[Shootout-list] X per second scoring system, resume
Einar Karttunen
ekarttun@cs.helsinki.fi
Fri, 1 Oct 2004 08:34:30 +0300
On 30.09 14:59, Brandon J. Van Every wrote:
> > Lets look at haskell for example (I don't think it is one of the worst
> > ones).
> >
> > If the timer is implemented with a signal then you run into the
> > following problems:
>
> Why are we talking 'if' ? Does Haskell have a timer call available or
> not? How is it implemented? Or is your point that Haskell is really
> really lazy and flow control for something like a system timer is always
> going to be really really weird? I imagine the Haskell community runs
> benchmarks on something, how do they solve their problems?
Haskell community likes usually to profile or just use time(1).
If two solutions are nearly the same in performance, then it is
easy to go for solution. The fact "language A seems to be 0.5% faster
than B" is not what we are after..
Also small differences can change order with different test platforms.
Of course Haskell does provide an interface to alarm(2), but that
does not make it easier to get the intermediate status of the
computation.
> Are you saying Haskell is silly and has done nothing practical to make
> it possible to accurately benchmark in the language?
No. I am saying that measuring the amount of work done before an
asynchronous interrupt occurs is nontrivial.
- Einar Karttunen