Well aren't we the propagandist? Why don't we examine the facts a bit? 3DNow is great, but all it really ends up doing is boost the soft FPU performance of the K6-2. Enabled on a Pentium II, I don't think it would make much of a difference at all, considering the fact that the P2 already has a strong, pipelined floating point unit. I don't want to get too deep into the details, but do consider the fact that Intel has to cater to more than just people who play games
Don't get me wrong, I certainly have no doubt that instruction-optimized 3D and SIMD are going to play a huge part in the upcoming processor wars. I'm sure AMD is hard at work on 3DNow 2, but as you know, Intel has a few tricks up its sleeve as well. Their upcoming Katmai New Instructions are optimized for 3D calculations, and seem much more complex and robust than AMD's offerings. Paired with the enhanced FPU core, this chip should prove to be a screamer.
Let's take a look at this more closely. The K6 family can't execute 3D-Now and floating point simultaneously (not that its sorry floating-point unit would help much anyway). Katmai, on the other hand, not only has a revised FPU core with 8 128-bit single precision packed registers, but it introduces a new CPU mode, something that hasn't been done since the 386. This new mode allows for MMX, floating point, AND KNI to be accessed simultaneously. Considering how strong KNI is supposed to perform against 3DNow, the addition of FPU alone should skyrocket Katmai to the top of the 3D performance charts, and keep it there until AMD (or someone else) comes out with 3DNow2 or implements their own KNI instruction set.