Hi, Arnaud,
ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 227, in <module>
use(config.device)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 214, in use
init_dev(device, preallocate=preallocate)
File "C:\Program Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py", line 121, in init_dev
dnn.dnn_available.msg)
b'C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find C:/Program: No such file or directory
C:/Program Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/bin/ld.exe: cannot find Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/../../../../x86_64-w64-mingw32/lib/../lib/default-manifest.o: No such file or directory
collect2.exe: error: ld returned 1 exit status'
Any other suggestion you have?
Thanks!
Post by Qinpeng Wang
Hi Pascal,
I have CUDA installed a while ago when I was still using the old GPU backend at C:\Program Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA folder to C root.
This most likely broke DLL registrations. Please don't do that.
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
It should work again.
---
I you really want to try with the current setup, you can try to set this
base_path=C:/CUDA/v7.5
Instead of the two above, but I'm pretty sure that will still break on the cuda DLLs.
Post by Qinpeng Wang
Then I installed cuDNN by copying unzipped files into corresponding folders inside CUDA folder. I'm using CUDA 7.5 and cuDNN 6.0 (Windows7). See my theanoc file in my original post to make sure I have pointed theano to the right location.
Yes this error is not related to Idflags.
Thanks!
This is a completely different error message.
This one has to do with cuDNN, not BLAS or MKL.
Where have you installed cuDNN?
Does _that_ error still happen when you do not ldflags?
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!
I have "mkl-service" package installed via conda (I have versions of
installed libraries in my original post), but somehow theano is unable
to locate the BLAS library automatically, hence the warning message.
Since the "gcc-v" output seems reasonable, could you please help me
figure out how to get rid of the error message in my original post? I
attach it here again for your convenience.
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support disabled
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 214, in use init_dev(device, preallocate=preallocate) File
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
line 121, in init_dev dnn.dnn_available.msg) RuntimeError: You enabled
cuDNN, but we aren't able to use it: cannot compile with cuDNN. We got
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned 1 exit status' |
Thanks!
Post by Qinpeng Wang
Hi Pascal,
Thanks for your reply!! Indeed I wasn't really sure how to set the
ldflags parameter, but if I don't have this line, I get this warning
WARNING (theano.tensor.blas): Using NumPy C-API based
implementation for
Post by Qinpeng Wang
BLAS functions.
This is just a warning, the worse issue would be a slight slow down in
some CPU operations.
Post by Qinpeng Wang
After inspecting theano/tensor/blas_headers.py
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py <https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>
<https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py <https://github.com/Theano/Theano/blob/82a3ec977b7c8630d92871e70d62e5a2c52cea62/theano/tensor/blas_headers.py>>>,
Post by Qinpeng Wang
|if not config.blas.ldflags: logger.warning('Using NumPy C-API based
implementation for BLAS functions.')|
So I realize I may have to have ldflags in the theaorc file to
get rid
Post by Qinpeng Wang
of this warning message. *How to set ldflags correctly?*
These flags are for specifying where an optimized BLAS library is, when
it is not found automatically.
If you do not have an optimized BLAS library installed, then it should
stay empty.
Did you install the "mkl-service" package in conda?
This is usually the easiest way to install an optimized BLAS that will
be automatically found.
Post by Qinpeng Wang
I was asking around what the correct output of "gcc -v" would
look like,
This seems reasonable.
Post by Qinpeng Wang
|Using built-in specs. COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=C:/Program\
Files/Anaconda3/envs/te/Library/mingw-w64/bin/../lib/gcc/x86_64-w64-mingw32/5.3.0/lto-wrapper.exe
Post by Qinpeng Wang
Target: x86_64-w64-mingw32 Configured with: ../gcc-5.3.0/configure
--prefix=/mingw64 --with-local-prefix=/mingw64/local
--build=x86_64-w64-mingw32 --host=x86_64-w64-mingw32
--target=x86_64-w64-mingw32
--with-native-system-header-dir=/mingw64/x86_64-w64-mingw32/include
--libexecdir=/mingw64/lib
--with-gxx-include-dir=/mingw64/include/c++/5.3.0 --enable-bootstrap
--with-arch=x86-64 --with-tune=generic
--enable-languages=c,lto,c++,objc,obj-c++,fortran,ada
--enable-shared
Post by Qinpeng Wang
--enable-static --enable-libatomic --enable-threads=posix
--enable-graphite --enable-fully-dynamic-string
--enable-libstdcxx-time=yes --disable-libstdcxx-pch
--disable-libstdcxx-debug --enable-version-specific-runtime-libs
--disable-isl-version-check --enable-lto --enable-libgomp
--disable-multilib --enable-checking=release --disable-rpath
--disable-win32-registry --disable-nls --disable-werror
--disable-symvers --with-libiconv --with-system-zlib
--with-gmp=/mingw64
Post by Qinpeng Wang
--with-mpfr=/mingw64 --with-mpc=/mingw64 --with-isl=/mingw64
--with-pkgversion='Rev5, Built by MSYS2 project'
--with-bugurl=https://sourceforge.net/projects/msys2 <https://sourceforge.net/projects/msys2>
<https://sourceforge.net/projects/msys2 <https://sourceforge.net/projects/msys2>> --with-gnu-as
Post by Qinpeng Wang
--with-gnu-ld Thread model: posix gcc version 5.3.0 (Rev5, Built by
MSYS2 project)|
Lastly, I have restarted the computer, the error message stayed
the same.
Post by Qinpeng Wang
Thank you so much!
Your "ldflags" option in the "[blas]" section is not correct,
it should
Post by Qinpeng Wang
at least start with -I or -l, and there should be quotation
marks
Post by Qinpeng Wang
around
the path since there is a space in "Program Files".
What happens if you do not include this line in the .theanorc?
Are you able to run `gcc -v` from inside the conda env, and
what is the
Post by Qinpeng Wang
output?
Also, it may seem stupid, but have you restarted your
computer, or at
Post by Qinpeng Wang
least the terminal, since you installed m2w64-toolchain?
Post by Qinpeng Wang
Hi, all,
I'm trying to setup theano 1.0.1 with CUDA 7.5 and cuDNN 6.0
(Windows7).
Post by Qinpeng Wang
I created a new virtual environment with conda and installed
required
Post by Qinpeng Wang
|conda install numpy scipy mkl-service libpython
m2w64-toolchain
Post by Qinpeng Wang
conda
Post by Qinpeng Wang
install theano pygpu |
|theano 1.0.1 python 3.5.5 numpy 1.14.2 scipy 1.0.0
m2w64-toolchain 5.3.0
Post by Qinpeng Wang
mkl-service 1.1.2 pygpu 0.7.5 libgpuarray 0.7.5 libpython
2.1 mkl
Post by Qinpeng Wang
2018.0.2 |
Post by Qinpeng Wang
The way I installed cuDNN is to copy files into corresponding
folders
Post by Qinpeng Wang
inside CUDA folder. Originally I installed CUDA at C:\Program
Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA
folder
Post by Qinpeng Wang
to C root.
Post by Qinpeng Wang
|python -c "import theano"|
|ERROR (theano.gpuarray): Could not initialize pygpu, support
disabled
Post by Qinpeng Wang
Traceback (most recent call last): File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 227, in <module> use(config.device) File "C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
Post by Qinpeng Wang
Post by Qinpeng Wang
line 214, in use init_dev(device, preallocate=preallocate)
File
Post by Qinpeng Wang
Post by Qinpeng Wang
"C:\Program
Files\Anaconda3\envs\te\lib\site-packages\theano\gpuarray\__init__.py",
You
Post by Qinpeng Wang
enabled
Post by Qinpeng Wang
cuDNN, but we aren't able to use it: cannot compile with
cuDNN.
Post by Qinpeng Wang
We got
Post by Qinpeng Wang
this error: b'C:/Program
cannot find C:/Program: No such file or directory C:/Program
cannot find
No such file or directory collect2.exe: error: ld returned
1 exit
Post by Qinpeng Wang
status' |
Post by Qinpeng Wang
|[global] floatX = float32 device = cuda [dnn] enabled = True
include_path=C:/CUDA/v7.5/include
library_path=C:/CUDA/v7.5/lib/x64
Post by Qinpeng Wang
Post by Qinpeng Wang
[blas] ldflags = C:/Program
Files/Anaconda3/envs/t/Library/bin
Post by Qinpeng Wang
mkl_rt |
Post by Qinpeng Wang
Can somebody help please?
Thanks!
--
---
You received this message because you are subscribed to
the Google
Post by Qinpeng Wang
Post by Qinpeng Wang
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails
from it,
Post by Qinpeng Wang
send
<javascript:> <javascript:>>.
Post by Qinpeng Wang
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>
Post by Qinpeng Wang
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>>.
Post by Qinpeng Wang
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it,
send
Post by Qinpeng Wang
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>
<https://groups.google.com/d/optout <https://groups.google.com/d/optout>>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google
Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
Pascal Lamblin
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>.
--
---
You received this message because you are subscribed to the Google Groups "theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to theano-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Loading...