From: terje.mathisen@nospicedham.tmsw.no   
      
   aen@nospicedham.spamtrap.com wrote:   
   > Hi!   
   >   
   > Have I got this right from Agner Fog's Optimizing subroutines in   
   > assembly language, pp 148-149?  Especially do I assume the maximum   
   > number correct?   
      
   It looks like you want both the decimal number and the check for it   
   being a palindrome, right?   
      
   Otherwise, if you justneed the test for palindromicity, and most inputs   
   will not be so, then you want a faster test, right?   
      
   Start by using the number of set bits to estimate how many digits you   
   will have, then extract the bottom and top digits with reciprocal muls   
   (using lookup table reciprocals) and compare them.   
      
   If you need both returns then the best you can do is to use a 64-bit   
   version of the algorithm I first posted maybe 20 (?) years ago which   
   splits the number into two parts, scales them into n.nnnnnn format and   
   then extract the digits in parallel starting from the top and using LEA   
   to multiply by 5 between each digit.   
      
   For 32-bit unsigned that algorithm uses about 30 cycles on a 10+ year   
   old cpu, the 64-bit version, with 3 or 4-way split, should do it in less   
   than twice that time.   
      
   Terje   
      
      
   --   
   -    
   "almost all programming can be viewed as an exercise in caching"   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)