Exaggerated? You must be wishing. Go check out the 3D benchmark on AMD homepage.
Well, I stand corrected, but not because of AMD's homepage, but because of Dr. Tom's tests. My original argument was that DirectX support alone isn't going to contribute a 50% performance increase. But it seems that 3D driver support will, and native support within the game itself will contribute even more.
The next cool game out is Trespasser. It will show you why K6-2 is better than PII.
Um, never heard of that game, and I'm a big games freak. Seems like Quake II and Unreal are better examples of K6-2's competitiveness. I haven't seen Unreal yet, but I was personally less than impressed with Quake II's graphics.
KNI will probably end up as a dead bead just like MMX.
Care to put money on that? It can be assumed that KNI will support single-precision SIMD just like 3D-Now. As for MMX being a dead bead, it's funny how AMD and Cyrix both chose to support MMX in their latest processors, eh? Whether or not you think MMX does any good is irrelevant. In fact, it shows just how well Intel can market such a technology which isn't as useful as it originally seemed. Imagine how well Intel can market MMX2 whose usefulness far outshines MMX.
In semiconductor, life and death is depended on the yield and die size.
And you think AMD has a better yield than Intel? But to be fair, AMD's yields will likely be more than half that of Intel's. Like I said before, AMD doesn't enjoy the economies of scale that Intel does, so it needs that smaller die size.
How many installed units must there be for you to consider critical mass?
Enough to overcome the marketing muscle of Intel. At the very worst, we'll have a standards war, where each company will be pushing their instruction set as the standard, much like a VHS vs. Beta war. AMD has the headstart, but Intel has the sheer size. It will be an interesting race, but either way, it's kind of dumb to assume that KNI will just wither away.
[useless "Intel sucks" drivel snipped]
Tenchusatsu |
