Question

I am trying to find out if the latest version of NumPy (2.0.0) is taking advantage of the updated Accelerate BLAS/LAPACK library, including ILP64.

Numpy

Numpy in their 2.0.0 release added support for the Apple Accelerate framework.

Having installed this version, I checked the BLAS/LAPACK linking:

>>> import numpy as np
>>> np.show_config()

Build Dependencies:
  blas:
    detection method: pkgconfig
    found: true
    include directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/include
    lib directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/lib
    name: blas
    openblas configuration: unknown
    pc file directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/lib/pkgconfig
    version: 3.9.0
  lapack:
    detection method: pkgconfig
    found: true
    include directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/include
    lib directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/lib
    name: lapack
    openblas configuration: unknown
    pc file directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/lib/pkgconfig
    version: 3.9.0

While it says found: true, I am unsure if this is using the Accelerate-linked versions of BLAS/LAPACK.

SciPy

The SciPy documentation (1.13.X) mentions that:

A recent version of OpenBLAS, MKL or ATLAS. The Accelerate BLAS library is no longer supported.
– Scipy Toolchain Roadmap

An archived SciPy GitHub-wiki entry details the reasoning for this missing support.

>>> import scipy as sp
>>> sp.show_config()

Build Dependencies:
  blas:
    detection method: pkgconfig
    found: true
    include directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/include
    lib directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/lib
    name: blas
    openblas configuration: unknown
    pc file directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/lib/pkgconfig
    version: 3.9.0
  lapack:
    detection method: pkgconfig
    found: true
    include directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/include
    lib directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/lib
    name: lapack
    openblas configuration: unknown
    pc file directory: /opt/homebrew/Caskroom/miniconda/base/envs/numpy/lib/pkgconfig
    version: 3.9.0

…this looks just like the configuration of NumPy 2.0.0… which leads me to believe the NumPy linking may not be working. Perhaps this is an issue with Conda?

Related:

• How to check BLAS/LAPACK linking in NumPy/SciPy