Different results on different machines

Hi All

I made thermal conductivity calculations of nanowires using the same input files on two different machines.
But the results I got are completely different ! I dont know the way out


If the 2 runs tracked each other closely (based on the thermo output)
for a few 100s or 1000s of steps and then diverged, that is normal
behavior for round-off differences in 2 different runs, similar
to what might happen if you ran on twice on different #s of processors.

In that case, there is statistical variation in the quantity you are measuring.