That's a given. 20% above GF4600 levels imply roughly 15000 3DMarks. The only ones who are in this ballpark are running seriously overclocked P4 systems. (And equally seriously overclocked 4600s. )
When it was introduced, 3DMark2001 was regarded as stressing the video card almost exclusively. For the hardware available at the time, this was understandable. With todays better performing cards, this is no longer even close to being the case (Game tests).
Too bad you're comparing minimum framerate with average framerate, or those graphs might actually mean something when placed next to each other. Or were you just using them to say that the Kyro II can be faster than a GeForce 2 in some situations, and therefore trying to claim that it should somehow get a better 3DMark score, even though it doesn't support half of the features 3DMark tests for?