Low Voltage CPU

limee · Post by **limee** » Fri Oct 31, 2003 1:41 pm

Hello, I've been looking for another CPU/Mobo combo that wouldn't use too much electricity. I'm considering perhaps an AMD Barton 2700+ as they are quite cheap right now and are unlocked. The mobo on my mind is the Albatron KX18D PROII, but I'm not exactly sure of it's undervolting/clocking options. Read the review for it here:

http://www.anandtech.com/mb/showdoc.html?i=1872&p=1

Does anyone have any suggestions? Also, how much idling voltage use could be reduced by underclocking/volting?

silvervarg · Post by **silvervarg** » Mon Nov 03, 2003 6:39 am

If you want to go real low on electricity or heat dissipation you might also consider reducing the clock multiplier.
If you even think about going with lower multiplier that default I can't see a reason to pick the Barton 2700+ over the Barton 2500+.

The board looks very good on the review you linked. The only thing that bothers me slightly is the lack of holes around the processor. This will limit you choice of CPU-heatsink a bit.

wsc · Post by **wsc** » Mon Nov 03, 2003 12:53 pm

I second the vote for Barton 2500 (if for no other reason, it's cheaper than the 2700). I also would recommend you take a look at the Abit NF7-s because it has Vcore down to 1.1V and it also has the four mounting holes around the processor.

silvervarg · Post by **silvervarg** » Mon Nov 03, 2003 1:09 pm

Another good option for motherboard is Aopen AK79G series (with integrated graphics) or AK79D series (without integrated graphics).
All of them undervolt to 1.1 volts and they have holes around the CPU socket.
Note that the MK (mATX) series does not undervolt.

More info on Aopen boards

limee · Post by **limee** » Tue Nov 04, 2003 5:07 pm

The AK79D-400 Max has alot of features and one of the few to have integrated SATA. The only thing that worries me was HardOCP's review
which stated stability problems, even before overclocking. Any thoughts on this?

http://www.hardocp.com/article.html?art=NTI1

Also, is there any way to calculate a rough estimate of voltage consumption if clock multiplier has been reduced by x amount?

silvervarg · Post by **silvervarg** » Wed Nov 05, 2003 12:59 am

Sorry to here that there might be stability problems on the AK79D-400. Actually this was the first time I had heard of that problem.

Also, is there any way to calucculate a rough estimate of voltage consumption if clock multiplier has been reded by x amount?

I guess you meant wattage (=heat).
There is a program written by someone here at SPCR that I have seen a like to. It is a windows program called "CPU Power".
In that program you can pick a processor. Then you will see the stock vCore and clock (MHz) settings along with the max wattage.
Then you can freely change VCore and Clock and see how it affects the wattage.
Here is an example:
Athlon XP Barton 2500+ (1.65)
Max power dissipation 68.3
VCore 1.65
Clock 1833

CPU (new)
Vcore 1.25
Clock 1833
New power dissipation: 39.1W (-42.6%)

This example setting is used by at least one member here at SPCR with a stable system.
The theory this builds upon is that effect (W) is proportional to clock and to vcore^2.
So lowering the vcore is essential. Lowering the clock-speed is of much less importance. However, by lowering the clockspeed you can often lower the vcore even more.