Question

I am asked:

Using your raspberry pi, write a python script that determines the randomness
of /dev/random and /dev/urandom. Read bytes and histogram the results.
Plot in matplotlib. For your answer include the python script.

I am currently lost on the phrasing “determines the randomness.”

I can read from urandom and random with:

#rb - reading as binary
devrndm = open("/dev/random", 'rb')
#read to a file instead of mem?
rndmdata = devrndm.read(25) #read 25bytes

or

with open("/dev/random", 'rb') as f:
    print repr(f.read(10))

I think the purpose of this exercise was to find out that urandom is faster and has a larger pool than random. However if I try to read anything over ~15 the time to read seems to increase exponentially.

So I am lost right now on how to compare ‘randomness’. If I read both urandom and random in to respective files how could I compare them?

Answer 1

Could it be as simple as:

In [664]: f = open("/dev/random", "rb")
In [665]: len(set(f.read(256)))
Out[665]: 169


In [666]: ff = open("/dev/urandom", "rb")
In [667]: len(set(ff.read(256)))
Out[667]: 167


In [669]: len(set(f.read(512)))
Out[669]: 218

In [670]: len(set(ff.read(512)))
Out[670]: 224

ie. asking for 256 bytes doesn’t give back 256 unique values. So you could plot increasing sample sizes against the unique count until it reaches the 256 saturation point.