Tiny computers vs. Big Languages

Wolfgang Strobl ws at mystrobl.de
Mon Aug 26 21:37:07 CEST 2002


Andrew MacIntyre <andymac at bullseye.apana.org.au>:

>On Sun, 25 Aug 2002, Wolfgang Strobl wrote:
>
>> - To my surprise, the Python version of compare is noticeably faster
>> with Python 2.3 (cvs version from friday) than with 2.2.1:  42 µs
>> instead of 50 µs for a single call.
>
>There have been a number of performance enhancements applied to CVS of
>late, including efforts to reduce internal function call overhead.
>
>2.3 should be generally a better performer than 2.2, although new-style
>classes still appear to be carrying more baggage than GvR would like.

Anyway, I'm glad that these performance enhancements where quite
noticeable in an inner loop of a real world example.  On the other
hand, a 20 % better perfomence isn't really good enough for my
purposes.  I'd like to ask whether anybody has an idea about how to
tackle algorithmic problems of this kind. Perhaps I'my spoiiled by
implementing stuff like this in low level languages like asm, pli or c
too often.   Don't misunderstand it as a critique - it annoys me that
implementing "compare" in c or assembler feels "more natural" to me
(and is _much_ faster)  than implementing the function in Python.


-- 
Wir danken für die Beachtung aller Sicherheitsbestimmungen



More information about the Python-list mailing list