Windows 10 with anaconda, installation error

Hi ,

I am trying to install apex on my windows 10. I have installed currently anaconda with cuda 10 (cuda toolkit) and pytorch. That works well. But when I launch the command pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" . I get an error about NVCC not found. Is there a way to install apex on windows10 using anaconda ?

Config variable ‘Py_DEBUG’ is unset, Python ABI tag may be incorrect Config variable ‘WITH_PYMALLOC’ is unset, Python ABI tag may be incorrect Created temporary directory: C:\Users\Shiro\AppData\Local\Temp\pip-ephem-wheel-cache-lad1ick2 Created temporary directory: C:\Users\Shiro\AppData\Local\Temp\pip-req-tracker-ntiu8rps Created requirements tracker ‘C:\Users\Shiro\AppData\Local\Temp\pip-req-tracker-ntiu8rps’ Created temporary directory: C:\Users\Shiro\AppData\Local\Temp\pip-install-xc91fwpi Processing c:\users\shiro\apex Created temporary directory: C:\Users\Shiro\AppData\Local\Temp\pip-req-build-fgj_ek8b Added file:///C:/Users/Shiro/apex to build tracker ‘C:\Users\Shiro\AppData\Local\Temp\pip-req-tracker-ntiu8rps’ Running setup.py (path:C:\Users\Shiro\AppData\Local\Temp\pip-req-build-fgj_ek8b\setup.py) egg_info for package from file:///C:/Users/Shiro/apex Running command python setup.py egg_info torch.version = 1.1.0 running egg_info creating pip-egg-info\apex.egg-info writing pip-egg-info\apex.egg-info\PKG-INFO writing dependency_links to pip-egg-info\apex.egg-info\dependency_links.txt writing top-level names to pip-egg-info\apex.egg-info\top_level.txt writing manifest file ‘pip-egg-info\apex.egg-info\SOURCES.txt’ reading manifest file ‘pip-egg-info\apex.egg-info\SOURCES.txt’ writing manifest file ‘pip-egg-info\apex.egg-info\SOURCES.txt’ Source in c:\users\shiro\appdata\local\temp\pip-req-build-fgj_ek8b has version 0.1, which satisfies requirement apex==0.1 from file:///C:/Users/Shiro/apex Removed apex==0.1 from file:///C:/Users/Shiro/apex from build tracker ‘C:\Users\Shiro\AppData\Local\Temp\pip-req-tracker-ntiu8rps’ Skipping bdist_wheel for apex, due to binaries being disabled for it. Installing collected packages: apex Created temporary directory: C:\Users\Shiro\AppData\Local\Temp\pip-record-o0vdkzsy Running command ‘C:\Users\Shiro\Anaconda3\envs\DL-torch\python.exe’ -u -c ‘import setuptools, tokenize;file=’“'”‘C:\Users\Shiro\AppData\Local\Temp\pip-req-build-fgj_ek8b\setup.py’“'”‘;f=getattr(tokenize, ‘"’“‘open’”’“‘, open)(file);code=f.read().replace(’”‘"’\r\n’“'”‘, ‘"’"’\n’“'”‘);f.close();exec(compile(code, file, ‘"’“‘exec’”’“‘))’ --cpp_ext --cuda_ext install --record ‘C:\Users\Shiro\AppData\Local\Temp\pip-record-o0vdkzsy\install-record.txt’ --single-version-externally-managed --compile torch.version = 1.1.0 Informationÿ: impossible de trouver des fichiers pour le(s) modŠle(s) sp‚cifi‚(s). Traceback (most recent call last): File “<string>”, line 1, in <module> File “C:\Users\Shiro\AppData\Local\Temp\pip-req-build-fgj_ek8b\setup.py”, line 62, in <module> raise RuntimeError(”–cuda_ext was requested, but nvcc was not found. Are you sure your environment has nvcc available? If you’re installing within a container from https://hub.docker.com/r/pytorch/pytorch, only images whose names contain ‘devel’ will provide nvcc.“) RuntimeError: --cuda_ext was requested, but nvcc was not found. Are you sure your environment has nvcc available? If you’re installing within a container from https://hub.docker.com/r/pytorch/pytorch, only images whose names contain ‘devel’ will provide nvcc. Running setup.py install for apex … error Cleaning up… Removing source in C:\Users\Shiro\AppData\Local\Temp\pip-req-build-fgj_ek8b Removed build tracker ‘C:\Users\Shiro\AppData\Local\Temp\pip-req-tracker-ntiu8rps’ ERROR: Command “‘C:\Users\Shiro\Anaconda3\envs\DL-torch\python.exe’ -u -c ‘import setuptools, tokenize;file=’”'”‘C:\Users\Shiro\AppData\Local\Temp\pip-req-build-fgj_ek8b\setup.py’“'”‘;f=getattr(tokenize, ‘"’“‘open’”’“‘, open)(file);code=f.read().replace(’”‘"’\r\n’“'”‘, ‘"’"’\n’“'”‘);f.close();exec(compile(code, file, ‘"’“‘exec’”’“‘))’ --cpp_ext --cuda_ext install --record ‘C:\Users\Shiro\AppData\Local\Temp\pip-record-o0vdkzsy\install-record.txt’ --single-version-externally-managed --compile” failed with error code 1 in C:\Users\Shiro\AppData\Local\Temp\pip-req-build-fgj_ek8b
Exception information: Traceback (most recent call last): File “C:\Users\Shiro\Anaconda3\envs\DL-torch\lib\site-packages\pip_internal\cli\base_command.py”, line 178, in main status = self.run(options, args) File “C:\Users\Shiro\Anaconda3\envs\DL-torch\lib\site-packages\pip_internal\commands\install.py”, line 414, in run use_user_site=options.use_user_site, File “C:\Users\Shiro\Anaconda3\envs\DL-torch\lib\site-packages\pip_internal\req_init_.py”, line 58, in install_given_reqs **kwargs File “C:\Users\Shiro\Anaconda3\envs\DL-torch\lib\site-packages\pip_internal\req\req_install.py”, line 951, in install spinner=spinner, File “C:\Users\Shiro\Anaconda3\envs\DL-torch\lib\site-packages\pip_internal\utils\misc.py”, line 776, in call_subprocess % (command_desc, proc.returncode, cwd)) pip._internal.exceptions.InstallationError: Command “‘C:\Users\Shiro\Anaconda3\envs\DL-torch\python.exe’ -u -c ‘import setuptools, tokenize;file=’”‘“‘C:\Users\Shiro\AppData\Local\Temp\pip-req-build-fgj_ek8b\setup.py’”’“';f=getattr(tokenize, '”‘“‘open’”’“‘, open)(file);code=f.read().replace(’”‘"’\r\n’“'”‘, ‘"’"’\n’“'”‘);f.close();exec(compile(code, file, ‘"’“‘exec’”’“‘))’ --cpp_ext --cuda_ext install --record ‘C:\Users\Shiro\AppData\Local\Temp\pip-record-o0vdkzsy\install-record.txt’ --single-version-externally-managed --compile” failed with error code 1 in C:\Users\Shiro\AppData\Local\Temp\pip-req-build-fgj_ek8b
Config variable ‘Py_DEBUG’ is unset, Python ABI tag may be incorrect Config variable ‘WITH_PYMALLOC’ is unset, Python ABI tag may be incorrect

[Shiro-LK]

Hi thanks for the response. I finally found the issue, it seems that installing cuda with anaconda does not install nvcc. I install manually cuda and other thing without anaconda, then the installation of apex works.

Issue solved

[TotzkePaul]

I used conda forge to get apex installed and ‘working’. https://anaconda.org/conda-forge/nvidia-apex

"Original ImportError was: ModuleNotFoundError(“No module named ‘amp_C’)”

I’m still interested in getting it complied on Windows 10 for CUDA 10.2. Is this possible right now for windows on the latest version? see https://gist.github.com/ax3l/9489132

Is this the right way to install the toolkit in conda? conda install -c anaconda cudatoolkit