I kinda doubt any article that thinks the X1000 series did audio.
Also full of issues; fist to think 2-3% difference is statistically significant
, second only seeing improvement on a second run (which usually is faster on same settings [as is known to 3Dmarkers for years]) not run both back to back with multiple runs. Then also you are running HD-audio versus no HD-Audio, not AMD-Audio versus non-AMD-HD-Audio.
All it says to me is maybe enabling HD-Audio slows down gameplay, wow, guess what, happened on my Audigy too.
Tests need to be done properly and in a way that resembles scientific method, this 'review' is little better than a forum post.