Thank you

[eschnett]

This is just a heads-up. I am one of the maintainers of the Einstein Toolkit https://einsteintoolkit.org which contains code to solve the Einstein equations on HPC systems with various architectures. We have, over the years, created a module to support explicit SIMD vectorization in very much the same spirit as this library (see https://bitbucket.org/cactuscode/cactusutils/src/master/Vectors/).

I have always been looking for a community standard for such a library, and I hope that std-simd will be / become this standard. At first glance, you seem to have all the required features (various architectures, integer and logical vectors, masked operations, etc.), obviously with a much cleaner design than our code.

All the best! Depending on whether I can convince my colleagues, we might start contributing towards this library – or towards VcDevel/Vc instead if you think that would be more appropriate.