It's been a while since I used PyPy, but JITs in general warm up over time. If you have hot loops with heavy arithmetic and no branches, that's usually best cas

It's been a while since I used PyPy, but JITs in general warm up over time. If you have hot loops with heavy arithmetic and no branches, that's usually best case scenario for JITs. If you have branchy non-uniform control flow, that's the worse scenario. So it really depends on your usage - you may be paying the JIT overhead costs with little benefit.