OSError: The CUDA lib64 path could not be located in /usr/lib64

Question

我该如何解决这个问题？

(cv) jalal@klein:~/computer_vision/py-faster-rcnn/lib$ make
python setup.py build_ext --inplace
Traceback (most recent call last):
  File "setup.py", line 58, in <module>
    CUDA = locate_cuda()
  File "setup.py", line 55, in locate_cuda
    raise EnvironmentError('The CUDA %s path could not be located in %s' % (k, v))
OSError: The CUDA lib64 path could not be located in /usr/lib64
Makefile:2: recipe for target 'all' failed
make: *** [all] Error 1

我正在使用这个： https://github.com/rbgirshick/py-faster-rcnn

Answer 1

为那些可能不再使用这个流行软件（Cuda ~5）的人使用以下更改解决了这个问题：

在~/computer_vision/py-faster-rcnn/lib/setup.py

变化：

cudaconfig = {'home':home, 'nvcc':nvcc,
              'include': pjoin(home, 'include'),
              'lib64': pjoin(home, 'lib64')}

至：

cudaconfig = {'home':home, 'nvcc':nvcc,
              'include': pjoin(home, 'include'),
              'lib64': pjoin(home, 'lib')}

如果您使用的是 Cuda7.5，您可能需要将其保留为 lib64，否则可能会出错。

Answer 2

如果没有设置CUDAHOME，也可能会出现这个错误。尝试：

$ export CUDAHOME=/usr/local/cuda-8.0

Answer 3

对于Ubuntu 17.04

将 /home 文件夹中的 .bashrc 文件中的 CUDAHOME 设置为，

export CUDAHOME=/usr/lib/nvidia-cuda-toolkit/

然后执行source ~/.bashrc

记得打开一个新的终端才能使用新的环境变量。

从

更改setup.py文件

cudaconfig = {'home':home, 'nvcc':nvcc,
                  'include': pjoin(home, 'include'),
                  'lib64': pjoin(home, 'lib64')}

到

cudaconfig = {'home':home, 'nvcc':nvcc,
               'include': '/usr/include',
               'lib64': '/usr/lib/x86_64-linux-gnu'}

还需要安装 g++5，因为 nvcc 不支持 g++6。

sudo apt install g++-5

OSError: The CUDA lib64 path could not be located in /usr/lib64

OSError: The CUDA lib64 path could not be located in /usr/lib64

python

ubuntu

gpu

path