Re: [Gcl-devel] gcd proposal

[Andrei Zorine]

I found the algorithm in Knuth's TAOCP vol.2. Then I decided to find out 
 how Maxima comeputes gcd and finally found out that it delegates it to 
gcl. Then I patched my gcl-2.5.2 (and sent the patch to you). Later I 
found words 'bin-gcd' in grep's output. I think it's ok to have binary 
gcd in gcl instead of Maxima. I haven't seen how exactly the new code is 
installed, I don't know what to look at. I hope you simply replaced the 
get-gcd's body thith the new one.

I have also found that new gcd code works faster for fixnums than 
clisp's one, but still is second with bignums. The next posible step 
would be either to implement binary gcd for bignums, or to call mpn_gcd 
directly from get-gcd. I don't know which is faster yet. This needs more 
testing. (and reading Knuth as well :))


Camm Maguire wrote:
> Greetings!  One other thought on this.  Given that you found it in
> lisp in maxima, you probably need/want this code inlined in compiled
> lisp functions calling gcd (with declared fixnum args of course).
> Please let me know if this is the case, and we can instruct the
> optimizer to inline this snippet.
>
> Take care,