Interesting to see that their test results are often very good (most of the NIST tests), but in certain other tests (Diehard Overlapping sums) there is p value inflation. I wonder where that comes from.