Page 1 of 1

d945gclf2 (DUAL CORE ATOM) power consumtion

Posted: Mon Sep 15, 2008 4:59 pm
by andrewb
So we measured the power consumption, rail by rail for the D945GCLF2, here are the results.... The amazing thing is that the F945GCLF (single core) consumes only 1watt less (actually 750 mwatts)

3.3V rail < 300mA
5V rail = 2.8A (mainly the 945 Chipser)
12V rail < 200mA
-12V rail = 0.05A
5VSB < 300mA

The total power consumption floats around 1.65A @ 12V. We used a picoPSU-90 PSU and measured the input with a fluke current probe.

Hope this helps,
-A

Re: d945gclf2 (DUAL CORE ATOM) power consumtion

Posted: Mon Sep 15, 2008 5:00 pm
by andrewb
BTW, here is the board: http://www.mini-box.com/Intel-D945GCLF2 ... tegory=100
andrewb wrote:So we measured the power consumption, rail by rail for the D945GCLF2, here are the results.... The amazing thing is that the F945GCLF (single core) consumes only 1watt less (actually 750 mwatts)

3.3V rail < 300mA
5V rail = 2.8A (mainly the 945 Chipser)
12V rail < 200mA
-12V rail = 0.05A
5VSB < 300mA

The total power consumption floats around 1.65A @ 12V. We used a picoPSU-90 PSU and measured the input with a fluke current probe.

Hope this helps,
-A

Posted: Tue Sep 16, 2008 8:11 am
by tnynyn
very nice.

didnt even know it was out already..

Posted: Tue Sep 16, 2008 8:26 am
by Luminair
good work soldier ! !

Posted: Wed Sep 17, 2008 2:53 pm
by webwiznz
Great job. Are these figures at idle or loaded? How did you load the board?

Can you also confirm that its excluding drives, usb and fans?

Thanks from the mp3car thread!

http://www.mp3car.com/vbulletin/general ... tom-7.html

Re: d945gclf2 (DUAL CORE ATOM) power consumtion

Posted: Wed Sep 17, 2008 6:00 pm
by andrewb
The test was done idle and full load, the differences in loading were minor. Looks like the D945 chipset is the energy vampire here, the ATOM CPU loading is insignificant.

I took the system from BIOS post to full XP windows and variations in
andrewb wrote:nificant, e.g. 2.9A vs 2.7A on the 5V rail.
The 12V rail (where the CPU is getting V(core) from was fluctuaing in between 100 and 200mA. Go figure...

-A
andrewb wrote:BTW, here is the board: http://www.mini-box.com/Intel-D945GCLF2 ... tegory=100
andrewb wrote:So we measured the power consumption, rail by rail for the D945GCLF2, here are the results.... The amazing thing is that the F945GCLF (single core) consumes only 1watt less (actually 750 mwatts)

3.3V rail < 300mA
5V rail = 2.8A (mainly the 945 Chipser)
12V rail < 200mA
-12V rail = 0.05A
5VSB < 300mA

The total power consumption floats around 1.65A @ 12V. We used a picoPSU-90 PSU and measured the input with a fluke current probe.

Hope this helps,
-A

Posted: Wed Sep 17, 2008 9:47 pm
by webwiznz
I was planning on doing these measurements, and if your test is clear enough i wont have to try to locate a DC clampmeter.

However, could you please confirm:

- your figures exclude HDD, optical, usb, fans(ie MoB only)
- if you can, document loaded and unloaded figures, and specifiy how you loaded the baord

Mike chin used 2 x cpu burn to load the cpu plus 3dmark to load the north bridge, it would be the latter i think more likely to be the issue in this case. If the i945 is rated at TDP of 22Watts, then at 5 volts thats 4.4amps compared to your 2.8 amps. Thats a resonable discrpency.
With the majority of DC-Dc baords maxing out at 5-6 amps on the 5volt rail , by the time you take into account an hdd and optical, plus 1-2 usb gadgets you are running perilously close to those psu's capacity.

Hence we need to be clear about it.

Appreciate your help with this. It will inform many many folks PSU buying decisions.

Posted: Thu Sep 18, 2008 12:49 pm
by andrewb
I understand what you are saying, but TDP cannot be reached with software, IMHO. From what I saw on my oscilloscope was basically a flat line, no matter what, with little 10-40ms spikes of current a little over 3A.

>With the majority of DC-Dc baords maxing out at 5-6 amps on the 5volt >rail , by the time you take into account an hdd and optical, plus 1-2 usb

Understood. With 2.5' HDD you will be at 3 - 3.5A. CDROM is not an 'always on' device. Something like picoPSU-90 or 120 will handle as it trips at currents greater than 8A. Continous you can do 6-7A.

Sorry, I don't have the time to do more agressive loading. :(

I like this board very much (D945GCLF2), I think it is superior to the VIA/C7, if you are looking at CPU perfomance and price. I can't wait to see 945GME based designs, which have at least 3x less power consumption. Fanless designs, oh yeah! :)

-A

webwiznz wrote:I was planning on doing these measurements, and if your test is clear enough i wont have to try to locate a DC clampmeter.

However, could you please confirm:

- your figures exclude HDD, optical, usb, fans(ie MoB only)
- if you can, document loaded and unloaded figures, and specifiy how you loaded the baord

Mike chin used 2 x cpu burn to load the cpu plus 3dmark to load the north bridge, it would be the latter i think more likely to be the issue in this case. If the i945 is rated at TDP of 22Watts, then at 5 volts thats 4.4amps compared to your 2.8 amps. Thats a resonable discrpency.
With the majority of DC-Dc baords maxing out at 5-6 amps on the 5volt rail , by the time you take into account an hdd and optical, plus 1-2 usb gadgets you are running perilously close to those psu's capacity.

Hence we need to be clear about it.

Appreciate your help with this. It will inform many many folks PSU buying decisions.