W1zzard said,
"We can see that both GPUs run different clocks, with the first one being typically higher clock and voltage while the second one uses more balanced clocks and voltages."
That's the most interesting... And do they have two sets of dynamic control components? Or did they figure a way to provide separate outputs using the same voltage controller. Heck I don't pick out the INA219 power monitor chips (there probably there someplace) or anything in appearance to the Richtek RT8802A voltage controller which had its own little PCB on the reference GTX680? That's the telling piece of this, if they have freed up some PCB real-estate and invested a new means of dynamic clock (less pricey?) components, then maybe a GTX670 with Clock Boost might see light. Has anyone paid attention to how AIB's 680 custom PCB's are implementing dynamic clock and what component?
It’s the new standard to compete to, for that it’s a marvel, the use of separate boost profiles is what gave it the lower Peak power (23%) over 680’s in SLI huge and the first time cost such dual card vs. SLI shows merit with no real difference in performance.
Here’s my question two 690’s in SLI, would all 4 GPU with one running a higher clock and voltage, while the other three run the balanced clocks? That be really "green" of it!
For those who need this level to play 2650x and above it will turn out to be good day when it arrives…? Well it still doesn’t play Metro 2003 @2650x... what’s up with that.