make faster Richards benchmark

[Peter Hansen]

Duncan Lissett wrote:

> Wilk <wilkSPAM at OUTflibuste.net> wrote in message news:<873c64zpu7.fsf at blakie.riol>...
> 
>>dlissett0 at yahoo.com (Duncan Lissett) writes:
>>
>>
>>>I'd appreciate any suggestions on how to make faster Python
>>>implementations of Richards benchmark. Perhaps there are obvious
>>>problems that can be corrected?
>>>
>>>http://www.lissett.com/ben/bench1.htm
>>
>>import psyco
>>psyco.full()
>>
>>2 times faster :-)
> 
> And Simon: "I just tried the benchmark with Psyco, and cut the run
> time for input=10000 from 8.1 seconds to 3.4"
> 
> Joseph, Simon, Wilk: thanks for the Psyco suggestion, and special
> thanks for going ahead and trying it - we get slightly better than x2.

What platform is this on?  On WinXP, with an AMD 2500+ chip, and lots
of memory etc., I'm still getting only 38% speedup, nowhere near 100%...

-Peter