[Shootout-list] New benchmark?

[Greg Buchholz]

--- Isaac Gouy <igouy2@yahoo.com> wrote:

> > I can see forcing the overall algorithm to be the same, but I
> > can't see why something like the implementation in the
> > reference paper
> 
> So maybe what we're looking for is "same algorithm" (and
> sometimes "same data-structure").
> 
  Or maybe we're talking about same worst case complexity? 
Especially if you're using a really high level language, and you
don't really know what algorithm your compiler eventually spits
out (i.e. maybe your compiler can convert O(n) lists to O(1)
arrays).  I'll throw out a very hypothetical example here.  Assume
we are talking about one of the recursive benchmarks.  Language
A's compiler recognizes the algorithm is tail-recursive and
optimizes so as to run in constant stack space.  Language B, on
the other hand, doesn't have the tail call optimization, but
instead has efficient continuations.  Would we allow B to use the
best features of the language, or is it better to test just how
bad recursive subroutines are?


Greg Buchholz




		
__________________________________ 
Do you Yahoo!? 
Yahoo! Mail - now with 250MB free storage. Learn more.
http://info.mail.yahoo.com/mail_250