C2Q seems to turn my power saving belief upside down :S

[arthurleung]

I just did some power usage testing on my X3350.

Stock Speed 2.66Ghz, Minimum voltage 1.1V: 161W Idle, 205W Full Load (13Mhz/Watt)
Overclocked 3.41Ghz, Moderate voltage 1.2875V: 158W Idle, 236W full Load (14.4Mhz/Watt)

I was expecting the idle power usage to be (significantly) lower with lower voltage (Ram,NB,SB) and frequency, but get quite an opposite :S

I'll give Ai Quiet a try and see if that lower the power usage more.

[Agent]

How are you testing the load? Through one of the kill-a-watt things?

It could be that the extra load is pushing the power draw into the part of the curve where your PSU is more efficient.

[Webby]

Yeah what Agent said also actually making a PC idle is a very hard thing to do, have you turned off everything anti-virus, firewall, page file you have to have absolutely nothing running except windows and even then you will get fluctuations the 3W difference you are seeing is say +- 1.5W of 159.5W that is an error of less than 1% I wouldn't expect your meter to have this level of accuracy.

Either way I would find it very hard to draw conclusions from data so close together.

[llololloy]

webby, surely the easiest way to test idle is to leave it in bios screen then test with multimeter.

[ajones]

webby, surely the easiest way to test idle is to leave it in bios screen then test with multimeter.

You wouldn't know what impact everything else in the hardware is having on the ultimate power draw then when you run the load test on a system with an OS and load application running. You need to make sure that the environment for testing idle draw is as similar as that under CPU only load.

[arthurleung]

Yeah what Agent said also actually making a PC idle is a very hard thing to do, have you turned off everything anti-virus, firewall, page file you have to have absolutely nothing running except windows and even then you will get fluctuations the 3W difference you are seeing is say +- 1.5W of 159.5W that is an error of less than 1% I wouldn't expect your meter to have this level of accuracy.

Either way I would find it very hard to draw conclusions from data so close together.

But then I'm not looking at the theoretical idle load, but real-life idle load, which includes those anti-virus, firewall, etc.

About the measurement error, since I'm using the same meter it would cancel out the systematic error, and even at 1% error it still show that the power usage is more or less the same, hence this post.

Which suprised me consider the voltage is 17% higher plus 30% higher clock speed. Guess the CPU doesn't really use much at idle.

[shomei]

I just did some power usage testing on my X3350.

Stock Speed 2.66Ghz, Minimum voltage 1.1V: 161W Idle, 205W Full Load (13Mhz/Watt)
Overclocked 3.41Ghz, Moderate voltage 1.2875V: 158W Idle, 236W full Load (14.4Mhz/Watt)

I was expecting the idle power usage to be (significantly) lower with lower voltage (Ram,NB,SB) and frequency, but get quite an opposite :S

I'll give Ai Quiet a try and see if that lower the power usage more.

You could try to take the measurements in linux where idle is much more "idle" than in windows and get better readings.

To get real-life idle load (as you say), you have to test for a really long period of time so that the error becomes as small as possible (for example, when you took your first measurement the OS might have used a lot of CPU power to do some of its "idle" processes while it didn't when you took the second)...
For the full load part is easier, you just use a benchmark for a relatively small amount of time..