Re: [SPOILER] Solution to Perl 'Expert' Quiz-of-the-Week #22

Randy W. Sims wrote:

> On 8/29/2004 5:08 AM, Randy W. Sims wrote:
>
>> This is a common optimization and applies to all languages. 
>> Generically, the goal is to perform each calculation only once. 
>> Although it's not always as obvious as the code above. In particular 
>> this implies reducing the number of temporaries which are created 
>> behind the scenes, especially in languages like perl. This is why 
>> complex one-line statements are often less efficent than the longer 
>> equivelant: the shorter code creates multiple and expensive temporaries.
>
>
> Benchmark: timing 10 iterations of compound, simple...
>   compound: 866 wallclock secs (726.63 usr +  0.94 sys = 727.57 CPU) @ 
>  0.01/s (n=10)
>     simple:  4 wallclock secs ( 3.52 usr +  0.05 sys =  3.58 CPU) @ 
> 2.80/s (n=10)
>          s/iter compound   simple
> compound   72.8       --    -100%
> simple    0.358   20246%       --
>

You must have been using the web2 list, not words. (names relative to 
this list)

The interesting thing is that with the 'words' list, compound is faster 
than simple:

D:\Code\QotW\22e>bench.pl words
           Rate compound   simple
compound 5.07/s       --     -63%
simple   13.9/s     174%       --


where as with 'web2', compound is slower:

D:\Code\QotW\22e>bench.pl web2
         s/iter compound   simple
compound   3.99       --     -45%
simple     2.18      83%       --


For reference:

D:\Code\QotW\22e>wc -l words web2
  25143 words
 234936 web2
 260079 total

My is is same as posted, I just added support for changing the dict on 
the command line.


Running on perl v5.8.4 built for MSWin32-x86-multi-thread (ActiveState 
build 810)

-- Rod Adams