[pypy-issue] [issue909] My code is slower under PyPy than under cPython
tracker at bugs.pypy.org
Sat Oct 15 22:01:47 CEST 2011
Ronny Pfannschmidt <Ronny.Pfannschmidt at gmx.de> added the comment:
the example is not running enough iterations to actually benefit from the jit
also the code s full of conditionals
pypy without the jit benefit is roughly half as fast as cpython, which means
your timing results are expected
try with a much lager sample - something that takes a least seconds to run on
cpython and has a few thousand iterations
just to test the basics i did the following after removing some print statements:
$ pypy-bin -m timeit -n 2000 -s 'import main' 'main.map_directions()'
2000 loops, best of 3: 1.74 msec per loop
$ pypy-bin -m timeit -s 'import main' 'main.map_directions()'
10 loops, best of 3: 3.98 msec per loop
$ python -m timeit -s 'import main' 'main.map_directions()'
100 loops, best of 3: 2.82 msec per loop
as you can see, pypy gets about twice as fast after hitting the jit treshold
to over-stress it i also did $ pypy-bin -m timeit -n 20000 -s 'import main'
20000 loops, best of 3: 1.71 msec per loop
but there doesn't seem to be more improvement (imho 0.03 msec can be considered
status: unread -> chatting
PyPy bug tracker <tracker at bugs.pypy.org>
More information about the pypy-issue