I seem to recall a good deal of the article being devoted to trying to figure out how the W2k machine could be 1/3 faster than the XP set up. They had assumed, (like I had) that since a Win2k/Office2k machine ran faster than a Win98/Office98 machine, the XP machine was going to rock.
Needless to say it was all rather disappointing.