Thank you for the feedback!

Thanks to your suggestion, in the project's future, I may conduct a performance test to compare the library against others, such as the ones you mentioned.

Thank you for the feedback!

According to this feedback data that I now have, I should add to the documentation that I plan to optimize the library in the future as the project progresses.

if the target is embedded systems, usually things like naive GEMM are faster than tiled approaches for small matrices (which are probably what's used on these systems).