- Dec 14, 2000
- 680
- 0
- 76
I have a watt meter hooked up directly to the power outlet of my house which tells me exactly what power my household is using and the money it costs. It allows electric companies to do remote metre reads as well.
After checking the accuracy using a few known devices with known power consumption figures, it definitely seems fairly accurate.
I then decided to do some experiments with my computers. Making sure all other devices in the household were off and after recording the stable base reading, I switched each machine on minus the monitor.
Some interesting results. One I was surprised at was the difference at idle between AUTO or OFFSET voltages vs a static/manual voltage.
My 2500K at 4.4Ghz (when turbo, 1.6Ghz idle all sleep states on) uses exactly the same power at idle and under light loads with a static 1.30v vs an offset or AUTO voltage (which reduces voltages to really low values at idle/light loads).
I can set a static voltage to a lower value and receive better stability and lower temps and power usage at 4.4Ghz vs using offsets/AUTO adjustment. This means I use less power under full load with a static voltage.
So it seems that it is better to use a static voltage for overclocking, as CPU must be good at shutting everything down internally under idle and low loads regardless of the external voltage.
The CPU may suffer from more electro-migration perhaps, but I doubt it will affect the chips lifespan much at a low voltage like 1.3v.
Now I always thought static voltages would increase power usage - not the case from what I can see, it seems to lower power usage when overclocking. Some others get similar results http://mattgadient.com/2012/05/21/sandy-bridge-undervolting-offset-vs-manual-observations/
However I am aware other sites show there is an increase in idle power use.
After checking the accuracy using a few known devices with known power consumption figures, it definitely seems fairly accurate.
I then decided to do some experiments with my computers. Making sure all other devices in the household were off and after recording the stable base reading, I switched each machine on minus the monitor.
Some interesting results. One I was surprised at was the difference at idle between AUTO or OFFSET voltages vs a static/manual voltage.
My 2500K at 4.4Ghz (when turbo, 1.6Ghz idle all sleep states on) uses exactly the same power at idle and under light loads with a static 1.30v vs an offset or AUTO voltage (which reduces voltages to really low values at idle/light loads).
I can set a static voltage to a lower value and receive better stability and lower temps and power usage at 4.4Ghz vs using offsets/AUTO adjustment. This means I use less power under full load with a static voltage.
So it seems that it is better to use a static voltage for overclocking, as CPU must be good at shutting everything down internally under idle and low loads regardless of the external voltage.
The CPU may suffer from more electro-migration perhaps, but I doubt it will affect the chips lifespan much at a low voltage like 1.3v.
Now I always thought static voltages would increase power usage - not the case from what I can see, it seems to lower power usage when overclocking. Some others get similar results http://mattgadient.com/2012/05/21/sandy-bridge-undervolting-offset-vs-manual-observations/
However I am aware other sites show there is an increase in idle power use.
Last edited:
