Additional Resources

The cannonical resources[1] for OpenBLAS are of course the OpenBLAS wiki and the the Netlib BLAS documentation, especially the bibliography and the final BLAS techincal forum standard.

• Naming conventions are noted here

• The reference documentation has the scheme too

• Also elaborated in the embedded context here

Conventions and Implementations

Beyond the standard netlib reference, here are some more common implementations of interest, with above average documentation[2]:

• Intel oneAPI MKL

• Arm performance library implementation (armpl.h)

• The IBM Engineering and Scientific Subroutine Library

• SGI’s Scientific Computing Software Library

Pedagogical Work

Here are some additional links for BLAS related information:

• Basic scientific python tutorial, with a notebook here

• This lecture on Numerical Linear Algebra Software

• Another lecture on Numerical Linear Algebra Background

• Some context on CPU performance

Useful extensions

• FlexiBLAS

  • Supported in SciPy

Python Ecosystem

• PyPackaging Native Overview

• SciPy document on debugging linear algebra issuse

• Tarballs built on CI

• NumPy documentation on using different BLAS

Testing this is still a matter of manual timing apparently.

---

[1]

Barring the wiki page, which can be a good place to start too.

[2]

This post has some NumPy specific context, as does the documentation