how can i make numpy use BLAS and LAPACK after being installed?

Question

Numpy in my computer is quite slow and I am suspecting that it is not using BLAS or LAPACK. When I run numpy.config.show(), I get the following output:

atlas_threads_info:
  NOT AVAILABLE
blas_opt_info:
    libraries = ['f77blas', 'cblas', 'atlas']
    library_dirs = ['C:\\local\\lib\\atlas\\sse3']
    define_macros = [('NO_ATLAS_INFO', -1)]
    language = c
atlas_blas_threads_info:
  NOT AVAILABLE
lapack_opt_info:
    libraries = ['lapack', 'f77blas', 'cblas', 'atlas']
    library_dirs = ['C:\\local\\lib\\atlas\\sse3']
    define_macros = [('NO_ATLAS_INFO', -1)]
    language = f77
atlas_info:
    libraries = ['lapack', 'f77blas', 'cblas', 'atlas']
    library_dirs = ['C:\\local\\lib\\atlas\\sse3']
    define_macros = [('NO_ATLAS_INFO', -1)]
    language = f77
lapack_mkl_info:
  NOT AVAILABLE
blas_mkl_info:
  NOT AVAILABLE
atlas_blas_info:
    libraries = ['f77blas', 'cblas', 'atlas']
    library_dirs = ['C:\\local\\lib\\atlas\\sse3']
    define_macros = [('NO_ATLAS_INFO', -1)]
    language = c
mkl_info:
  NOT

The problem is that those library directories do not exist on my computer. I have no idea how those directories have sneaked in to the numpy config. Is there anyway to change a setting so that numpy can use the optimized BLAS and LAPACK after being installed? Or, should I re-install it?

possible duplicate of http://stackoverflow.com/q/21671040/1461210 — ali_m, Oct 24 '15 at 10:53

score 0 · Answer 1 · answered Oct 23 '15 at 20:18

The key will be a version of a site.cfg file, tuned to your system. You can find numpy's example, site.cfg.example, here. When I did this on Linux, I cloned numpy from source, edited that file as described in the comments, and ran python setup.py install in the numpy directory, and it worked. Hope this helps.

score 0 · Answer 2 · answered Oct 24 '15 at 05:45

You are using blas. These are the libraries you are using.

 blas_opt_info:
    libraries = ['f77blas', 'cblas', 'atlas']
    library_dirs = ['C:\\local\\lib\\atlas\\sse3']
    define_macros = [('NO_ATLAS_INFO', -1)]
    language = c

You are using lapack.

lapack_opt_info:
    libraries = ['lapack', 'f77blas', 'cblas', 'atlas']
    library_dirs = ['C:\\local\\lib\\atlas\\sse3']
    define_macros = [('NO_ATLAS_INFO', -1)]
    language = f77

You are not using blas_mkl and lapack_mkl. These are proprietary implementations of those libraries.

I hope that you aren't thinking numpy will compare in speed to direct implementations of blas and lapack. It will be an order of magnitude slower.

how can i make numpy use BLAS and LAPACK after being installed?

2 Answers2