Speeding up Python

Cameron Laird claird at starbase.neosoft.com
Tue Oct 19 11:27:11 EDT 1999


In article <Pine.LNX.4.20.9910191510020.32336-100000 at fep132.fep.ru>,
Oleg Broytmann  <phd at phd.russ.ru> wrote:
>On 19 Oct 1999, TToommLL wrote:
>> I read that Industrial Light And Magic is using Python, but I don't see how
>> that is possible.  I need to do some heavy floating point and math.sin() number
>> crunching  but find that Python takes 40 times as long as C++.  I tried JPython
			.
			.
			.
>   Do low-level stuff in low-level languages, and high-level stuff in high
>lavael languages.
>
>   Do number crunching in C, compile your code into a python module using
>compiler optimisation, link with optimised (both in terms of algorithms and
>compiler) math library.
>   And glue all your modules together with Python!
			.
			.
			.
Also, acquaint yourself with NumPy.  It might speed
satisfying development by a large factor.
-- 

Cameron Laird           http://starbase.neosoft.com/~claird/home.html
claird at NeoSoft.com      +1 281 996 8546 FAX




More information about the Python-list mailing list