JB> If numbers can't be repeated, then the algorithm does not generate
JB> random numbers.  In order to draw a random number from a set of
JB> 65536, you'd have to "put the number back in the hat" after each
JB> draw. In other words, every one of the 65536 should have an equal
JB> chance of being drawn on every attempt, which means there are going
 
The catch is a 65536 generator is not good for the entire range, but is ood
for a range of say, 6 (dice rolls).  My CRC/Random number generators have 
een
tested by some RNG testers, and they are quite good random number generators
as longg as you don't want a full 65536 combinations.  If you want 65536
posible outputs, then you use the routine that calcs 2 ^ 32 combinations
modulus 65536.  You seemed to have missed that part of my message.
Most people don't understand that part of RNG's.  I probablt also don't make 
it appearent.  An RNG for 32 bits is really only good about 2/3's its range.  
By using a MODULUS you enable the RNG to repeat.  I ran my 32 bit RNG through 
a CHI-SQUARE test.  I have had to make some changes since I orginally posted 
the information here, in order to get it to pass the CHI-SQUARE test most the 
time. (Most RNG's only pass it about 90% of the time).  
If Fred gives any interrest in this, I will FREELY give him example code.  
ut
Fred seems not to think much of me, and I probably shouldn't post around here 
much, don't use commo much, if at all, any more, as I don't do any 
communications of that kind...
Carl Morris
--- Renegade v10-05 Exp
---------------