[Haskell-cafe] New Benchmark Under Review: Magic Squares

[Brent Fulgham]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


On Jul 4, 2006, at 5:20 PM, Daniel Fischer wrote:
>> I would propose modifying the other entries (since there are only a
>> handful) to match the output of your original solution.
>>
>> What do you think?
>
> Cool, though the problem of exploding runtime remains, it's only  
> pushed a
> little further. Now I get a 5x5 magig square in 1 s, a 6x6 in 5.4  
> s, but 7x7
> segfaulted after about 2 1/2 hours - out of memory, I believe.

Hrm.  Well, I still prefer the growth of search space in your version  
over the
original, since it *was* going from 0.01 s (3x3) to 0.10 (4x4) to 4  
hours (5x5).

Going 1s->5.4s ->x hours is at least a bit more controlled.....

I wonder if anyone can propose a slightly smaller problem, or a  
better algorithm?

Thanks,

- -Brent
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (Darwin)

iD8DBQFEq0b7zGDdrzfvUpURAumWAJ4itR4eayB3mj5hYEtxbK630mF4IgCeO3PA
qFF7cLTW4xk36J/nQOON+F4=
=C1xL
-----END PGP SIGNATURE-----