I've been fiddling around, trying to find my way to a maximally performant installation of NumPy on my new Apple MacBook Air M2 and have arrived at this NumPy configuration:
(env) dbanas@Davids-Air prj % python3 -c "import numpy as np; np.show_config()"
openblas64__info:
libraries = ['openblas64_', 'openblas64_']
library_dirs = ['/usr/local/lib']
language = c
define_macros = [('HAVE_CBLAS', None), ('BLAS_SYMBOL_SUFFIX', '64_'), ('HAVE_BLAS_ILP64', None)]
runtime_library_dirs = ['/usr/local/lib']
blas_ilp64_opt_info:
libraries = ['openblas64_', 'openblas64_']
library_dirs = ['/usr/local/lib']
language = c
define_macros = [('HAVE_CBLAS', None), ('BLAS_SYMBOL_SUFFIX', '64_'), ('HAVE_BLAS_ILP64', None)]
runtime_library_dirs = ['/usr/local/lib']
openblas64__lapack_info:
libraries = ['openblas64_', 'openblas64_']
library_dirs = ['/usr/local/lib']
language = c
define_macros = [('HAVE_CBLAS', None), ('BLAS_SYMBOL_SUFFIX', '64_'), ('HAVE_BLAS_ILP64', None), ('HAVE_LAPACKE', None)]
runtime_library_dirs = ['/usr/local/lib']
lapack_ilp64_opt_info:
libraries = ['openblas64_', 'openblas64_']
library_dirs = ['/usr/local/lib']
language = c
define_macros = [('HAVE_CBLAS', None), ('BLAS_SYMBOL_SUFFIX', '64_'), ('HAVE_BLAS_ILP64', None), ('HAVE_LAPACKE', None)]
runtime_library_dirs = ['/usr/local/lib']
Supported SIMD extensions in this NumPy install:
baseline = NEON,NEON_FP16,NEON_VFPV4,ASIMD
found = ASIMDHP
not found = ASIMDDP,ASIMDFHM
I was curious as to whether the installed OpenBLAS runtime libraries were native. So, I started to probe, but...
(env) dbanas@Davids-Air prj % ls /usr/local/lib/
ls: /usr/local/lib/: No such file or directory
(env) dbanas@Davids-Air prj % ls /usr/local/
bin
(env) dbanas@Davids-Air prj % ls /usr/local/bin
2to3 idle3.11 pydoc3 python3-config python3.11-config
2to3-3.11 pip3 pydoc3.11 python3-intel64 python3.11-intel64
idle3 pip3.11 python3 python3.11 stack
So, how did I get through a successful installation of NumPy pointing it at non-existent OpenBLAS libraries?!
