Are any of these still needed, though? I kinda assumed they would be like Duff's Device and be a nifty historical relic, but firmly in the "Let the compiler do this for you" category in modern C.
Many of those are not standard functions (or were not, until very recently). You have to implement it somehow, so having a ready-made collection that also performs well is very nice to have.
What code specifically do you write instead of the code in TFA, in order to have the optimizer generate the code in TFA? In general compiling with optimizations enabled has not produced particularly good code in my experience.
