From: terje.mathisen@nospicedham.tmsw.no   
      
   asetofsymbols@nospicedham.gmail.com wrote:   
   > The algo would be as this   
   >   
   > a:=input   
   > r:=0   
   > A: r:=r*10+(a mod 10); a:=a div 10   
   >    if a!=0 then goto A   
   >    if input==r then return true   
   >    return false   
   >   
   That algorithm, on a 64-bit machine with a good compiler, will require   
   one 64x64->128 MUL (and a shift) in order to calculate a div 10, then an   
   IMUL (64x64->64) and SUB to get a mod 10, while the r*10 part will   
   probably be handled with a LEA plus a shift, i.e. about 10-15 cycles per   
   digit.   
      
   Multiply by 20 digits and you end up with 300-450 cycles, vs ~60 for my   
   3-way uint64_t to decimal conversion, plus the final string reciprocal   
   test which takes about 15-30 cycles for naive code.   
      
   OTOH, if input numbers are often relatively small (5 digits or less),   
   then it would be OK to do it your way.   
      
   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)