From: Shawn Hargreaves <Shawn AT talula DOT demon DOT co DOT uk>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: your ma
Date: Fri, 11 Jul 1997 23:17:36 +0100
Organization: None
Distribution: world
Message-ID: <U+cmrAAAErxzEwN1@talula.demon.co.uk>
References: <199707101349 DOT AA237002579 AT typhoon DOT rose DOT hp DOT com>
 <33C68250 DOT 301AAEDC AT execulink DOT com>
NNTP-Posting-Host: talula.demon.co.uk
MIME-Version: 1.0
Lines: 37
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Precedence: bulk

Jeff Weeks writes:
>I just stubled upon a funky intruction that might do what you what.  I
>couldn't find any use for it myself... but I guess you just did :)
>
>The two instuctions are BSF (bit scan forward) and BSR (bit scan
>reverse).  They take two operands and do the following:

There is one big problem with those instructions: they are _slow_. On a
486 they can take from 6 to 42 clocks, depending on how soon a bit is
found. On a Pentium they can range from 7 to 73 clocks, which make them
pretty pointless since you can get similar speeds from a little
shift+test loop. 

As I think someone already mentioned, if you are willing to use a bit of
inline asm you can just test the value and then check the parity flag.
For example:

int parity(int val)
{
   unsigned char result;

   asm (
      " orl %1, %1 ; "        /* test the input */
      " setpb %b0 ; "         /* and store the result */

      : "=g" (result)
      : "r" (val)
      : "cc"
   );

   return result;
}


--
Shawn Hargreaves - shawn AT talula DOT demon DOT co DOT uk - http://www.talula.demon.co.uk/
Beauty is a French phonetic corruption of a short cloth neck ornament.