View Single Post
Posts: 1,680 | Thanked: 3,685 times | Joined on Jan 2011
#217
Here is a plot of my SR values against frequency.

The X axis is frequency.

The Y axis is calculated power. Power is calculated from P=C*F*V^2. Where P is power in Watts, C is capacitance in Farads, V^2 is volts squared measured in Volts.

The voltage is calculated by the base voltage plus the number of voltage steps(0.0125V per step). So step 20 is (0.0125V*20) +0.6V=0.85V

Unfortunately I have no idea what the gate capacitance of the chip is. I have set it as 0.1F which is INSANELY high, however it is just a multiplication factor so it simply scales the output of the equation. So the power values are total nonsense however the shape of the graph is what is important.




Code:
Frequency	Step	Voltage	Power
0	0	0.6125	0.0375
125	20	0.85	9.03125
250	23	0.8875	19.69140625
500	36	1.05	55.125
550	38	1.075	63.559375
600	40	1.1	72.6
720	46	1.175	99.405
805	50	1.225	120.8003125
850	52	1.25	132.8125
900	54	1.275	146.30625

As you can see the voltage steps increase linearly, the power required increases exponentially. From this I would conclude that the 'most bang for buck' is in fact at 600MHz.

//EDIT: Real Conclusions

You can see from the graph that the difference in power use between 125-250MHz is negligible. When you consider the performance hit you take by activating 125MHz it makes sense to avoid it. Here doubling the frequency takes a tiny amount of power.

after 250MHz the increase in power use is almost linear up to 600MHz. Above 600 the power use line changes to a much higher gradient. This means increases in speed come at a much higher power cost.

meh, TBC.
__________________
N900: One of God's own prototypes. A high-powered mutant of some kind never even considered for mass production. Too weird to live, and too rare to die.

Last edited by vi_; 2012-04-05 at 09:21.
 

The Following 10 Users Say Thank You to vi_ For This Useful Post: