yesplease wrote:
For whatever reason, AMD doesn't go below 1.1V on the X2 chips, which is why their power consumption is relatively high compared to the C2D chips that drop to .85V IIRC, with power consumption being related to the square of voltage and all that's nearly a 70% reduction in power, compounded by, as others have mentioned, the differences between 90nm and 45nm chips. It's kinda odd that AMD wouldn't drop down to at least .9V with the X2s since that would result in a significant reduction in power consumption in that P-state. I have an old Sempron 64 that likely uses a little less energy than my Via C7 at .9V/1ghz, so AMD chips are certainly capable of lower voltages at 90nm IME.
The reason I would guess why AMD doesn't go lower is because there is a clear connection between threshold voltage (i.e., the voltages at which transistors turn on and off) and capacitance, and a connection between capacitance and clock (in that each clock needs to allow enough time for the "capacitances to charge up," in some sense)
Decreasing the core voltage requires you to at the same time lower the threshold voltage, which requires (given any particular fabrication process) higher capacitance on the gates, which means longer time to charge and thus lower clock speeds and usually lower performance.
You can somewhat account for these tradeoffs in the microarchitecture (squeezing some extra clock speed from lower-voltage circuits) but more fundamentally, Intel is taking advantage here of their much-improved fab process technology to build transistors that have better overall performance characteristics, allowing them to lower threshold voltage without slowing down clock too much.