asm intrinsics for speed

Hi Slava,

I'm building (what amounts to) an olap database engine in factor, and 
part of that involves linear searching mmapped files. Basically the 
faster I can do linear searching operations, the less I need to use 
sorted indexes and the more data I can fit into memory (esp with 
compression). I've currently got a bunch of C code doing searches, but 
ideally I'd like to not rely on a separate C dll in the released code or 
require people to have the gcc tool chain to compile it.

I decided to experiment with using asm intrinsics to implement the tight 
loops, and took 'count' as a simple testcase. I'm searching a 48M file 
of 32bit integers.

By comparison some C code (minus the mmaping stuff):

int count_items(int *from,int *to,int item) {
   int count = 0;
   while(from != to){
     if(*from == item) count++;
     from ++;
   }
   return count;
}

Which when compiled with -O3 takes about 80ms to search the file on a 
processor locked at 800MHZ with the filebuffers running warm. The best I 
could do with factor in this setup was a few seconds (which I think is 
still pretty good).