EW>Herman, you wrote to Darin McBride:
 HS> Since you obviously fail to comprehend this I shall demonstrate.
 HS> Loop1 :-
 HS> #define bSize(99999)
EW>There isn't anything to optimize. Macro parameters should be
EW>identifiers. ;-)
 HS> Oh, but you being so proffessional and all, you probably
 HS> implement loops that look like this..
  HS> Loop2 :-
EW>I would use memcpy() since I know the arrays don't overlap. It is
EW>shorter, better readable and probably faster.
 HS> Loop3 :-
EW>Why waste all that precious space, watch your cache and use Duff's
EW>device
I did use Duffs device. Look at the message again if you doubt.
EW>/* original code by Tom Duff, adapted from the Jargon File */
EW>/* I am not sure whether it is PD */
EW>   int n = (bufSize+3)/4;
EW>   char *to=buf1, *from=buf2;
EW>   assert ( bufSize > 0 );
EW>   switch ( bufSize % 4 )
EW>   {
EW>   case 0:        do {  *to++ = *from++;
EW>   case 3:              *to++ = *from++;
EW>   case 2:              *to++ = *from++;
EW>   case 1:              *to++ = *from++;
EW>                      } while ( n-- > 0 );
EW>   }
EW>Bye,
EW> Erik
... I'm not dead, I'm metabolically challenged.
--- Ezycom V1.48g0 01fd016b
---------------