python - 如何将 NumPy 链接到 Eigen3，或 OpenBLAS、GotoBLAS2、ATLAS？

Question

我安装了 Anaconda3(使用 Python3.7)，发现 NumPy 扩展通过 numpy.__config__.show() 链接到 MKL:

mkl_info:
    libraries = ['mkl_rt']
    library_dirs = ['C:/Anaconda\\Library\\lib']
    define_macros = [('SCIPY_MKL_H', None), ('HAVE_CBLAS', None)]
    include_dirs = ['C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\include', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\lib', 'C:/Anaconda\\Library\\include']
blas_mkl_info:
    libraries = ['mkl_rt']
    library_dirs = ['C:/Anaconda\\Library\\lib']
    define_macros = [('SCIPY_MKL_H', None), ('HAVE_CBLAS', None)]
    include_dirs = ['C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\include', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\lib', 'C:/Anaconda\\Library\\include']
blas_opt_info:
    libraries = ['mkl_rt']
    library_dirs = ['C:/Anaconda\\Library\\lib']
    define_macros = [('SCIPY_MKL_H', None), ('HAVE_CBLAS', None)]
    include_dirs = ['C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\include', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\lib', 'C:/Anaconda\\Library\\include']
lapack_mkl_info:
    libraries = ['mkl_rt']
    library_dirs = ['C:/Anaconda\\Library\\lib']
    define_macros = [('SCIPY_MKL_H', None), ('HAVE_CBLAS', None)]
    include_dirs = ['C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\include', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\lib', 'C:/Anaconda\\Library\\include']
lapack_opt_info:
    libraries = ['mkl_rt']
    library_dirs = ['C:/Anaconda\\Library\\lib']
    define_macros = [('SCIPY_MKL_H', None), ('HAVE_CBLAS', None)]
    include_dirs = ['C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\include', 'C:\\Program Files (x86)\\IntelSWTools\\compilers_and_libraries_2016.4.246\\windows\\mkl\\lib', 'C:/Anaconda\\Library\\include']

就我使用 Ryzen CPU 而言，所以我希望我的 NumPy 链接到 Eigen3、OpenBLAS、GotoBLAS2、ATLAS？哪些 conda 命令可以帮助我获取它？

Answer 1

首先在 conda-forge channel 上查询可用的 numpy 构建列表:

conda search -c conda-forge numpy

在列表中查找在其构建版本中包含 openblas 的 numpy 构建。然后使用 conda install 命令安装该版本。例如:

conda install -c conda-forge numpy=1.15.4=py37_blas_openblash442142e_1000