[MLton] MLTon speed (Languages Shootout)

Chris Clearwater chris@detriment.org
Tue, 01 Mar 2005 14:53:35 -0800

Alexandre wrote:

> Hello,
> just see updated computer language shootout benchmark  
> (http://shootout.alioth.debian.org/benchmark.php? 
> test=all&lang=all&sort=cpu&xfullcpu=1&xmem=1&xloc=0&ackermann=1&wc=3&ech 
> o=5&fannkuch=3&fasta=3&fibo=1&harmonic=1&heapsort=4&mandelbrot=4&matrix= 
> 3&nbody=3&nsieve=3&nsievebits=2&objinst=5&methcall=5&pidigits=2&random=3 
> &regexmatch=4&reversefile=4&spellcheck=4&hello=0&moments=2&sumcol=3&takf 
> p=1&process=2&message=3&wordfreq=5) and it seems that SML/NJ  
> implementation became faster... I mean that there is no such big  
> perfomance lag as it was later. Can you comment this (I'm just 
> curious  why)?
> Regards,
> /Alexandre.

It appears SML/NJ has all the tests for the shootout implemented.  Most 
implementations are missing a few and don't recieve points for those 
tests. If you give all the tests which have a "(new)" suffix a weight of 
0, SML/NJ falls pretty down the list. Also I think that the shootout 
mailing list (shootout-list@lists.alioth.debian.org) would probally be 
the most appropraite place for such a question as it doesn't involve 
MLton at all :-)