Giving a final word on the K6-2 is certainly not as easy as it used to be with previous CPUs. As already pointed out in the introduction, the K6-2 is facing a few pretty serious problems. Will the 3D game developers jump on the 3DNow! train? Will the people have enough belief in Socket 7? Will the Taiwanese chipset manufacturers offer convincing solutions for the Super7 platform?
There are three different levels where performance increase can be achieved by implementing 3DNow!:
- Possible Performance Gain
- 3D Game Engine - up to 100%
- DirectX 6 Support - up to 30%
- 3D Card Driver - up to 15%
The concept of the K6-2 is certainly a very good approach, being more than half a year earlier than Intel with its Katmai CPU, using the MMX2 or 'KNI' instructions and offering a similar solution to 3DNow!. The K6-2 could mix up the 3D gaming area by a whole lot, since the performance offered by 3DNow! could change the CPU hirarchy significantly. I've got a comment from AMD's Director of Technical Marketing, Lance Smith (alias SFX), which should say it all: "The 3DNow! is faster than Intel's best FPU and on a Super7 bus without a backside L2!". I don't doubt that the potential of the K6-2 is close to amazing, however am I not in the position to foresee if the answers to the above questions will be all 'yes'. AMD will do as much as they can, generating an attractive environment for game developers, offering a 3DNow! vector compiler soon and they will hopefully kick some serious butt back there at the Taiwanese chipset and motherboard manufacturers. All in all it's going to be up to you users out there, if you will have enough belief in the K6-2. If enough units are sold, the game developers will be pushed to do something about this, the chipset manufacturers will smell the money too and the future of Socket 7 will be safe. It's up to you guys. It is certainly better to keep a serious competitor to Intel, isn;t it? Or do you like the situation on the OS market, where Microsoft can play around with us as much as they like?