For comparison purposes:

17⁷² = 3.911310908*10⁸⁸ max. attempts

63⁷² = 3.568778227*10¹²⁹ max. attempts

95⁷² = 2.489428061*10¹⁴² max. attempts

The max. attempts calculation can be a little misleading since it is a maximum, and the average password crack attempt will probably succeed in about 1/2 that number on average.

Additionally, if there is any additional known information about the password you can cut down the # of max. attempts considerably.  For instance with the prior knowledge that your password is composed of two GUID's cast to character format and concatenated, I automatically know that the following 8 character positions are always hyphens: 9, 14, 19, 24, 45, 50, 55, 60.  I also know that the remaining 64 character positions only have 16 possible characters each (0-9, A-F).  So the new calculation is:

16⁶⁴ * 1⁸ = 1.157920892*10⁷⁷

The estimates in the article you linked to appear to be using Pentium 100 MHz class machines in their estimates.  Those estimates should probably take common 2.x GHz (and higher) processor machines into consideration as well.