Question about test comparisons

I have recently took a closer look at the test used to compare Intel's and AMD's processors (3.2 vs 3200+). I am concerned over one of the test used however (outside synthetic tests). The test was the XMpeg test. As a consumer, i know the impact that Tom's Hardware has to me, buying my next generation components. And since i do alot of video encoding, i want the best materials for the criteria.

What struck me was that the iDCTs used were biased in a way. Since the only SSE iDCT that comes with XMpeg was Peter Gubanov's "DVD2AVI SSEMMX". Yet the iDCT used for the athlon is most likely Peter Gubanov's "DVD2AVI MMX" or some other iDCT that was MMX optimized (which i learned from the test setup and details page). Why didn't this site use "MCMAB 3dNow! enh." iDCT for the athlon to do the comparisons, as it would give the exact results the athlon and the p4 _are_ capable of? It's like using Direct3D on games like Unreal or Unreal Tournament on a voodoo3 when Glide is such a better option.

If i wasn't an informed or experienced reader, i could have been totaly mislead.