there are basically two measures of "good" when it comes to RNG's...
you can just do a histogram to 1 billion... and check the output over some interval.
you can look at what
wikipedia[
^] says
Or use: P. L'Ecuyer and R. Simard, TestU01: "A C Library for Empirical Testing of Random Number Generators", ACM Transactions on Mathematical Software, 33, 4, Article 22, August 2007.
Never used it myself.