Does profiling remove some optimisations?

[andrew cooke]

I have a core loop that is critical to performance.  The code is at
http://www.acooke.org/lepl/api/lepl.parser-pysrc.html#trampoline

If I write a separate "optimised" version of that function for when
"monitor" is empty, with all the "if monitor" tests removed, the profiler
(cProfile) indicates a 10% reduction in time spent in the loop.

But if I run the same code 100 times under timeit, without profiling, I
see no difference in total time.  The process is CPU bound.

I am pretty sure the improvement seen in the prfiler is repeatable and
less than the noise in the profiling, so why don't the timeit times
change?  The only reason I can think of is that the profiler is blocking
some kind of optimisation (which seems odd, since that makes profiling
somewhat pointless).

Full disclosure - the code linked above is not quite identical to my
current workspace (I have replaced the append/pop with an explicit index
into an array whose size is managed separately).

Andrew