Graphing the Linux PRNG

As mentioned previously, I’ve been tinkering with the Linux pseudo-random number generator. I’m now ready to present my findings having had plenty of time to hack up code and mess with it. I even graphed the output with gnuplot thanks to the awesome interface provided by SciPy!

I basically stuck to using the cryptographically unsafe /dev/urandom because it’s orders of magnitude faster than using /dev/random which will usually only give you around 8 bytes at a time. Check out my frequency plots of different sample sizes (the x axes are the byte numbers from 0-255):

You can see that the distribution of stuff is pretty random. Cool eh? I used Python to do it. You can grab my actual code and the pretty-printed HTML version (thank you KDevelop!).

  1. No comments yet.

  1. No trackbacks yet.