cupy: Can't install via Pip
Tried to install via pip install cupy but got the following error:
Collecting cupy
Using cached cupy-2.3.0.tar.gz
Complete output from command python setup.py egg_info:
cc1plus: warning: command line option '-Wstrict-prototypes' is valid for C/ObjC but not for C++
cc1plus: warning: command line option '-Wstrict-prototypes' is valid for C/ObjC but not for C++
/home/ubuntu/anaconda2/envs/thingy/bin/../lib/gcc/x86_64-conda_cos6-linux-gnu/7.2.0/../../../../x86_64-conda_cos6-linux-gnu/bin/ld: cannot find -lcuda
Options: {'profile': False, 'linetrace': False, 'annotate': False, 'no_cuda': False}
Include directories: ['/usr/local/cuda/include']
Library directories: ['/usr/local/cuda/lib64']
Cannot build a stub file.
Original error: command '/home/ubuntu/anaconda2/envs/thingy/bin/x86_64-conda_cos6-linux-gnu-c++' failed with exit status 1
**************************************************
*** WARNING: Cannot link libraries: ['cublas', 'cuda', 'cudart', 'curand', 'cusparse', 'nvrtc', 'nvToolsExt']
*** WARNING: Skip installing cuda support
*** WARNING: Check your LDFLAGS environment variable
**************************************************
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/tmp/pip-build-r2jn275v/cupy/setup.py", line 32, in <module>
ext_modules = cupy_setup_build.get_ext_modules()
File "/tmp/pip-build-r2jn275v/cupy/cupy_setup_build.py", line 378, in get_ext_modules
extensions = make_extensions(arg_options, compiler, use_cython)
File "/tmp/pip-build-r2jn275v/cupy/cupy_setup_build.py", line 268, in make_extensions
raise Exception('Your CUDA environment is invalid. '
Exception: Your CUDA environment is invalid. Please check above error log.
----------------------------------------
Command "python setup.py egg_info" failed with error code 1 in /tmp/pip-build-r2jn275v/cupy/
About this issue
- Original URL
- State: closed
- Created 6 years ago
- Comments: 17 (8 by maintainers)
Actually,
pip install --pre cupy-cuda90solved the problem.Cloud you use latest version? You can run
pip install cupy --precommand. We improved message during installation. Maybe you can get more useful information.And, we provides binary package as experimental ( link ).
As we now provide wheels (cupy-cudaXX) for stable releases, generally you don’t have to specify
--pre(unless you want to install pre-release (alpha/beta/rc) versions). See the installation guide for details. It now covers the installation steps to install wheels: https://docs-cupy.chainer.org/en/stable/install.html#install-cupy