In my screenshot, you could observe that when we are successively doing the same operation, bytecode will be generated for the first time.
And, as the same operation is encountered again, instead to re-interpreting, it executes the pre-generated bytecode.
So, it takes lesser amount of time in the 2nd time.
This is again the reason for executing it in same amount of time in the next further loops.