Quote Originally Posted by cadaveca View Post
um, 1000w not enough?



actually i thought about this...but if you were right, then when i loaded 3d up, the system should crash from lack of power(2x2900XT's, phys-X), however, I have not had any issues in that regard.

when you think about it tho...cpu is supplied by 12v line, 8-phases...12v/8= 1.5v....issue begins @ 1.5v approx....
Quote Originally Posted by jonny_ftm View Post
You're right, after all, the +12v rails amperage limit is only theorical, as 99.99% of the PSUs seems "forgotting" the power protection if one of the +12v rails exceeds the 18-20A limit fixed by intel. So, as many claims, it is the combined +12v Amperage that should play a role instead of the +12v limit per rail (18-20A officially to respect intel specifications).

In most calculators, a Q6600 G0 at 3.8GHz and 1.6 vcore can drain near 250W and 20-23A. If the PSU respects the combined +12v amperage "lease", the PWM would be the limiting factor.

If the CPU, after asking for about 20A will overload the 8 PWM feeding it, they'll enter a protection phase (EPU) thus limiting the current going to the CPU. At least, this is the most logical theory. In that case, only a PCB revision would solve it, with stronger PWM components
So you guys know that the 12V feeds the CPU. I always thought it was the 3.3V rails... So an x-phase regulator actually divides the supply voltage by x?