Yeah, none of these tight loop benchmark tasks should be happening in python, which is why you won't see anywhere NEAR this kind of speedup with "normal" python code, that makes any attempt to use the many c python libraries that exists for these (like the built in sort functions, ffs).

Over the years, I've tried pypy and whatnot, but I've only ever seen between -30% and 30% speedups, depending on the code.