The good DrHyde is quite correct. Another alternative would be to use integers:

for(my $x = 0; $x <= 100; ++$x){
     print $x/1000,"\n";
}
[download]

Your choice should depend on how much accuracy you really need, presuming that you will use $x in actual calculations, not just a list of values. I'd avoid the overhead of Math::BigFloat unless it was really necessary.