Hacker News Reader

Not general or universal. Only for pre-trained data and with abysmal worst cases.

I'm not positive what you mean here. Are you saying the discovered algorithm isn't actually good? Didn't LLVM accept it?