[MLton] Re: [Haskell-cafe] fastest Fibonacci numbers in the West

Wesley W. Terpstra wesley@terpstra.ca
Thu, 27 Jan 2005 19:51:24 +0100


On Thu, Jan 27, 2005 at 01:33:17PM -0500, Matthew Fluet wrote:
> > Excedrin reported the bug on #sml last night.  I saw it and ran a
> > couple of tests.  The bug he reported was for William's version of the
> > code with an input of 99999999.  I was not able to reproduce the bug
> > using the gmp-4.1.2-2 RPM on my Linux machine.  
> 
> Hmm... I seem to be running on gmp-4.1.2 on the Linux machines I ran the 
> tests with.

I used:
ii  mlton          20041109-1     Optimizing compiler for Standard ML
ii  libgmp3-dev    4.1.4-5        Multiprecision arithmetic library developers
on a debian system.

> By having wesley compute  exp (fib, n + 1)  instead of  exp (fib, n).

Yes, I counted from 1. :-)

I think a good thing to check would be a C version using libgmp.
That might help in determining if MLton has any of the blame...

-- 
Wesley W. Terpstra