[Shootout-list] Latest Updates

[Brent A. Fulgham]

Isaac Gouy wrote:

>--- "Brent A. Fulgham" <bfulgham@debian.org> wrote:
>  
>
>>I'm probably not going to add too many more languages, since it's 
>>already taking hours to run the tests. 
>>    
>>
>Could the test approach be changed the reduce the runtime?
>
>Is every microbenchmark re-tested everytime?
>  
>
No.  The long runtimes are only when I periodically do a 'make clean' 
and rerun everything.  I sometimes do this to make sure things are 
updating properly.  Otherwise, only source file changes spur new runs.

The problem is more the annoyance of waiting for ElastiC or even Erlang 
to finish certain programs.  I'm going to take a look at the Erlang 
implementations, to see if I can figure anything out.  I'm afraid 
Elastic is just downright slow :-)

>Smarter algorithm rather than faster machine ;-)
>  
>
Yes -- that's quite true.  Nowadays there is a mindset that Moore's law 
will save us from our own excesses.  However, anyone who has read the 
excellent "Programming Pearls" or Knuth's algorithms books has seen the 
evidence of the massive changes in runtimes found by correcting 
inefficient algorithms...

-Brent