[issue14757] INCA: Inline Caching meets Quickening in Python 3.3

[Stefan Krah]

Stefan Krah <stefan-usenet at bytereef.org> added the comment:

> > Modules/_decimal/tests/bench.py:
> > --------------------------------
> >
> > Not much change for floats and decimal.py, 8-10% slowdown for _decimal!
> 
> This result is not unexpected, as I have no inline cached versions of
> functions using this module. The derivatives I generate work for Long,
> Float and Complex numbers (plus Unicode strings and some others.)

But I couldn't detect a speedup for either float or int in bench.py. Also,
in perf.py I'm consistently getting slower results for float with the patch.

> there is a clear need, of course I can look into that and add these
> derivatives (as I said, there are still some 40+ opcodes unused.)

I'd say that the minimum requirement for a performance enhancement
patch is that there aren't any slowdowns. :)

I'm getting a 9% speedup for mandelbrot and an 18% speedup for spectralnorm.
That's nice, but many benchmarks that Antoine and I have posted show
slowdowns of 15-18%.

----------

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue14757>
_______________________________________