Vick:
I know the effect you describe. It is no bug or memory problem, it has to do with the starting point in random space. Illustration of what happen. Suppose the numbers below are steps in random space. Step 46 Contains a solution from wich LottoSynchronizer picks 1 or 2 numbers.
When you start the prediction, this could happen:
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47...
Or this could happen:
41 42 43 44 45 46 47 48 49 50 ...
You see, in the second run, the starting point in random space was much closer then in the first try, thus the second run will come up much faster with numbers. It is not possible to control this starting point, since we don't know at forehand where the next solution will be situated.
The same problem comes up in the version used for the Free Jackpot Chase. Only much worse since the criteria to pick it's numbers are harder. That's why 60 people are needed. When the "full solution" is for example situated in step 5598732 we hope that at least 1 computer is started close enough to that point, to reach the solution in time. So, in fact even an old pII could come up with the right answers. It just depends...