<igouy / yahoo.com> wrote:
> It scales badly
> 
>         0.496s gcc 5x5 (without print statements)
> 30055.098s gcc 6x6 (without print statements)
> 
> imo If Peter has figured out what he's trying to do then it would be
> smart to find a better way of doing it rather than scrabbling for small
> percentage improvements.

Sure it would, though his web page implies that different algorithms
aren't welcome. For example, <http://jsnell.iki.fi/tmp/latin.lisp>
will find the squares for a 6x6 in 180 seconds [*], compared to 13000
seconds for the C version with -O3. Both times are without output.

If the results are printed, the Lisp code will take about 1500
seconds, which doesn't make for a very sensible "time taken for real
work" / "time taken for output" ratio.

[*] X2 3800+, SBCL 0.9.15.6

-- 
Juho Snellman