is pretty amazing. E.g., try https://pv.github.io/numpy-bench/#/regressions where you can see specific functions in numpy, and how commits changed their
performance exactly. One could imagine making something similar for the
Sage library, which is full of similar compute-intensive well-defined functions.