[Cryptech Tech] My analysis of the Cryptech RNG data

[Russ Housley]

Thanks for providing the rosc_entropy_2014-12-18.bin.  It contains 160000000 bytes of very solid random data.

I found a bug in the program that I was using, and that bug lead me astray on the analysis of the previous file.  Sorry for the alarm.  All is well.

I wrote a routine to do a broader look at the constant value test, the the results are quite impressive.  Here they are:

Probability of adjacent bytes with the same value is 1/2**8
Ideal value for repeated bytes is 625000
Count of repeated bytes: 624051

Probability of three adjacent bytes with the same value is 1/2**16
Ideal value for repeated 3-byte blocks is 2441
Count of 3-byte matches: 2400

Probability of x y x y pattern is 1/2**16
Ideal value for the x y x y pattern is is 2441
Count of x y x y pattern: 2431

Probability of x y x x pattern is 1/2**16
Ideal value for the x y x x pattern is 2441
Count of x y x x pattern: 2381

Probability of four adjacent bytes with the same value is 1/2**24
Ideal value for repeated 4-byte blocks is 9
Count of 4-byte matches: 9

Russ